Pedagogical Rules and their Relationship to …...rules, guided by UG and relatively independent of...
Transcript of Pedagogical Rules and their Relationship to …...rules, guided by UG and relatively independent of...
Applied Linguistics 282 286ndash308 Oxford University Press 2007
doi101093applinamm015 Advance Access Published on 7 June 2007
Pedagogical Rules and theirRelationship to Frequency in theInput Observational and EmpiricalData from L2 French
BRUCE ANDERSON
University of California Davis
Corpus-based research has shown that the frequency of use of particular
grammatical structures and lexis in English is not always congruent with the
content or ordering of explicit rules in pedagogical materials The present study
provides an additional example from French focusing on word-order rules
related to adjective position and the (in)congruity of those rules with classroom
input and texts written for and by native French speakers In addition it
demonstrates how this (in)congruity leads to a particular and particularly
nonnativelike performance on the part of even highly proficient L1 English-
speaking learners in their evaluation of the acceptability of contextualized
French sentences This state of affairs leads us to re-examine the debate
surrounding native speaker usage as both the basis of pedagogical norms and the
goal of foreign language instruction
Frequency of use of particular grammatical constructions and lexis has played
an important role in both linguistic theory and (second) language acquisition
The present study examines frequency of use from the perspective of the
applied linguist as intermediary between the worlds of the theorist-
researcher and the teacher-practitioner (Koike and Liskin-Gasparro 2003
263) We investigate the extent to which the pedagogical treatment of
French adjective position typically found in American textbooks is congruent
with word-order frequency patterns in corpora based on authentic French-
language texts andor with oral input on adjective position in a typical
American university classroom setting This leads us to an investigation of
how certain (in)congruities may manifest themselves in learnersrsquo accept-
ability judgments when they are presented with lsquoexpectedrsquo and lsquounexpected
but nativelikersquo word orders within context The cumulative results from these
investigations and the importance of frequency patterns in the field of
applied linguistics more generally are discussed with respect to the on-going
pedagogical versus descriptive norm debate (Owen 1993 Widdowson 2000
see also Gass et al 2002) and the related native-speaker-usage-as-goal debate
(Blyth 2003)
THREE PERSPECTIVES ON FREQUENCY
Within linguistic theory frequency has long been associated with the notion of
markedness Greenberg (1966 67) in fact suggested that frequency is the
primary criterion for unmarked versus marked word order though one must be
careful to distinguish markedness in cross-linguistic terms from markedness
within a given language For example whereas van Riemsdijk (1978) argued
for preposition pied-piping in wh-questions (eg To whom did Cathy give a book)
as the more frequent and thus unmarked construction cross-linguistically Ross
(1977) observed that pied-piping has been undergoing neutralization in favor
of preposition stranding (eg Who[m] did Cathy give the book to) making it the
more frequent and thus unmarked construction in modern (colloquial) English
The relationship between frequency and markedness within linguistic
theory as exemplified above has in turn been investigated from a second
perspectivemdashthat of (second) language acquisition theorymdashby Mazurkewich
(1984) and Bardovi-Harlig (1987) Although Mazurkewichrsquos study appeared
to illustrate the strength of cross-linguistic markedness in determining
acquisitional sequences through empirical data suggesting the acquisition of
pied-piping before stranding among French- and Inuktitut-speaking ESL
learners Bardovi-Harligrsquos replication and re-interpretation of that study
showed just the opposite order illustrating the equal or greater importance of
the lsquodistribution of data available as inputrsquo (1987 400)1
Some twenty years later frequency effects in (second) language acquisition
continue to be addressed in the literature most particularly concerning the
extent to which first (L1) andor second (L2) language epistemology can be
defined in such terms Ellis (2002) for example provided a sweeping
examination of studies on input frequency effects in L1 processing claiming
that such effects are not simply a key determinant of acquisition but its sole
determinant Linguistic units are abstracted from language use and language
rules are in effect lsquostructural regularities that emerge from learnersrsquo lifetime
analysis of the distributional characteristics of the language inputrsquo (2002
144) This stands in stark contrast to so-called rationalist or innatist accounts
in which frequency effects are epiphenomenal (see for example Eubank and
Greggrsquos (2002) peer commentary of Ellis)
Bley-Vroman (2004) has taken an intermediate stance in arguing that
language acquisition at least in L1 involves both the lsquoformation of abstract
rules guided by UG and relatively independent of frequency effectsrsquo and
the lsquopiecemeal instance-based acquisition of particular structures which can
be substantially affected by the statistical structure of languagersquo (2004 269)
L2 acquisition according to Bley-Vroman appears to depart significantly
from this two-part process resulting in learner grammars that are based
significantly moremdashor perhaps whollymdashon lsquowhat they have heard and how
oftenrsquo (2004 263)
Bley-Vromanrsquos (2004) claims are partly based on data from Bley-Vroman
and Yoshinaga (2000) a study in which native English speakers and L1
BRUCE ANDERSON 287
Japanese learners of English were asked to judge the acceptability of multiple
wh-questions of the type Who ate what and Who sat where wherein
the sentence-final wh-word is a complement of the verb versus questions of
the type Who sang where wherein the sentence-final wh-word is an adjunct
Irrespective of this complementndashadjunct distinction the frequency of use of
the three question types was found to be quite low Only 122 examplars
beginning with who could be found in the 56 million word corpus
Bley-Vroman consulted 106 (or 87 percent) were specifically of the
who what type Native English speakers whose judgments were hypothe-
sized to follow from abstract rules favored both who wh[complement]
structures and disfavored the who wh[adjunct] structure despite the across-
the-board low frequency of all such structures in the corpus consulted The
learner group as expected favored the (relatively) predominant who what
type to the near exclusion of the other two seemingly paying no heed to the
complementndashadjunct distinction
Though certainly useful from the standpoint of L2 acquisition studies both
in terms of methodology and epistemological arguments frequency patterns
culled from authentic native-speaker corpora do not as Bley-Vroman (2004
270) pointed out necessarily reflect what learners are exposed to and it is
this latter type of corpus that is crucial in L2 acquisition studies because it
represents the input with which a learning algorithm would interact
The potential dichotomy between frequency patterns in native-speaker
usage and what classroom-based learners are exposed to (viewed broadly as
the sum total of textbook presentations authentic texts used in the
classroom and lsquoteacher talkrsquo) represents a third and especially applied
linguistic perspective on frequency An early study by Holmes (1988) on the
presentation of expressions of doubt and uncertainty in ESL textbooks for
example found that textbooks largely favored modal verbs while
simultaneously neglecting alternatives such as adjectives tag questions and
fallndashrise intonation (1988 40) This led her to raise concerns that a simplicity
criterion in the formulation of textbook rules may come at the cost of
naturalness and thus she concurred with Kasper (1979 277) that the
reduced range of expressions of modality witnessed in learnersrsquo production
can in some cases be viewed as teaching induced
More recent cases of mismatch have come to light through the corpus-
based methodology of Biber (1988 1995) and Biber et al (1998) and the
corpus-based Longman Grammar of Spoken and Written English (Biber et al
1999) A recent study by Biber and Reppen (2002) for instance examined
six ESL textbooks all of which presented attributive adjectives as the
principle means of modifying nouns Two-thirds of these textbooks included
some treatment of participial adjectives (those ending in -ed and -ing)
but only one discussed the use of nouns rather than adjectives as
prenominal modifiers (eg tomato sauce pencil case) However Biber et alrsquos
(1999) analysis of Englishmdashbased on a corpus of conversational transcripts
fictional literature newspaper writing and academic prosemdashshowed
288 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that (a) participial adjectives are quite rare across the four registers
(b) common attributive adjectives are in fact used the most in conversation
to modify nouns but most nouns actually have no modification at all in that
register and (c) nouns as prenominal modifiers are quite frequent in
newspaper writing and nearly as common as adjectives
Such corpus-based findings Biber and Reppen (2002) suggested should
result in participial forms being given little attention in intermediate to
advanced ESL materials and nouns as prenominal modifiers (and the various
and complex meaning relations they evince) being given more prominent
coverage In other words a descriptive norm for noun modification established
through corpus-based research should serve asmdashor at the very least
significantly inform the elaboration ofmdasha pedagogical norm (See OrsquoConnor Di
Vito 1991 for a similar argument based on frequency data from French) This
seemingly simple equation is not without its critics however (eg Owen 1993
Widdowson 2000) and we re-examine their criticisms in the discussion section
in light of the findings presented here on French adjective position
RESEARCH QUESTIONS
Although the potential for mismatch between native-speaker usage and
pedagogical materials has received some attention in the field of applied
linguistics the present study is unique in also offering a glimpse of the
potential for mismatch between pedagogical materials and ambient language
use in the classroom and as a result the potential for mismatch between
native speaker and L2 learner intuitions of acceptable word order(s) The
following three research questions are therefore addressed
1 To what extent are textbook presentations of French adjective position
congruent with authentic text sources across a number of genres
2 What is the nature of ambient language use involving adjective position
in the L2 French classroom and to what extent is it congruent with
textbook presentations of adjective position
3 To what extent are intuitions of acceptability among classroom learners
biased by any such (in)congruities
Because each of the three research questions represents a mini-study in
and of itself involving a different study methodology and type of data
analysis I address each question in its own section providing first an
overview of the methodology used and then a presentation of results
FRENCH ADJECTIVE POSITION IN TEXTBOOKS ANDAUTHENTIC SOURCES
The complexity of English adjectival modification pointed out in Biber and
Reppen (2002) is compounded in French by the fact that adjectives appear
BRUCE ANDERSON 289
in both adnominal positions prenominally (pre-N) and postnominally
(post-N) following certain lexical and contextual (pragmatic) constraints as
to their interpretation in one position versus the other (see eg Delmonier
1980 for a general review Delbecque 1990 for an account within cognitive
grammar Anderson 2002 for a compositional account within generative
syntax) This state of affairs leads us to question whether the traditional
textbook rules relating to adjectival modification in French reflect
attested native-speaker language use any better than that uncovered by
Biber and Reppen in authentic English-language texts across a number of
genres
English-speaking learners of French at least in the typical American
university setting are provided with three adjective placement rules
which take the form of a general rule with two exceptions Most adjectives
in French are said to be post-N (as in (1)) though a few are pre-N (as in (2))
and a few more are said to change meaning based on their position
(as in (3))
(1) une voiture rougelsquoa red carrsquo
(2) une petite maisonlsquoa small housersquo
(3) a un bijou cher un cher bijoulsquoan expensive jewelrsquo lsquoa cherished jewelrsquob une eglise ancienne une ancienne eglise
lsquoan ancient churchrsquo lsquoa former churchrsquo
First- and second-year textbooks of French published in the USA are
strikingly similar in the statement of such rules and in their order of
presentation though they differ to some degree as to which adjectives are to
be included in the (putative) lexical classes constituting exceptions to the
general post-N rule The textbooks used in the classrooms observed for the
present study are no different In the first-year textbook Chez nous (Valdman
and Pons 1997) students are first told that most adjectives follow the noun
(1997 73) but that the twelve adjectives listed in (4) typically and
exceptionally precede the noun (1997 241)
(4) a jeune vieux lsquoyoung oldrsquob nouveau lsquonewrsquoc petit grand gros lsquosmalllittle bigtall largersquod beau lsquohandsomersquoe joli lsquoprettyrsquof bon mauvais lsquogood badrsquog premier dernier lsquofirst lastrsquo
In the second-year textbook Quant a moi (Bragger and Rice 1996) the basic
post-N rule is reviewed as is the class of adjectives that exceptionally appear
prenominally (1996 33) Note however that this class of exceptions has
eleven members instead of twelve having lost some members from the Chez
290 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
Nous list in (4) (eg bon premier dernier gros) but gaining others (eg long
lsquolongrsquo court lsquoshortrsquo and autre lsquootherrsquo)
Bragger and Rice (1996) then present the second exception to the basic
post-N rule wherein students are told that lsquoa number of adjectives change
meaning depending on whether theyrsquore placed after or before a noun
In general an adjective placed after the noun retains its basic concrete
meaning The same adjective placed before the noun is used in an abstract or
figurative mannerrsquo (1996 34 emphasis in original) The seven adjectives that
apparently fall into this class are listed in (5) The first English gloss captures
the meaning derived from pre-N position the second gloss that derived from
post-N position Note that Bragger and Rice put the adjective grand in both
lists of exceptions In other words grand is apparently exceptionally pre-N
and exceptionally variable at the same time2
(5) ADJECTIVE PRE-N POST-Na ancien lsquoformerrsquo vs lsquoold ancientrsquob cher lsquodear well-lovedrsquo vs lsquoexpensiversquoc dernier lsquolast latest (in a series)rsquo vs lsquolast (before this one)rsquod grand lsquogreatrsquo vs lsquotall largersquoe pauvre lsquopoor (unfortunate)rsquo vs lsquopoor (not rich)rsquof prochain lsquonext (in a series)rsquo vs lsquonext (after this one)rsquog propre lsquoownrsquo vs lsquocleanrsquo
To summarize the pedagogical treatment of French adjective position
commonly found in university-level American textbooks presents post-N
position as the most frequent and typical (frac14unmarked) adjective order
the less frequent and exceptional (frac14marked) pre-N order is either required
when the adjective is a member of a seemingly fixed lexical class such as
the list in (4) or licensed by a particular interpretation that differs from that
obtained in post-N position such as the list in (5) This difference in
interpretation is often though not always captured through distinct English
glosses
Methodology
In order to determine the extent to which the pedagogical treatment just
summarized is congruent with authentic French texts three corpora were
consulted Wilmetrsquos (1980) corpus of literary texts (approximately 10 million
words 3835 adjectives) Forsgrenrsquos (1978) corpus of newspaper texts from Le
Monde and LrsquoExpress (1 million words 829 adjectives) and Larssonrsquos (1993)
corpus of travel catalogues travel guides and other non-literary prose (15
million words 113 adjectives)3 Using Wilmetrsquos corpus as the point of
comparison given that it contains the highest number of discrete lexemes
(3835 adjectives) an arbitrary cut-off point of 25 attestations or more was
used creating a list of 205 adjectives4 The total number of attestations pre-N
attestations and post-N attestations for each of the 205 adjectives across
BRUCE ANDERSON 291
the three corpora was then determined by adding Forsgrenrsquos and Larssonrsquos
data to that found in Wilmet if available
Results
The 205 most frequently attested adjectives in French texts appear to fall
along a continuum at one end of which we find 68 adjectives (or 332
percent) appearing either only in post-N position (nfrac1434) or typically in that
position (nfrac14 34) and at the other end of which we find 15 adjectives (or 73
percent) typically appearing in pre-N position as shown in Table 1 Note that
there are no adjectives appearing only in pre-N position The remaining 122
adjectives (or 595 percent) are regularly attested in both positions with
some having what we might call an affinity for post-N position (nfrac1441) or
pre-N position (nfrac14 29) but others being highly variable (nfrac1452)
Reminiscent of Holmesrsquo (1988) critique of the simplicity versus naturalness
of textbook rules such data raise some problems for the seemingly
straightforward fixed-order rules of adjective position exemplified in (1)ndash(3)
and the associated lists in (4) and (5) A textbook rule to the effect that
adjectives typically appear after the noun in Frenchmdasha word order that
clearly contrasts with what one finds in Englishmdashis only partly descriptively
Table 1 Frequency groupings of adjective position in written French texts
Frequency grouping Type count(nfrac14 205)
Percent oftotal
Examples
Exclusively pre-N (0post-N position)
0 0 mdash
Typically pre-N (below10 post-N position)
15 75 bon lsquogoodrsquo fameuxlsquo(in)famousrsquo petit lsquosmallrsquo
Variable but with anaffinity for pre-Nposition (10ndash39post-N position)
29 14 charmant lsquocharmingrsquo etrangelsquostrangersquo miserable lsquomiserablersquo
Highly variable (40ndash59post-N position)
52 255 delicieux lsquodeliciousrsquo joyeuxlsquojoyousrsquo puissant lsquopowerfulrsquo
Variable but with anaffinity for post-Nposition (60ndash90post-N position)
41 20 amer lsquobitterrsquo tiede lsquolukewarmrsquoserieux lsquoseriousrsquo
Typically post-N (above90 post-N)
34 17 blanc lsquowhitersquo inconnulsquounknownrsquo sauvage lsquowildrsquo
Exclusively post-N(100 post-Nposition)
34 17 allemand lsquoGermanrsquo dorelsquogoldenrsquo quotidien lsquodailyrsquo
292 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
accurate It is accurate in the sense that the ends of the continuum are
not equivalent in their percentages There are many more adjectives
appearing either only or typically in post-N position than is the case for pre-N
position It is inaccurate in the sense that it ignores the distributional
behavior of 595 percent of the most frequently used adjectives which can
appear in both positions French is not as the rule implies the mirror
opposite of English but instead is more accurately viewed as an lsquoANArsquo
(adjectivendashnounndashadjective) language as opposed to both lsquosystematic ANrsquo
languages like English and German and lsquosystematic NArsquo languages like
Indonesian and Thai (Cinque 1994 99ndash101)
The data also lead us to question the criteria textbooks use for including
certain adjectives in the two lists in (4) and (5) For the so-called prenominal
adjectives a frequency of 90 percent or greater in that position would
establish that adjectives such as excellent fameux lsquo(in)famousrsquo moindre
lsquoslightestrsquo and vrai lsquorealtruersquo need to be added That list itself implies a fixed
lexical class to be memorized though in reality it does nothing more
than provide a handful of frequently attested adjectives (nfrac14 15 [75 percent])
that typically appear just in that position Many more adjectives can
appear in pre-N position as well For the so-called change-of-meaning
adjectives the data show that the list in (5) falls woefully short in terms of
representing those adjectives that actually do regularly appear in both
positions (nfrac14 52)
Even if it were to be established that such rules are accurate with respect
to colloquial French as spoken between native speakers and thus appropriate
for oral proficiency-oriented classrooms such rules obviously fail to capture
the state of affairs in written texts particularly literature to which learners at
later course levels are exposed Yet even students at earlier course levels
occasionally find counter-examples to what they are taught The four cases of
lsquounexpectedrsquo pre-N position in (6) come from the cultural readings of two
first-year and two second-year US textbooks respectively
(6) a La France a un excellent systeme de transports publics(Magnan et al 2002 223)lsquoFrance has an excellent system of public transportrsquo
b Cela explique le grand nombre de magnifiques chateauxdans la region (Valdman and Pons 1997 395)lsquoThis explains the great number of magnificent chateaux inthe regionrsquo
c [U]ne veritable explosion de la technologie (St Onge etal 2003 71) lsquoa veritable explosion of technologyrsquo
d Lrsquohistoire est une douloureuse suite de tragedieshumaines (Oates and Dubois 2003 208)lsquoIts history is a painful series of human tragediesrsquo
One might wish to argue that the examples provided in (6) demonstrate
that inputmdashat least classroom-related inputmdashis not as contradictory to
descriptive norms as argued in this section Examples of the type in (6)
BRUCE ANDERSON 293
however are only occasionally found in textbook reading materials they are
included here to demonstrate that even when making a concerted effort to
write level-appropriate cultural readings textbook authors sometimes break
the very grammar rules they present elsewhere in the textbook suggesting a
true disparity between rules and lsquonaturalisticrsquo usage5 (I return to the issue of
descriptively accurate statements about French adjective position in the
discussion section)
FREQUENCY OF ADJECTIVE USE AND POSITIONIN CLASSROOM INPUT
Methodology
In order to get a general sense of how well the classroom setting as primary
input source reflects either textbook rules or the corpora-based descriptive
norm just summarized frequency counts of adjective type (ie discrete
lexemes) and their position in classroom input were collected by the
researcher from 60 hours of in-class observations of university-level French
courses at a large Midwestern American university Ten hours were spent in
each of six courses spanning the curriculummdashfrom the four semesters
comprising the first- and second-year language skills curriculum (referred to
as Levels 1a 1b 2a and 2b below) to a third-year conversation course
(Level 3) and a fourth-year lecture course on French civilization (Level 4)
Each of these courses was taught almost exclusively in French by a mix of
native and nonnative instructors none of whom were aware of the purpose
of this observation period Class observations were conducted on days when
adjective position was not the focus of explicit instruction as this would have
more than likely biased the frequency counts
The first- and second-year courses consisted of explicit grammar
instruction drills and communicative activities normally involving students
in pair or group work The conversation and civilization courses involved
no explicit grammar instruction the former involved mostly teacherndash
student and studentndashstudent oral communication based on previous course
readings and video segments whereas the latter was almost exclusively
devoted to instructor-delivered lectures Error correction was minimal in all
courses and no case of error correction involving adjective position was
witnessed during the observation period (despite the fact that some errors
were made)
Oral input on adjective position in all courses came from one of three
sources the instructor audio aides (such as cassette and videotape) and
the students themselves Written input in the first- and second-year
courses came from two sources visual aids (such as handouts transparencies
writing on the blackboard) and passages from the textbook that were read
silently or aloud in class At Level 3 written materials included authentic
newspaper and magazine articles in addition to visual aides at Level 4
294 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
a number of works of French literature were read outside of class Note
that written input in these latter two course levels is qualitatively the same
as that used in the corpora compiled by Forsgren (1978) Wilmet (1980) and
Larsson (1993)
At issue during this period of investigation was classroom input that could
be seen andor heard by all students in the classroom (as well as the
researcher) Input from students involved in pair and group work was
therefore excluded from analysis Material to be read outside of class was
also excluded as there was no guarantee that all students had actually
read such material Although the totality of inputmdashincluding sources of input
the learner might have voluntarily sought out outside of classmdashcould
obviously never be exhaustively tallied the exclusion of certain forms of
input was first and foremost a matter of logistics (ie not having the
appropriate funds equipment facilities and time to sound record then
transcribe all classroom input from every participant in every situation of
use) Despite this drawback I see no reason to suspect that student-delivered
input to other students during pair- or group-work would be qualitatively
different (at least in terms of adjective position) from the input delivered by
students that was audible to the entire class this latter source of input
already constitutes 10ndash30 percent of all sources of input tallied as part of the
observation period
Each instance of adjective use during each class session was recorded with
pen and paper by the researcher A coding scheme was devised whereby an
11 17-inch sheet of paper was divided into quadrants labeled according to
the source of the input lsquoinstructorrsquo lsquostudentrsquo lsquoaudiorsquo and lsquowrittenrsquo Each
quadrant was then subdivided according to the position of adjectival
modification lsquopre-Nrsquo lsquopost-Nrsquo lsquopredicatersquo and lsquootherrsquo (to be described
below) Input of the form une bonne idee lsquoa good idearsquo by the instructor was
therefore recorded as an instance of the adjective bonne in the instructor
quadrant pre-N sub-quadrant Hatch marks next to the adjective bonne were
used for any further exemplars of this adjective (including its masculine form
bon) in this position from this source I acknowledge that mere human error
caused by fatigue etc may have resulted in tallies that are less accurate than
if each session were recorded on tape Though the results presented below
should be considered with some caution then occasional errors could not
have reached a significant enough level to detract from the highly salient
patterns of usage that we will find
The resulting sixty sheets each representing one 50-minute class session
were then tallied in order to determine (a) the relative contribution of the
four sources of input (b) the average token (exemplar) count of adjectives
per 50-minute class period (c) the ratio of pre-N to post-Npredicate tokens
(d) the type count (number of discrete lexemes) entering into pre-N and
post-N positions and (e) the frequency of variation in positionmdashthat is the
extent to which a given adjective appeared at least once in both pre-N and
post-N position during the ten hours of observation
BRUCE ANDERSON 295
Results
The instructor not surprisingly was the primary provider of input in the
classroom (See Figure 1 available to online subscribers at http
applijoxfordjournalsorg) This was especially the case in the first-year
courses (Levels 1a 1b) the second- and third-year courses (Levels 2a 2b 3)
showed a greater proportion of input from students and from audio aides
The instructor was by far the predominant source of input in the fourth-year
course (Level 4) given that it involved instructor-delivered lectures on
French civilization with minimal class participation on the part of students
Concerning the average number of times an adjective occurred in the
input during any one 50-minute class period we find a general increase with
level from 42 tokens at Level 1a to 156 tokens per class at Level 4 (See
Figure 2 available to online subscribers at httpapplijoxfordjournalsorg)
Students therefore heard an adjective used anywhere from one to three
times per minute on average
Aggregate data on the frequency with which adjectives appeared in pre-N
post-N and predicate positions or in other constructions (eg used in
isolation in elliptical constructions such as la charmante lsquothe charming [one]rsquo
and following indefinite pronouns as in quelqursquoun de charmant lsquosomeone
charmingrsquo) demonstrate that post-N and predicate positions were generally
equally frequent and together constitute fully 75 percent of the ratio whereas
pre-N position never surpassed 27 percent across course levels Other
constructions were minimal (See Figure 3 available to online subscribers at
httpapplijoxfordjournalsorg)
This roughly 75 percentndash25 percent ratio was then reanalyzed in order to
consider the number of distinct lexemes (lsquotypesrsquo) that make up the
percentages We find for example that only 15 lexemes make up the 14
percent of pre-N adjective use at Level 1a This number gradually increased
with course level but only reached a total of 41 by Level 4 Post-N position
on the other hand admitted a much larger variety of adjectives in classroom
input from 71 at Level 1a to 323 by Level 4 (See Figure 4 available to online
subscribers at httpapplijoxfordjournalsorg) Students therefore got
relatively little input on pre-N position and what input they did get was
made up of a relatively small class of adjectives being used over and over
again and which already figure in the textbook-determined list of pre-N
adjectives in (4) (See Appendix A available to online subscribers at http
applijoxfordjournalsorg for a list of adjectives appearing in pre-N position
by course level)
Finally we find that very few adjectives varied in position during the 10
hours of in-class observation There was no variation attested at Level 1a
and only four to seven adjectives varied in position from Level 1b through to
Level 3 Moreover some of the input provided to learners in these courses
in the form of other studentsrsquo output was nonnativelike6 At the most
advanced undergraduate level learners got somewhat more input involving
296 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
variation in position where 20 adjectives did so (See Figure 5 available to
online subscribers at httpapplijoxfordjournalsorg) Again most (but not
all) of these adjectives already figure into the textbook-determined list of
variable adjectives in (5) (See Appendix B available to online subscribers at
httpapplijoxfordjournalsorg for a list of adjectives varying in position by
course level)
To summarize the observational data just presented the ambient language
of the classroom clearly mimics the pedagogical treatment that American
university-level classroom learners receive from textbooks Although
adjectives are in general a very frequent part of the input the rate of
pre-N position (at roughly 25 percent) is essentially static across levels
pointing to the fact that classroom input across levels is in no way
contradictory to what learners are initially taught Post-N position is basic
(unmarked) and pre-N position is exceptional (marked) The adjectives that
do appear in pre-N position in classroom input are in most cases those that
learners already expect to appear in that position (cf the list in (4) and
Appendix A) The few adjectives that appear in both positions are again in
most cases those that learners expect to do so (cf the list in (5) and
Appendix B)
By the end of the first-year level then both instruction and classroom
input would lead learners to expect that adjectives will appear in post-N
position with the notable exception of the class of adjectives in (4) (those in
(5) having not yet been introduced) By the second-year level it could be
argued that instruction provides examples in the form of so-called change of
meaning adjectives that variation in position is to be associated with
variation in interpretation (and perhaps by implication always to
interpretation) However learnersrsquo explicit knowledge of interpretative
differences as exemplified by the adjectives in (5) typically consists of
arbitrary and unsystematic English glosses By arbitrary I mean that
instruction never provides an adequate explanation for why the pre-N
position of cher should correspond to beloved or for its post-N position to
correspond to expensive7 Though telling students following Bragger and Rice
(1996) that the meaning in post-N position is lsquoconcretersquo and the meaning in
pre-N position is lsquofigurativersquo might (somehow) help it does not account for
other adjectives in (5) like prochain lsquonextrsquo This latter adjective also
demonstrates the fact that English glosses are unsystematic in that prochain
translates as lsquonextrsquo in either position
Before considering our final research question I note that no research on
adjective position in native French speaker discourse is available to date
An anonymous Applied Linguistics reviewer questions the validity of compar-
ing primarily spoken classroom discoursemdashor any other kind of spoken
discoursemdashto written texts without first demonstrating that adjective position
is insensitive to this difference in medium While recognizing the potentially
important distinction in medium and the need for further research in this
area the validity of comparing classroom input to written texts in the present
BRUCE ANDERSON 297
study lies in the fact that such a comparison roughly corresponds to the
curriculum of typical university-level courses an emphasis on listening
comprehension and oral communication in the first- and second-year
language courses leading to an emphasis on reading comprehension
(typically literature-based) and proficiency in writing in the third- and
fourth-year courses Were it to turn out that the degree of variation in
adjective position found in written texts is restricted to just that medium
revisions to pedagogy would necessarily be restricted to writing-intensive
(composition creative composition stylistics) courses at the third- and
fourth-year levels (I return to this issue in the discussion section)
LEARNER INTUITIONS OF ACCEPTABILITY
Thus far we have seen an incongruity between textbook presentations of
adjective position and authentic text sources across a number of written text
genres (research question 1) and that ambient language use involving
adjective position in the classroom is clearly more congruent with and might
therefore be taken to reinforce the explicit rules found in textbooks
(research question 2) We are now in a position to address our third research
questionmdashthe extent to which intuitions of acceptability among classroom
learners are biased by the congruity between pedagogical treatment and
classroom input frequency patterns An empirical study previously reported
in Anderson (2002 2007) can shed some light on this question
Methodology
Anderson (2002 2007) investigated the acquisition of various morphosyn-
tactic properties of French noun phrases by administering an acceptability
judgment task in which participants had to evaluate the appropriateness of
sentences featuring among other things the pre- versus post-N position of
adjectives Each of the adjectives selected were lsquohighly variablersquo with respect
to position (cf Table 1) yet crucially do not figure in textbook lists of
adjectives said to change meaning based on position (cf the list in (10))
These include lourd lsquoheavyrsquo violent lsquoviolentrsquo charmant lsquocharmingrsquo precieux
lsquopreciousrsquo epais lsquothickrsquo etrange lsquostrangersquo and delicieux lsquodeliciousrsquo (See
Appendix C available to online subscribers at httpapplijoxfordjournals
org for test sentences) Study participants included English-speaking learners
of French in the second third and fourth years of undergraduate study
(nfrac14 80) who were attending courses at the same institution at which the
observational data were collected a group of advanced learners (nfrac14 20)
consisting of graduate student instructors at that same institution as well as
secondary education teachers and a group of native French speakers (nfrac14 27)
roughly matched in age and education level
Each adjective appeared in test sentences once prenominally and once
postnominally with each test sentence preceded once by a context
298 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
motivating acceptance and once by a context motivating rejection of one or
the other adjective order for a total of four positionndashinterpretation pairings
per adjective These were randomly distributed across two testing
instruments administered a week apart After reading each context and
paired test sentence study participants were directed to check boxes
labeled lsquofinersquo lsquooddrsquo or lsquocannot decidersquo to the prompt This particular sentence
sounds ___ in this context
As an example each of the sentences in (7) featuring the adjective lourd
lsquoheavyrsquo was paired with each of two contexts
(7) a Pierre doit laisser la lourde valise avec son amib Pierre doit laisser la valise lourde a lrsquoaeroport
lsquoPierre has to leave the heavy suitcase with his friendatthe airportrsquo
In the first context motivating a unique noun-referent interpretation
there was one and only one suitcase which happened to be heavy and
which lsquoPierrersquo had to leave behind with his friend In the second context
motivating a multiple noun-referent interpretation there were a number of
suitcases only one of which was heavy and it was this one that lsquoPierrersquo had
to leave behind at the airport (See Appendix D available to online
subscribers at httpapplijoxfordjournalsorg for the complete version of
each context)
When viewed solely in terms of truth-conditional semantics each of the
four positionndashinterpretation pairings is compatible save for one pre-N
position (7a) in a multiple noun-referent context This is because the singular
definite article la and pre-N adjective lourde together presuppose a unique
suitcase in context (which happens to be heavy) This presupposition
does not hold when the adjective is in post-N position where all that is
required is that there be at least one such suitcase in context The testing
instrument then allows us to examine how classroom learners react to
an lsquoexpectedrsquo word order (ie sentences like (7b) in either context) as well
as to an lsquounexpectedrsquo but semantically and pragmatically appropriate
word order (ie sentences like (7a) but only in the unique noun-referent
context)
Results
As the mean acceptance rates from native French speakers in Table 2 show
truth-conditional semantics was not the only factor determining acceptance
rates Two of the pairings received essentially chance-level rates of
acceptance (pre-N position in a multiple noun-referent context at 49
percent and post-N position in a unique context at 52 percent) whereas the
other two pairings showed much clearer above- or below-chance rates (pre-N
position in a unique noun-referent context at 74 percent and post-N
position in a multiple noun-referent context at 37 percent)
BRUCE ANDERSON 299
As was argued in Anderson (2002 2007) results for each of the four
positionndashinterpretation pairings among the native French speakers are
explainable in terms of a disproportionate interaction between semantics and
pragmatic implicature in particular the Gricean maxim of quantity An
utterance should be no more and no less informative than is required for the
current purposes of the exchange (Grice 1975) The positionndashinterpretation
pairing that was lsquotruersquo in terms of semantics and also the most pragmatically
appropriate was accepted at the highest rate (74 percent) the pairing that
was also lsquotruersquo in terms of semantics but was pragmatically the least
appropriate was accepted at the lowest rate (37 percent)8
Do learner group response patterns show the same bias toward pragmatic
implicature in their acceptance rates The comparison of mean acceptance
rates between native French speakers and learner groups in Table 3 shows
that learners performed in a particular (and particularly non-Frenchlike) way
in which the interaction goes in the opposite direction favoring semantics
over pragmatics
Recall that the only positionndashinterpretation pairing that was semantically
incompatible was pre-N position (7a) in a multiple noun-referent context
given the presupposition of uniqueness of the noun referent All learner
groups did indeed accept this pairing at the lowest rate of the four and with
the exception of the second-year learners (at thorn9 percent) did so more
strongly than the native French speakers at mean acceptance rates that were
11ndash15 percent lower)9 However they also accepted at much lower mean
rates (12 to 23 percent) the most felicitous positionndashinterpretation pairing
for the native French speakers pre-N position in a unique noun-referent
context Recall also that post-N position (7b) was semantically compatible
under either interpretation All learner groups accepted these two positionndash
interpretation pairings at much higher rates than the native French speakers
(thorn23 to thorn31 percent) including the least felicitous positionndashinterpretation
pairing for the native French speakersmdashpost-N position in a multiple noun-
referent context This is precisely what one would expect from the
pedagogical treatment these learners are provided and the congruity of
that treatment with classroom input which under-represents actual usage in
Table 2 Native French speaker mean acceptance rates for the uniquenonunique noun-referent distinction
Prenominal (AN)order ()
Postnominal (NA)order ()
Unique noun referent 74 52
Nonunique noun referent 49 37
Source Adapted from Anderson 2002 2007
300 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
text-based corpora and as we have now seen differs from native speaker
intuitions of overall (semantico-pragmatic) acceptability
Although a bias toward or even strict adherence to explicit pedagogical
rules in evaluating such sentences may not seem all that surprising what does
stand out in the data from Anderson (2002 2007) is that these differential
rates in acceptance between native and nonnative speakers are as true for the
advanced group in that studymdashconsisting of graduate student instructors and
second education teachersmdashas they are for the intermediate (second-year)
learner group This would seem to indicate no increased sensitivity to
contextual (pragmatic) factors in adjective position over the course of studying
French even though the advanced group had been exposed to much more
authentic input through texts written for and by native French speakers (ie
the descriptive norm uncovered in our examination of authentic text corpora)
DISCUSSION
What should be clear from the findings presented here is that rules of French
adjective placement that one typically finds in (at least American) textbooks
though not plainly incorrect are not fully descriptively accurate either at
least with respect to written French across several text genres In particular
such rules portray pre-N and post-N positions as mutually exclusive for
almost all adjectives when in fact they are not
Similar discrepancies are often found between rules for and by native
speakers and what those speakers actually do One need look no further than
the French for an example with colloquial speech often at odds with the
prescriptive bon usage and the admonishments of the Academie francaise One
might therefore also expect to find a discrepancy between textbook rules and
what teachers do in the classroom andor what students find to be
lsquoacceptablersquo in terms of word order As we have seen however this is not
the case These rules are faithfully reflected in classroom language use by
instructors to such an extent that when asked to evaluate the acceptability
of lsquoexpectedrsquo and lsquounexpected but nativelikersquo word orders (and one would
presume in production as well) both instructors and students paid much
less attention to problems in pragmatic implicature than did native speakers
a situation which does not appear to change even in the face of counter-
examples occasionally provided in the textbooks they use (see (6) above) and
most certainly in the written texts that they read at later course levels As
pointed out by an anonymous reviewer this situation can be considered a
testament to the power of classroom instruction (instructor input in
particular) if instruction is changed he or she reasons it is likely that
classroom learners will reflect this change
But should instruction be changed If so who along the continuum from
the theorist-researcher to the teacher-practitioner should inform and guide
this change At which level In what way Such questions are at the very
heart of two on-going debates that I re-examine before proposing some
BRUCE ANDERSON 301
answers the pedagogical versus descriptive norm debate (eg Owen 1993
Widdowson 2000 and the collection of articles in Gass et al 2002) and
the interrelated native-speaker-usage-as-goal debate in foreign language
pedagogy (eg the collection of articles in Blyth 2003)
On the whole the approach to instruction on French adjective
position outlined earlier appears advantageous in its simplicity providing
easily remembered rules of thumb that learners can apply in language
production That this approach to instruction does not as we have seen
straightforwardly follow the descriptive norm is a situation that linguists
utilizing a corpus-based methodology believe can and should be remedied
Conrad (2000) for instance sees in corpus-based research an immediate way
in which linguistics can inform language pedagogy freeing language teachers
from lsquohav[ing] to rely on judgments of grammatical accuracy or on native
speakersrsquo intuition about what sounds naturalrsquo (2000 555) Although the
subsequent study by Biber and Reppen (2002) uses a disclaimer to the effect
that frequency should not be the sole factor in material design they
nevertheless suggest that lsquoa selective revision of pedagogy to reflect actual
use as shown by frequency studies could result in radical changes that
facilitate the learning process for studentsrsquo (2002 199) Owen (1993)
however provides examples of how a total reliance on a corpus does not
necessarily yield better observation and more importantly questions
whether better observation automatically equates to better explanation
(1993 168) (but see Francis and Sinclair 1994 for rebuttal) For his part
Widdowson (2000) objects to the kind of equation just described as a
case of linguistics applied a mistaken belief that what is textually attested
lsquouniquely represents real language and that this reality should define the
foreign language subjectrsquo (2000 8) (but see Stubbs 2001 for rebuttal) What
is needed according to Widdowson is not simply linguistics applied
to pedagogy but a true role for applied linguistics an approach to usage that
also recognizes native speakersrsquo awareness of and opinions about such usage
The need for a truly applied linguistics perspective on native speaker
usage (including frequency patterns) is embodied in a pedagogical norm
originally conceptualized in Valdman (1989) and elaborated upon in
Bardovi-Harlig and Gass (2002) as
[A] combination of language systems and forms selected by linguistsand pedagogues to serve as the immediate language target ortargets that learners seek to acquire during their language studyIn other words pedagogical norms represent a mid-point or seriesof mid-points for learners as they progress toward acquiringnative language norms [thus constituting] a form of languagethat is acceptable to native speakers but easier to learn than the fullnative language system (p 3)
The textbook word-order rules for French adjective position reviewed
earlier would first appear to constitute an appropriate pedagogical norm
302 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
under Bardovi-Harlig and Gassrsquos (2002) definition because the learner is able
to produce a subset of the totality of possible utterances deemed acceptable to
native speakers which may turn out to be even more the case in spoken than
in written French (though the difference in mediums remains to be studied)
The learner data presented here from Anderson (2002 2007) however
provides one piece of evidence that textbook rules and classroom input do not
constitute a sufficient pedagogical norm as learners have stopped in their
progress toward acquiring native speaker (ie descriptively accurate) norms
As the articles in Blyth (2003) attest however it is highly debatable
whether the lsquoNative Speakerrsquo (as an idealized abstraction) should be the
appropriate model for the contemporary learner For Kramsch (2002 2003)
too much of a focus on an authentic native speaker norm reduces
foreign language (FL) study to a constant lsquoapproximation to and conformity
with stereotypical native speakers thus often silencing other FL-specific
forms of language potentialrsquo (2002 60) Language learning in instructional
settings on my understanding of Kramschrsquos (2003) argument should not be
viewed solely in a teleological sensemdashas language acquisition directed
toward a fixed nativelike end-pointmdashbut rather as language appropriation
by learners in the sense that they are constructing a linguistic and even
cultural identity lsquoin the interstices of national languages and on the
margins of monolingual speakersrsquo territoriesrsquo (2003 260) The political nature
of the debate as highlighted by Kramsch becomes all the more apparent
when one additionally considers the status of the foreign language on
the world stage (eg French as an international language in retreat)
and how deeply the normative attitudes of its speakers run (eg Francersquos
three-century-long history of acceptance and maintenance of the standard
(Lodge 1993))
The foregoing synopsis of the debates surrounding native speaker usage as
both the goal of foreign language instruction and the extent to which
pedagogical norms need apply has highlighted the real world complexities
involved in what would otherwise seem to be a simple equation Revise
pedagogy to reflect actual use as shown by frequency studies and this
will result in a more targetlike performance on the part of learners So we
return then to the questions posed earlier (1) Should instruction be
changed (2) If so who along the continuum from the theorist-researcher to
the teacher-practitioner should inform and guide this change (3) At which
level (4) In what ways
With respect to our first question because one can empirically
demonstrate that the pedagogical norm for French adjective position
prevents learners from eventually acquiring native speaker (ie descriptively
accurate) norms instruction should be changed so thatmdashirrespective of
whether the attainment of such a norm is necessarily desirablemdashlearners
have both a choice in selecting one grammatical or lexical form over the other
and an awareness of the meaning potential of each choice (following Kramsch
2002 71)
BRUCE ANDERSON 303
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
THREE PERSPECTIVES ON FREQUENCY
Within linguistic theory frequency has long been associated with the notion of
markedness Greenberg (1966 67) in fact suggested that frequency is the
primary criterion for unmarked versus marked word order though one must be
careful to distinguish markedness in cross-linguistic terms from markedness
within a given language For example whereas van Riemsdijk (1978) argued
for preposition pied-piping in wh-questions (eg To whom did Cathy give a book)
as the more frequent and thus unmarked construction cross-linguistically Ross
(1977) observed that pied-piping has been undergoing neutralization in favor
of preposition stranding (eg Who[m] did Cathy give the book to) making it the
more frequent and thus unmarked construction in modern (colloquial) English
The relationship between frequency and markedness within linguistic
theory as exemplified above has in turn been investigated from a second
perspectivemdashthat of (second) language acquisition theorymdashby Mazurkewich
(1984) and Bardovi-Harlig (1987) Although Mazurkewichrsquos study appeared
to illustrate the strength of cross-linguistic markedness in determining
acquisitional sequences through empirical data suggesting the acquisition of
pied-piping before stranding among French- and Inuktitut-speaking ESL
learners Bardovi-Harligrsquos replication and re-interpretation of that study
showed just the opposite order illustrating the equal or greater importance of
the lsquodistribution of data available as inputrsquo (1987 400)1
Some twenty years later frequency effects in (second) language acquisition
continue to be addressed in the literature most particularly concerning the
extent to which first (L1) andor second (L2) language epistemology can be
defined in such terms Ellis (2002) for example provided a sweeping
examination of studies on input frequency effects in L1 processing claiming
that such effects are not simply a key determinant of acquisition but its sole
determinant Linguistic units are abstracted from language use and language
rules are in effect lsquostructural regularities that emerge from learnersrsquo lifetime
analysis of the distributional characteristics of the language inputrsquo (2002
144) This stands in stark contrast to so-called rationalist or innatist accounts
in which frequency effects are epiphenomenal (see for example Eubank and
Greggrsquos (2002) peer commentary of Ellis)
Bley-Vroman (2004) has taken an intermediate stance in arguing that
language acquisition at least in L1 involves both the lsquoformation of abstract
rules guided by UG and relatively independent of frequency effectsrsquo and
the lsquopiecemeal instance-based acquisition of particular structures which can
be substantially affected by the statistical structure of languagersquo (2004 269)
L2 acquisition according to Bley-Vroman appears to depart significantly
from this two-part process resulting in learner grammars that are based
significantly moremdashor perhaps whollymdashon lsquowhat they have heard and how
oftenrsquo (2004 263)
Bley-Vromanrsquos (2004) claims are partly based on data from Bley-Vroman
and Yoshinaga (2000) a study in which native English speakers and L1
BRUCE ANDERSON 287
Japanese learners of English were asked to judge the acceptability of multiple
wh-questions of the type Who ate what and Who sat where wherein
the sentence-final wh-word is a complement of the verb versus questions of
the type Who sang where wherein the sentence-final wh-word is an adjunct
Irrespective of this complementndashadjunct distinction the frequency of use of
the three question types was found to be quite low Only 122 examplars
beginning with who could be found in the 56 million word corpus
Bley-Vroman consulted 106 (or 87 percent) were specifically of the
who what type Native English speakers whose judgments were hypothe-
sized to follow from abstract rules favored both who wh[complement]
structures and disfavored the who wh[adjunct] structure despite the across-
the-board low frequency of all such structures in the corpus consulted The
learner group as expected favored the (relatively) predominant who what
type to the near exclusion of the other two seemingly paying no heed to the
complementndashadjunct distinction
Though certainly useful from the standpoint of L2 acquisition studies both
in terms of methodology and epistemological arguments frequency patterns
culled from authentic native-speaker corpora do not as Bley-Vroman (2004
270) pointed out necessarily reflect what learners are exposed to and it is
this latter type of corpus that is crucial in L2 acquisition studies because it
represents the input with which a learning algorithm would interact
The potential dichotomy between frequency patterns in native-speaker
usage and what classroom-based learners are exposed to (viewed broadly as
the sum total of textbook presentations authentic texts used in the
classroom and lsquoteacher talkrsquo) represents a third and especially applied
linguistic perspective on frequency An early study by Holmes (1988) on the
presentation of expressions of doubt and uncertainty in ESL textbooks for
example found that textbooks largely favored modal verbs while
simultaneously neglecting alternatives such as adjectives tag questions and
fallndashrise intonation (1988 40) This led her to raise concerns that a simplicity
criterion in the formulation of textbook rules may come at the cost of
naturalness and thus she concurred with Kasper (1979 277) that the
reduced range of expressions of modality witnessed in learnersrsquo production
can in some cases be viewed as teaching induced
More recent cases of mismatch have come to light through the corpus-
based methodology of Biber (1988 1995) and Biber et al (1998) and the
corpus-based Longman Grammar of Spoken and Written English (Biber et al
1999) A recent study by Biber and Reppen (2002) for instance examined
six ESL textbooks all of which presented attributive adjectives as the
principle means of modifying nouns Two-thirds of these textbooks included
some treatment of participial adjectives (those ending in -ed and -ing)
but only one discussed the use of nouns rather than adjectives as
prenominal modifiers (eg tomato sauce pencil case) However Biber et alrsquos
(1999) analysis of Englishmdashbased on a corpus of conversational transcripts
fictional literature newspaper writing and academic prosemdashshowed
288 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that (a) participial adjectives are quite rare across the four registers
(b) common attributive adjectives are in fact used the most in conversation
to modify nouns but most nouns actually have no modification at all in that
register and (c) nouns as prenominal modifiers are quite frequent in
newspaper writing and nearly as common as adjectives
Such corpus-based findings Biber and Reppen (2002) suggested should
result in participial forms being given little attention in intermediate to
advanced ESL materials and nouns as prenominal modifiers (and the various
and complex meaning relations they evince) being given more prominent
coverage In other words a descriptive norm for noun modification established
through corpus-based research should serve asmdashor at the very least
significantly inform the elaboration ofmdasha pedagogical norm (See OrsquoConnor Di
Vito 1991 for a similar argument based on frequency data from French) This
seemingly simple equation is not without its critics however (eg Owen 1993
Widdowson 2000) and we re-examine their criticisms in the discussion section
in light of the findings presented here on French adjective position
RESEARCH QUESTIONS
Although the potential for mismatch between native-speaker usage and
pedagogical materials has received some attention in the field of applied
linguistics the present study is unique in also offering a glimpse of the
potential for mismatch between pedagogical materials and ambient language
use in the classroom and as a result the potential for mismatch between
native speaker and L2 learner intuitions of acceptable word order(s) The
following three research questions are therefore addressed
1 To what extent are textbook presentations of French adjective position
congruent with authentic text sources across a number of genres
2 What is the nature of ambient language use involving adjective position
in the L2 French classroom and to what extent is it congruent with
textbook presentations of adjective position
3 To what extent are intuitions of acceptability among classroom learners
biased by any such (in)congruities
Because each of the three research questions represents a mini-study in
and of itself involving a different study methodology and type of data
analysis I address each question in its own section providing first an
overview of the methodology used and then a presentation of results
FRENCH ADJECTIVE POSITION IN TEXTBOOKS ANDAUTHENTIC SOURCES
The complexity of English adjectival modification pointed out in Biber and
Reppen (2002) is compounded in French by the fact that adjectives appear
BRUCE ANDERSON 289
in both adnominal positions prenominally (pre-N) and postnominally
(post-N) following certain lexical and contextual (pragmatic) constraints as
to their interpretation in one position versus the other (see eg Delmonier
1980 for a general review Delbecque 1990 for an account within cognitive
grammar Anderson 2002 for a compositional account within generative
syntax) This state of affairs leads us to question whether the traditional
textbook rules relating to adjectival modification in French reflect
attested native-speaker language use any better than that uncovered by
Biber and Reppen in authentic English-language texts across a number of
genres
English-speaking learners of French at least in the typical American
university setting are provided with three adjective placement rules
which take the form of a general rule with two exceptions Most adjectives
in French are said to be post-N (as in (1)) though a few are pre-N (as in (2))
and a few more are said to change meaning based on their position
(as in (3))
(1) une voiture rougelsquoa red carrsquo
(2) une petite maisonlsquoa small housersquo
(3) a un bijou cher un cher bijoulsquoan expensive jewelrsquo lsquoa cherished jewelrsquob une eglise ancienne une ancienne eglise
lsquoan ancient churchrsquo lsquoa former churchrsquo
First- and second-year textbooks of French published in the USA are
strikingly similar in the statement of such rules and in their order of
presentation though they differ to some degree as to which adjectives are to
be included in the (putative) lexical classes constituting exceptions to the
general post-N rule The textbooks used in the classrooms observed for the
present study are no different In the first-year textbook Chez nous (Valdman
and Pons 1997) students are first told that most adjectives follow the noun
(1997 73) but that the twelve adjectives listed in (4) typically and
exceptionally precede the noun (1997 241)
(4) a jeune vieux lsquoyoung oldrsquob nouveau lsquonewrsquoc petit grand gros lsquosmalllittle bigtall largersquod beau lsquohandsomersquoe joli lsquoprettyrsquof bon mauvais lsquogood badrsquog premier dernier lsquofirst lastrsquo
In the second-year textbook Quant a moi (Bragger and Rice 1996) the basic
post-N rule is reviewed as is the class of adjectives that exceptionally appear
prenominally (1996 33) Note however that this class of exceptions has
eleven members instead of twelve having lost some members from the Chez
290 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
Nous list in (4) (eg bon premier dernier gros) but gaining others (eg long
lsquolongrsquo court lsquoshortrsquo and autre lsquootherrsquo)
Bragger and Rice (1996) then present the second exception to the basic
post-N rule wherein students are told that lsquoa number of adjectives change
meaning depending on whether theyrsquore placed after or before a noun
In general an adjective placed after the noun retains its basic concrete
meaning The same adjective placed before the noun is used in an abstract or
figurative mannerrsquo (1996 34 emphasis in original) The seven adjectives that
apparently fall into this class are listed in (5) The first English gloss captures
the meaning derived from pre-N position the second gloss that derived from
post-N position Note that Bragger and Rice put the adjective grand in both
lists of exceptions In other words grand is apparently exceptionally pre-N
and exceptionally variable at the same time2
(5) ADJECTIVE PRE-N POST-Na ancien lsquoformerrsquo vs lsquoold ancientrsquob cher lsquodear well-lovedrsquo vs lsquoexpensiversquoc dernier lsquolast latest (in a series)rsquo vs lsquolast (before this one)rsquod grand lsquogreatrsquo vs lsquotall largersquoe pauvre lsquopoor (unfortunate)rsquo vs lsquopoor (not rich)rsquof prochain lsquonext (in a series)rsquo vs lsquonext (after this one)rsquog propre lsquoownrsquo vs lsquocleanrsquo
To summarize the pedagogical treatment of French adjective position
commonly found in university-level American textbooks presents post-N
position as the most frequent and typical (frac14unmarked) adjective order
the less frequent and exceptional (frac14marked) pre-N order is either required
when the adjective is a member of a seemingly fixed lexical class such as
the list in (4) or licensed by a particular interpretation that differs from that
obtained in post-N position such as the list in (5) This difference in
interpretation is often though not always captured through distinct English
glosses
Methodology
In order to determine the extent to which the pedagogical treatment just
summarized is congruent with authentic French texts three corpora were
consulted Wilmetrsquos (1980) corpus of literary texts (approximately 10 million
words 3835 adjectives) Forsgrenrsquos (1978) corpus of newspaper texts from Le
Monde and LrsquoExpress (1 million words 829 adjectives) and Larssonrsquos (1993)
corpus of travel catalogues travel guides and other non-literary prose (15
million words 113 adjectives)3 Using Wilmetrsquos corpus as the point of
comparison given that it contains the highest number of discrete lexemes
(3835 adjectives) an arbitrary cut-off point of 25 attestations or more was
used creating a list of 205 adjectives4 The total number of attestations pre-N
attestations and post-N attestations for each of the 205 adjectives across
BRUCE ANDERSON 291
the three corpora was then determined by adding Forsgrenrsquos and Larssonrsquos
data to that found in Wilmet if available
Results
The 205 most frequently attested adjectives in French texts appear to fall
along a continuum at one end of which we find 68 adjectives (or 332
percent) appearing either only in post-N position (nfrac1434) or typically in that
position (nfrac14 34) and at the other end of which we find 15 adjectives (or 73
percent) typically appearing in pre-N position as shown in Table 1 Note that
there are no adjectives appearing only in pre-N position The remaining 122
adjectives (or 595 percent) are regularly attested in both positions with
some having what we might call an affinity for post-N position (nfrac1441) or
pre-N position (nfrac14 29) but others being highly variable (nfrac1452)
Reminiscent of Holmesrsquo (1988) critique of the simplicity versus naturalness
of textbook rules such data raise some problems for the seemingly
straightforward fixed-order rules of adjective position exemplified in (1)ndash(3)
and the associated lists in (4) and (5) A textbook rule to the effect that
adjectives typically appear after the noun in Frenchmdasha word order that
clearly contrasts with what one finds in Englishmdashis only partly descriptively
Table 1 Frequency groupings of adjective position in written French texts
Frequency grouping Type count(nfrac14 205)
Percent oftotal
Examples
Exclusively pre-N (0post-N position)
0 0 mdash
Typically pre-N (below10 post-N position)
15 75 bon lsquogoodrsquo fameuxlsquo(in)famousrsquo petit lsquosmallrsquo
Variable but with anaffinity for pre-Nposition (10ndash39post-N position)
29 14 charmant lsquocharmingrsquo etrangelsquostrangersquo miserable lsquomiserablersquo
Highly variable (40ndash59post-N position)
52 255 delicieux lsquodeliciousrsquo joyeuxlsquojoyousrsquo puissant lsquopowerfulrsquo
Variable but with anaffinity for post-Nposition (60ndash90post-N position)
41 20 amer lsquobitterrsquo tiede lsquolukewarmrsquoserieux lsquoseriousrsquo
Typically post-N (above90 post-N)
34 17 blanc lsquowhitersquo inconnulsquounknownrsquo sauvage lsquowildrsquo
Exclusively post-N(100 post-Nposition)
34 17 allemand lsquoGermanrsquo dorelsquogoldenrsquo quotidien lsquodailyrsquo
292 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
accurate It is accurate in the sense that the ends of the continuum are
not equivalent in their percentages There are many more adjectives
appearing either only or typically in post-N position than is the case for pre-N
position It is inaccurate in the sense that it ignores the distributional
behavior of 595 percent of the most frequently used adjectives which can
appear in both positions French is not as the rule implies the mirror
opposite of English but instead is more accurately viewed as an lsquoANArsquo
(adjectivendashnounndashadjective) language as opposed to both lsquosystematic ANrsquo
languages like English and German and lsquosystematic NArsquo languages like
Indonesian and Thai (Cinque 1994 99ndash101)
The data also lead us to question the criteria textbooks use for including
certain adjectives in the two lists in (4) and (5) For the so-called prenominal
adjectives a frequency of 90 percent or greater in that position would
establish that adjectives such as excellent fameux lsquo(in)famousrsquo moindre
lsquoslightestrsquo and vrai lsquorealtruersquo need to be added That list itself implies a fixed
lexical class to be memorized though in reality it does nothing more
than provide a handful of frequently attested adjectives (nfrac14 15 [75 percent])
that typically appear just in that position Many more adjectives can
appear in pre-N position as well For the so-called change-of-meaning
adjectives the data show that the list in (5) falls woefully short in terms of
representing those adjectives that actually do regularly appear in both
positions (nfrac14 52)
Even if it were to be established that such rules are accurate with respect
to colloquial French as spoken between native speakers and thus appropriate
for oral proficiency-oriented classrooms such rules obviously fail to capture
the state of affairs in written texts particularly literature to which learners at
later course levels are exposed Yet even students at earlier course levels
occasionally find counter-examples to what they are taught The four cases of
lsquounexpectedrsquo pre-N position in (6) come from the cultural readings of two
first-year and two second-year US textbooks respectively
(6) a La France a un excellent systeme de transports publics(Magnan et al 2002 223)lsquoFrance has an excellent system of public transportrsquo
b Cela explique le grand nombre de magnifiques chateauxdans la region (Valdman and Pons 1997 395)lsquoThis explains the great number of magnificent chateaux inthe regionrsquo
c [U]ne veritable explosion de la technologie (St Onge etal 2003 71) lsquoa veritable explosion of technologyrsquo
d Lrsquohistoire est une douloureuse suite de tragedieshumaines (Oates and Dubois 2003 208)lsquoIts history is a painful series of human tragediesrsquo
One might wish to argue that the examples provided in (6) demonstrate
that inputmdashat least classroom-related inputmdashis not as contradictory to
descriptive norms as argued in this section Examples of the type in (6)
BRUCE ANDERSON 293
however are only occasionally found in textbook reading materials they are
included here to demonstrate that even when making a concerted effort to
write level-appropriate cultural readings textbook authors sometimes break
the very grammar rules they present elsewhere in the textbook suggesting a
true disparity between rules and lsquonaturalisticrsquo usage5 (I return to the issue of
descriptively accurate statements about French adjective position in the
discussion section)
FREQUENCY OF ADJECTIVE USE AND POSITIONIN CLASSROOM INPUT
Methodology
In order to get a general sense of how well the classroom setting as primary
input source reflects either textbook rules or the corpora-based descriptive
norm just summarized frequency counts of adjective type (ie discrete
lexemes) and their position in classroom input were collected by the
researcher from 60 hours of in-class observations of university-level French
courses at a large Midwestern American university Ten hours were spent in
each of six courses spanning the curriculummdashfrom the four semesters
comprising the first- and second-year language skills curriculum (referred to
as Levels 1a 1b 2a and 2b below) to a third-year conversation course
(Level 3) and a fourth-year lecture course on French civilization (Level 4)
Each of these courses was taught almost exclusively in French by a mix of
native and nonnative instructors none of whom were aware of the purpose
of this observation period Class observations were conducted on days when
adjective position was not the focus of explicit instruction as this would have
more than likely biased the frequency counts
The first- and second-year courses consisted of explicit grammar
instruction drills and communicative activities normally involving students
in pair or group work The conversation and civilization courses involved
no explicit grammar instruction the former involved mostly teacherndash
student and studentndashstudent oral communication based on previous course
readings and video segments whereas the latter was almost exclusively
devoted to instructor-delivered lectures Error correction was minimal in all
courses and no case of error correction involving adjective position was
witnessed during the observation period (despite the fact that some errors
were made)
Oral input on adjective position in all courses came from one of three
sources the instructor audio aides (such as cassette and videotape) and
the students themselves Written input in the first- and second-year
courses came from two sources visual aids (such as handouts transparencies
writing on the blackboard) and passages from the textbook that were read
silently or aloud in class At Level 3 written materials included authentic
newspaper and magazine articles in addition to visual aides at Level 4
294 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
a number of works of French literature were read outside of class Note
that written input in these latter two course levels is qualitatively the same
as that used in the corpora compiled by Forsgren (1978) Wilmet (1980) and
Larsson (1993)
At issue during this period of investigation was classroom input that could
be seen andor heard by all students in the classroom (as well as the
researcher) Input from students involved in pair and group work was
therefore excluded from analysis Material to be read outside of class was
also excluded as there was no guarantee that all students had actually
read such material Although the totality of inputmdashincluding sources of input
the learner might have voluntarily sought out outside of classmdashcould
obviously never be exhaustively tallied the exclusion of certain forms of
input was first and foremost a matter of logistics (ie not having the
appropriate funds equipment facilities and time to sound record then
transcribe all classroom input from every participant in every situation of
use) Despite this drawback I see no reason to suspect that student-delivered
input to other students during pair- or group-work would be qualitatively
different (at least in terms of adjective position) from the input delivered by
students that was audible to the entire class this latter source of input
already constitutes 10ndash30 percent of all sources of input tallied as part of the
observation period
Each instance of adjective use during each class session was recorded with
pen and paper by the researcher A coding scheme was devised whereby an
11 17-inch sheet of paper was divided into quadrants labeled according to
the source of the input lsquoinstructorrsquo lsquostudentrsquo lsquoaudiorsquo and lsquowrittenrsquo Each
quadrant was then subdivided according to the position of adjectival
modification lsquopre-Nrsquo lsquopost-Nrsquo lsquopredicatersquo and lsquootherrsquo (to be described
below) Input of the form une bonne idee lsquoa good idearsquo by the instructor was
therefore recorded as an instance of the adjective bonne in the instructor
quadrant pre-N sub-quadrant Hatch marks next to the adjective bonne were
used for any further exemplars of this adjective (including its masculine form
bon) in this position from this source I acknowledge that mere human error
caused by fatigue etc may have resulted in tallies that are less accurate than
if each session were recorded on tape Though the results presented below
should be considered with some caution then occasional errors could not
have reached a significant enough level to detract from the highly salient
patterns of usage that we will find
The resulting sixty sheets each representing one 50-minute class session
were then tallied in order to determine (a) the relative contribution of the
four sources of input (b) the average token (exemplar) count of adjectives
per 50-minute class period (c) the ratio of pre-N to post-Npredicate tokens
(d) the type count (number of discrete lexemes) entering into pre-N and
post-N positions and (e) the frequency of variation in positionmdashthat is the
extent to which a given adjective appeared at least once in both pre-N and
post-N position during the ten hours of observation
BRUCE ANDERSON 295
Results
The instructor not surprisingly was the primary provider of input in the
classroom (See Figure 1 available to online subscribers at http
applijoxfordjournalsorg) This was especially the case in the first-year
courses (Levels 1a 1b) the second- and third-year courses (Levels 2a 2b 3)
showed a greater proportion of input from students and from audio aides
The instructor was by far the predominant source of input in the fourth-year
course (Level 4) given that it involved instructor-delivered lectures on
French civilization with minimal class participation on the part of students
Concerning the average number of times an adjective occurred in the
input during any one 50-minute class period we find a general increase with
level from 42 tokens at Level 1a to 156 tokens per class at Level 4 (See
Figure 2 available to online subscribers at httpapplijoxfordjournalsorg)
Students therefore heard an adjective used anywhere from one to three
times per minute on average
Aggregate data on the frequency with which adjectives appeared in pre-N
post-N and predicate positions or in other constructions (eg used in
isolation in elliptical constructions such as la charmante lsquothe charming [one]rsquo
and following indefinite pronouns as in quelqursquoun de charmant lsquosomeone
charmingrsquo) demonstrate that post-N and predicate positions were generally
equally frequent and together constitute fully 75 percent of the ratio whereas
pre-N position never surpassed 27 percent across course levels Other
constructions were minimal (See Figure 3 available to online subscribers at
httpapplijoxfordjournalsorg)
This roughly 75 percentndash25 percent ratio was then reanalyzed in order to
consider the number of distinct lexemes (lsquotypesrsquo) that make up the
percentages We find for example that only 15 lexemes make up the 14
percent of pre-N adjective use at Level 1a This number gradually increased
with course level but only reached a total of 41 by Level 4 Post-N position
on the other hand admitted a much larger variety of adjectives in classroom
input from 71 at Level 1a to 323 by Level 4 (See Figure 4 available to online
subscribers at httpapplijoxfordjournalsorg) Students therefore got
relatively little input on pre-N position and what input they did get was
made up of a relatively small class of adjectives being used over and over
again and which already figure in the textbook-determined list of pre-N
adjectives in (4) (See Appendix A available to online subscribers at http
applijoxfordjournalsorg for a list of adjectives appearing in pre-N position
by course level)
Finally we find that very few adjectives varied in position during the 10
hours of in-class observation There was no variation attested at Level 1a
and only four to seven adjectives varied in position from Level 1b through to
Level 3 Moreover some of the input provided to learners in these courses
in the form of other studentsrsquo output was nonnativelike6 At the most
advanced undergraduate level learners got somewhat more input involving
296 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
variation in position where 20 adjectives did so (See Figure 5 available to
online subscribers at httpapplijoxfordjournalsorg) Again most (but not
all) of these adjectives already figure into the textbook-determined list of
variable adjectives in (5) (See Appendix B available to online subscribers at
httpapplijoxfordjournalsorg for a list of adjectives varying in position by
course level)
To summarize the observational data just presented the ambient language
of the classroom clearly mimics the pedagogical treatment that American
university-level classroom learners receive from textbooks Although
adjectives are in general a very frequent part of the input the rate of
pre-N position (at roughly 25 percent) is essentially static across levels
pointing to the fact that classroom input across levels is in no way
contradictory to what learners are initially taught Post-N position is basic
(unmarked) and pre-N position is exceptional (marked) The adjectives that
do appear in pre-N position in classroom input are in most cases those that
learners already expect to appear in that position (cf the list in (4) and
Appendix A) The few adjectives that appear in both positions are again in
most cases those that learners expect to do so (cf the list in (5) and
Appendix B)
By the end of the first-year level then both instruction and classroom
input would lead learners to expect that adjectives will appear in post-N
position with the notable exception of the class of adjectives in (4) (those in
(5) having not yet been introduced) By the second-year level it could be
argued that instruction provides examples in the form of so-called change of
meaning adjectives that variation in position is to be associated with
variation in interpretation (and perhaps by implication always to
interpretation) However learnersrsquo explicit knowledge of interpretative
differences as exemplified by the adjectives in (5) typically consists of
arbitrary and unsystematic English glosses By arbitrary I mean that
instruction never provides an adequate explanation for why the pre-N
position of cher should correspond to beloved or for its post-N position to
correspond to expensive7 Though telling students following Bragger and Rice
(1996) that the meaning in post-N position is lsquoconcretersquo and the meaning in
pre-N position is lsquofigurativersquo might (somehow) help it does not account for
other adjectives in (5) like prochain lsquonextrsquo This latter adjective also
demonstrates the fact that English glosses are unsystematic in that prochain
translates as lsquonextrsquo in either position
Before considering our final research question I note that no research on
adjective position in native French speaker discourse is available to date
An anonymous Applied Linguistics reviewer questions the validity of compar-
ing primarily spoken classroom discoursemdashor any other kind of spoken
discoursemdashto written texts without first demonstrating that adjective position
is insensitive to this difference in medium While recognizing the potentially
important distinction in medium and the need for further research in this
area the validity of comparing classroom input to written texts in the present
BRUCE ANDERSON 297
study lies in the fact that such a comparison roughly corresponds to the
curriculum of typical university-level courses an emphasis on listening
comprehension and oral communication in the first- and second-year
language courses leading to an emphasis on reading comprehension
(typically literature-based) and proficiency in writing in the third- and
fourth-year courses Were it to turn out that the degree of variation in
adjective position found in written texts is restricted to just that medium
revisions to pedagogy would necessarily be restricted to writing-intensive
(composition creative composition stylistics) courses at the third- and
fourth-year levels (I return to this issue in the discussion section)
LEARNER INTUITIONS OF ACCEPTABILITY
Thus far we have seen an incongruity between textbook presentations of
adjective position and authentic text sources across a number of written text
genres (research question 1) and that ambient language use involving
adjective position in the classroom is clearly more congruent with and might
therefore be taken to reinforce the explicit rules found in textbooks
(research question 2) We are now in a position to address our third research
questionmdashthe extent to which intuitions of acceptability among classroom
learners are biased by the congruity between pedagogical treatment and
classroom input frequency patterns An empirical study previously reported
in Anderson (2002 2007) can shed some light on this question
Methodology
Anderson (2002 2007) investigated the acquisition of various morphosyn-
tactic properties of French noun phrases by administering an acceptability
judgment task in which participants had to evaluate the appropriateness of
sentences featuring among other things the pre- versus post-N position of
adjectives Each of the adjectives selected were lsquohighly variablersquo with respect
to position (cf Table 1) yet crucially do not figure in textbook lists of
adjectives said to change meaning based on position (cf the list in (10))
These include lourd lsquoheavyrsquo violent lsquoviolentrsquo charmant lsquocharmingrsquo precieux
lsquopreciousrsquo epais lsquothickrsquo etrange lsquostrangersquo and delicieux lsquodeliciousrsquo (See
Appendix C available to online subscribers at httpapplijoxfordjournals
org for test sentences) Study participants included English-speaking learners
of French in the second third and fourth years of undergraduate study
(nfrac14 80) who were attending courses at the same institution at which the
observational data were collected a group of advanced learners (nfrac14 20)
consisting of graduate student instructors at that same institution as well as
secondary education teachers and a group of native French speakers (nfrac14 27)
roughly matched in age and education level
Each adjective appeared in test sentences once prenominally and once
postnominally with each test sentence preceded once by a context
298 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
motivating acceptance and once by a context motivating rejection of one or
the other adjective order for a total of four positionndashinterpretation pairings
per adjective These were randomly distributed across two testing
instruments administered a week apart After reading each context and
paired test sentence study participants were directed to check boxes
labeled lsquofinersquo lsquooddrsquo or lsquocannot decidersquo to the prompt This particular sentence
sounds ___ in this context
As an example each of the sentences in (7) featuring the adjective lourd
lsquoheavyrsquo was paired with each of two contexts
(7) a Pierre doit laisser la lourde valise avec son amib Pierre doit laisser la valise lourde a lrsquoaeroport
lsquoPierre has to leave the heavy suitcase with his friendatthe airportrsquo
In the first context motivating a unique noun-referent interpretation
there was one and only one suitcase which happened to be heavy and
which lsquoPierrersquo had to leave behind with his friend In the second context
motivating a multiple noun-referent interpretation there were a number of
suitcases only one of which was heavy and it was this one that lsquoPierrersquo had
to leave behind at the airport (See Appendix D available to online
subscribers at httpapplijoxfordjournalsorg for the complete version of
each context)
When viewed solely in terms of truth-conditional semantics each of the
four positionndashinterpretation pairings is compatible save for one pre-N
position (7a) in a multiple noun-referent context This is because the singular
definite article la and pre-N adjective lourde together presuppose a unique
suitcase in context (which happens to be heavy) This presupposition
does not hold when the adjective is in post-N position where all that is
required is that there be at least one such suitcase in context The testing
instrument then allows us to examine how classroom learners react to
an lsquoexpectedrsquo word order (ie sentences like (7b) in either context) as well
as to an lsquounexpectedrsquo but semantically and pragmatically appropriate
word order (ie sentences like (7a) but only in the unique noun-referent
context)
Results
As the mean acceptance rates from native French speakers in Table 2 show
truth-conditional semantics was not the only factor determining acceptance
rates Two of the pairings received essentially chance-level rates of
acceptance (pre-N position in a multiple noun-referent context at 49
percent and post-N position in a unique context at 52 percent) whereas the
other two pairings showed much clearer above- or below-chance rates (pre-N
position in a unique noun-referent context at 74 percent and post-N
position in a multiple noun-referent context at 37 percent)
BRUCE ANDERSON 299
As was argued in Anderson (2002 2007) results for each of the four
positionndashinterpretation pairings among the native French speakers are
explainable in terms of a disproportionate interaction between semantics and
pragmatic implicature in particular the Gricean maxim of quantity An
utterance should be no more and no less informative than is required for the
current purposes of the exchange (Grice 1975) The positionndashinterpretation
pairing that was lsquotruersquo in terms of semantics and also the most pragmatically
appropriate was accepted at the highest rate (74 percent) the pairing that
was also lsquotruersquo in terms of semantics but was pragmatically the least
appropriate was accepted at the lowest rate (37 percent)8
Do learner group response patterns show the same bias toward pragmatic
implicature in their acceptance rates The comparison of mean acceptance
rates between native French speakers and learner groups in Table 3 shows
that learners performed in a particular (and particularly non-Frenchlike) way
in which the interaction goes in the opposite direction favoring semantics
over pragmatics
Recall that the only positionndashinterpretation pairing that was semantically
incompatible was pre-N position (7a) in a multiple noun-referent context
given the presupposition of uniqueness of the noun referent All learner
groups did indeed accept this pairing at the lowest rate of the four and with
the exception of the second-year learners (at thorn9 percent) did so more
strongly than the native French speakers at mean acceptance rates that were
11ndash15 percent lower)9 However they also accepted at much lower mean
rates (12 to 23 percent) the most felicitous positionndashinterpretation pairing
for the native French speakers pre-N position in a unique noun-referent
context Recall also that post-N position (7b) was semantically compatible
under either interpretation All learner groups accepted these two positionndash
interpretation pairings at much higher rates than the native French speakers
(thorn23 to thorn31 percent) including the least felicitous positionndashinterpretation
pairing for the native French speakersmdashpost-N position in a multiple noun-
referent context This is precisely what one would expect from the
pedagogical treatment these learners are provided and the congruity of
that treatment with classroom input which under-represents actual usage in
Table 2 Native French speaker mean acceptance rates for the uniquenonunique noun-referent distinction
Prenominal (AN)order ()
Postnominal (NA)order ()
Unique noun referent 74 52
Nonunique noun referent 49 37
Source Adapted from Anderson 2002 2007
300 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
text-based corpora and as we have now seen differs from native speaker
intuitions of overall (semantico-pragmatic) acceptability
Although a bias toward or even strict adherence to explicit pedagogical
rules in evaluating such sentences may not seem all that surprising what does
stand out in the data from Anderson (2002 2007) is that these differential
rates in acceptance between native and nonnative speakers are as true for the
advanced group in that studymdashconsisting of graduate student instructors and
second education teachersmdashas they are for the intermediate (second-year)
learner group This would seem to indicate no increased sensitivity to
contextual (pragmatic) factors in adjective position over the course of studying
French even though the advanced group had been exposed to much more
authentic input through texts written for and by native French speakers (ie
the descriptive norm uncovered in our examination of authentic text corpora)
DISCUSSION
What should be clear from the findings presented here is that rules of French
adjective placement that one typically finds in (at least American) textbooks
though not plainly incorrect are not fully descriptively accurate either at
least with respect to written French across several text genres In particular
such rules portray pre-N and post-N positions as mutually exclusive for
almost all adjectives when in fact they are not
Similar discrepancies are often found between rules for and by native
speakers and what those speakers actually do One need look no further than
the French for an example with colloquial speech often at odds with the
prescriptive bon usage and the admonishments of the Academie francaise One
might therefore also expect to find a discrepancy between textbook rules and
what teachers do in the classroom andor what students find to be
lsquoacceptablersquo in terms of word order As we have seen however this is not
the case These rules are faithfully reflected in classroom language use by
instructors to such an extent that when asked to evaluate the acceptability
of lsquoexpectedrsquo and lsquounexpected but nativelikersquo word orders (and one would
presume in production as well) both instructors and students paid much
less attention to problems in pragmatic implicature than did native speakers
a situation which does not appear to change even in the face of counter-
examples occasionally provided in the textbooks they use (see (6) above) and
most certainly in the written texts that they read at later course levels As
pointed out by an anonymous reviewer this situation can be considered a
testament to the power of classroom instruction (instructor input in
particular) if instruction is changed he or she reasons it is likely that
classroom learners will reflect this change
But should instruction be changed If so who along the continuum from
the theorist-researcher to the teacher-practitioner should inform and guide
this change At which level In what way Such questions are at the very
heart of two on-going debates that I re-examine before proposing some
BRUCE ANDERSON 301
answers the pedagogical versus descriptive norm debate (eg Owen 1993
Widdowson 2000 and the collection of articles in Gass et al 2002) and
the interrelated native-speaker-usage-as-goal debate in foreign language
pedagogy (eg the collection of articles in Blyth 2003)
On the whole the approach to instruction on French adjective
position outlined earlier appears advantageous in its simplicity providing
easily remembered rules of thumb that learners can apply in language
production That this approach to instruction does not as we have seen
straightforwardly follow the descriptive norm is a situation that linguists
utilizing a corpus-based methodology believe can and should be remedied
Conrad (2000) for instance sees in corpus-based research an immediate way
in which linguistics can inform language pedagogy freeing language teachers
from lsquohav[ing] to rely on judgments of grammatical accuracy or on native
speakersrsquo intuition about what sounds naturalrsquo (2000 555) Although the
subsequent study by Biber and Reppen (2002) uses a disclaimer to the effect
that frequency should not be the sole factor in material design they
nevertheless suggest that lsquoa selective revision of pedagogy to reflect actual
use as shown by frequency studies could result in radical changes that
facilitate the learning process for studentsrsquo (2002 199) Owen (1993)
however provides examples of how a total reliance on a corpus does not
necessarily yield better observation and more importantly questions
whether better observation automatically equates to better explanation
(1993 168) (but see Francis and Sinclair 1994 for rebuttal) For his part
Widdowson (2000) objects to the kind of equation just described as a
case of linguistics applied a mistaken belief that what is textually attested
lsquouniquely represents real language and that this reality should define the
foreign language subjectrsquo (2000 8) (but see Stubbs 2001 for rebuttal) What
is needed according to Widdowson is not simply linguistics applied
to pedagogy but a true role for applied linguistics an approach to usage that
also recognizes native speakersrsquo awareness of and opinions about such usage
The need for a truly applied linguistics perspective on native speaker
usage (including frequency patterns) is embodied in a pedagogical norm
originally conceptualized in Valdman (1989) and elaborated upon in
Bardovi-Harlig and Gass (2002) as
[A] combination of language systems and forms selected by linguistsand pedagogues to serve as the immediate language target ortargets that learners seek to acquire during their language studyIn other words pedagogical norms represent a mid-point or seriesof mid-points for learners as they progress toward acquiringnative language norms [thus constituting] a form of languagethat is acceptable to native speakers but easier to learn than the fullnative language system (p 3)
The textbook word-order rules for French adjective position reviewed
earlier would first appear to constitute an appropriate pedagogical norm
302 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
under Bardovi-Harlig and Gassrsquos (2002) definition because the learner is able
to produce a subset of the totality of possible utterances deemed acceptable to
native speakers which may turn out to be even more the case in spoken than
in written French (though the difference in mediums remains to be studied)
The learner data presented here from Anderson (2002 2007) however
provides one piece of evidence that textbook rules and classroom input do not
constitute a sufficient pedagogical norm as learners have stopped in their
progress toward acquiring native speaker (ie descriptively accurate) norms
As the articles in Blyth (2003) attest however it is highly debatable
whether the lsquoNative Speakerrsquo (as an idealized abstraction) should be the
appropriate model for the contemporary learner For Kramsch (2002 2003)
too much of a focus on an authentic native speaker norm reduces
foreign language (FL) study to a constant lsquoapproximation to and conformity
with stereotypical native speakers thus often silencing other FL-specific
forms of language potentialrsquo (2002 60) Language learning in instructional
settings on my understanding of Kramschrsquos (2003) argument should not be
viewed solely in a teleological sensemdashas language acquisition directed
toward a fixed nativelike end-pointmdashbut rather as language appropriation
by learners in the sense that they are constructing a linguistic and even
cultural identity lsquoin the interstices of national languages and on the
margins of monolingual speakersrsquo territoriesrsquo (2003 260) The political nature
of the debate as highlighted by Kramsch becomes all the more apparent
when one additionally considers the status of the foreign language on
the world stage (eg French as an international language in retreat)
and how deeply the normative attitudes of its speakers run (eg Francersquos
three-century-long history of acceptance and maintenance of the standard
(Lodge 1993))
The foregoing synopsis of the debates surrounding native speaker usage as
both the goal of foreign language instruction and the extent to which
pedagogical norms need apply has highlighted the real world complexities
involved in what would otherwise seem to be a simple equation Revise
pedagogy to reflect actual use as shown by frequency studies and this
will result in a more targetlike performance on the part of learners So we
return then to the questions posed earlier (1) Should instruction be
changed (2) If so who along the continuum from the theorist-researcher to
the teacher-practitioner should inform and guide this change (3) At which
level (4) In what ways
With respect to our first question because one can empirically
demonstrate that the pedagogical norm for French adjective position
prevents learners from eventually acquiring native speaker (ie descriptively
accurate) norms instruction should be changed so thatmdashirrespective of
whether the attainment of such a norm is necessarily desirablemdashlearners
have both a choice in selecting one grammatical or lexical form over the other
and an awareness of the meaning potential of each choice (following Kramsch
2002 71)
BRUCE ANDERSON 303
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
Japanese learners of English were asked to judge the acceptability of multiple
wh-questions of the type Who ate what and Who sat where wherein
the sentence-final wh-word is a complement of the verb versus questions of
the type Who sang where wherein the sentence-final wh-word is an adjunct
Irrespective of this complementndashadjunct distinction the frequency of use of
the three question types was found to be quite low Only 122 examplars
beginning with who could be found in the 56 million word corpus
Bley-Vroman consulted 106 (or 87 percent) were specifically of the
who what type Native English speakers whose judgments were hypothe-
sized to follow from abstract rules favored both who wh[complement]
structures and disfavored the who wh[adjunct] structure despite the across-
the-board low frequency of all such structures in the corpus consulted The
learner group as expected favored the (relatively) predominant who what
type to the near exclusion of the other two seemingly paying no heed to the
complementndashadjunct distinction
Though certainly useful from the standpoint of L2 acquisition studies both
in terms of methodology and epistemological arguments frequency patterns
culled from authentic native-speaker corpora do not as Bley-Vroman (2004
270) pointed out necessarily reflect what learners are exposed to and it is
this latter type of corpus that is crucial in L2 acquisition studies because it
represents the input with which a learning algorithm would interact
The potential dichotomy between frequency patterns in native-speaker
usage and what classroom-based learners are exposed to (viewed broadly as
the sum total of textbook presentations authentic texts used in the
classroom and lsquoteacher talkrsquo) represents a third and especially applied
linguistic perspective on frequency An early study by Holmes (1988) on the
presentation of expressions of doubt and uncertainty in ESL textbooks for
example found that textbooks largely favored modal verbs while
simultaneously neglecting alternatives such as adjectives tag questions and
fallndashrise intonation (1988 40) This led her to raise concerns that a simplicity
criterion in the formulation of textbook rules may come at the cost of
naturalness and thus she concurred with Kasper (1979 277) that the
reduced range of expressions of modality witnessed in learnersrsquo production
can in some cases be viewed as teaching induced
More recent cases of mismatch have come to light through the corpus-
based methodology of Biber (1988 1995) and Biber et al (1998) and the
corpus-based Longman Grammar of Spoken and Written English (Biber et al
1999) A recent study by Biber and Reppen (2002) for instance examined
six ESL textbooks all of which presented attributive adjectives as the
principle means of modifying nouns Two-thirds of these textbooks included
some treatment of participial adjectives (those ending in -ed and -ing)
but only one discussed the use of nouns rather than adjectives as
prenominal modifiers (eg tomato sauce pencil case) However Biber et alrsquos
(1999) analysis of Englishmdashbased on a corpus of conversational transcripts
fictional literature newspaper writing and academic prosemdashshowed
288 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that (a) participial adjectives are quite rare across the four registers
(b) common attributive adjectives are in fact used the most in conversation
to modify nouns but most nouns actually have no modification at all in that
register and (c) nouns as prenominal modifiers are quite frequent in
newspaper writing and nearly as common as adjectives
Such corpus-based findings Biber and Reppen (2002) suggested should
result in participial forms being given little attention in intermediate to
advanced ESL materials and nouns as prenominal modifiers (and the various
and complex meaning relations they evince) being given more prominent
coverage In other words a descriptive norm for noun modification established
through corpus-based research should serve asmdashor at the very least
significantly inform the elaboration ofmdasha pedagogical norm (See OrsquoConnor Di
Vito 1991 for a similar argument based on frequency data from French) This
seemingly simple equation is not without its critics however (eg Owen 1993
Widdowson 2000) and we re-examine their criticisms in the discussion section
in light of the findings presented here on French adjective position
RESEARCH QUESTIONS
Although the potential for mismatch between native-speaker usage and
pedagogical materials has received some attention in the field of applied
linguistics the present study is unique in also offering a glimpse of the
potential for mismatch between pedagogical materials and ambient language
use in the classroom and as a result the potential for mismatch between
native speaker and L2 learner intuitions of acceptable word order(s) The
following three research questions are therefore addressed
1 To what extent are textbook presentations of French adjective position
congruent with authentic text sources across a number of genres
2 What is the nature of ambient language use involving adjective position
in the L2 French classroom and to what extent is it congruent with
textbook presentations of adjective position
3 To what extent are intuitions of acceptability among classroom learners
biased by any such (in)congruities
Because each of the three research questions represents a mini-study in
and of itself involving a different study methodology and type of data
analysis I address each question in its own section providing first an
overview of the methodology used and then a presentation of results
FRENCH ADJECTIVE POSITION IN TEXTBOOKS ANDAUTHENTIC SOURCES
The complexity of English adjectival modification pointed out in Biber and
Reppen (2002) is compounded in French by the fact that adjectives appear
BRUCE ANDERSON 289
in both adnominal positions prenominally (pre-N) and postnominally
(post-N) following certain lexical and contextual (pragmatic) constraints as
to their interpretation in one position versus the other (see eg Delmonier
1980 for a general review Delbecque 1990 for an account within cognitive
grammar Anderson 2002 for a compositional account within generative
syntax) This state of affairs leads us to question whether the traditional
textbook rules relating to adjectival modification in French reflect
attested native-speaker language use any better than that uncovered by
Biber and Reppen in authentic English-language texts across a number of
genres
English-speaking learners of French at least in the typical American
university setting are provided with three adjective placement rules
which take the form of a general rule with two exceptions Most adjectives
in French are said to be post-N (as in (1)) though a few are pre-N (as in (2))
and a few more are said to change meaning based on their position
(as in (3))
(1) une voiture rougelsquoa red carrsquo
(2) une petite maisonlsquoa small housersquo
(3) a un bijou cher un cher bijoulsquoan expensive jewelrsquo lsquoa cherished jewelrsquob une eglise ancienne une ancienne eglise
lsquoan ancient churchrsquo lsquoa former churchrsquo
First- and second-year textbooks of French published in the USA are
strikingly similar in the statement of such rules and in their order of
presentation though they differ to some degree as to which adjectives are to
be included in the (putative) lexical classes constituting exceptions to the
general post-N rule The textbooks used in the classrooms observed for the
present study are no different In the first-year textbook Chez nous (Valdman
and Pons 1997) students are first told that most adjectives follow the noun
(1997 73) but that the twelve adjectives listed in (4) typically and
exceptionally precede the noun (1997 241)
(4) a jeune vieux lsquoyoung oldrsquob nouveau lsquonewrsquoc petit grand gros lsquosmalllittle bigtall largersquod beau lsquohandsomersquoe joli lsquoprettyrsquof bon mauvais lsquogood badrsquog premier dernier lsquofirst lastrsquo
In the second-year textbook Quant a moi (Bragger and Rice 1996) the basic
post-N rule is reviewed as is the class of adjectives that exceptionally appear
prenominally (1996 33) Note however that this class of exceptions has
eleven members instead of twelve having lost some members from the Chez
290 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
Nous list in (4) (eg bon premier dernier gros) but gaining others (eg long
lsquolongrsquo court lsquoshortrsquo and autre lsquootherrsquo)
Bragger and Rice (1996) then present the second exception to the basic
post-N rule wherein students are told that lsquoa number of adjectives change
meaning depending on whether theyrsquore placed after or before a noun
In general an adjective placed after the noun retains its basic concrete
meaning The same adjective placed before the noun is used in an abstract or
figurative mannerrsquo (1996 34 emphasis in original) The seven adjectives that
apparently fall into this class are listed in (5) The first English gloss captures
the meaning derived from pre-N position the second gloss that derived from
post-N position Note that Bragger and Rice put the adjective grand in both
lists of exceptions In other words grand is apparently exceptionally pre-N
and exceptionally variable at the same time2
(5) ADJECTIVE PRE-N POST-Na ancien lsquoformerrsquo vs lsquoold ancientrsquob cher lsquodear well-lovedrsquo vs lsquoexpensiversquoc dernier lsquolast latest (in a series)rsquo vs lsquolast (before this one)rsquod grand lsquogreatrsquo vs lsquotall largersquoe pauvre lsquopoor (unfortunate)rsquo vs lsquopoor (not rich)rsquof prochain lsquonext (in a series)rsquo vs lsquonext (after this one)rsquog propre lsquoownrsquo vs lsquocleanrsquo
To summarize the pedagogical treatment of French adjective position
commonly found in university-level American textbooks presents post-N
position as the most frequent and typical (frac14unmarked) adjective order
the less frequent and exceptional (frac14marked) pre-N order is either required
when the adjective is a member of a seemingly fixed lexical class such as
the list in (4) or licensed by a particular interpretation that differs from that
obtained in post-N position such as the list in (5) This difference in
interpretation is often though not always captured through distinct English
glosses
Methodology
In order to determine the extent to which the pedagogical treatment just
summarized is congruent with authentic French texts three corpora were
consulted Wilmetrsquos (1980) corpus of literary texts (approximately 10 million
words 3835 adjectives) Forsgrenrsquos (1978) corpus of newspaper texts from Le
Monde and LrsquoExpress (1 million words 829 adjectives) and Larssonrsquos (1993)
corpus of travel catalogues travel guides and other non-literary prose (15
million words 113 adjectives)3 Using Wilmetrsquos corpus as the point of
comparison given that it contains the highest number of discrete lexemes
(3835 adjectives) an arbitrary cut-off point of 25 attestations or more was
used creating a list of 205 adjectives4 The total number of attestations pre-N
attestations and post-N attestations for each of the 205 adjectives across
BRUCE ANDERSON 291
the three corpora was then determined by adding Forsgrenrsquos and Larssonrsquos
data to that found in Wilmet if available
Results
The 205 most frequently attested adjectives in French texts appear to fall
along a continuum at one end of which we find 68 adjectives (or 332
percent) appearing either only in post-N position (nfrac1434) or typically in that
position (nfrac14 34) and at the other end of which we find 15 adjectives (or 73
percent) typically appearing in pre-N position as shown in Table 1 Note that
there are no adjectives appearing only in pre-N position The remaining 122
adjectives (or 595 percent) are regularly attested in both positions with
some having what we might call an affinity for post-N position (nfrac1441) or
pre-N position (nfrac14 29) but others being highly variable (nfrac1452)
Reminiscent of Holmesrsquo (1988) critique of the simplicity versus naturalness
of textbook rules such data raise some problems for the seemingly
straightforward fixed-order rules of adjective position exemplified in (1)ndash(3)
and the associated lists in (4) and (5) A textbook rule to the effect that
adjectives typically appear after the noun in Frenchmdasha word order that
clearly contrasts with what one finds in Englishmdashis only partly descriptively
Table 1 Frequency groupings of adjective position in written French texts
Frequency grouping Type count(nfrac14 205)
Percent oftotal
Examples
Exclusively pre-N (0post-N position)
0 0 mdash
Typically pre-N (below10 post-N position)
15 75 bon lsquogoodrsquo fameuxlsquo(in)famousrsquo petit lsquosmallrsquo
Variable but with anaffinity for pre-Nposition (10ndash39post-N position)
29 14 charmant lsquocharmingrsquo etrangelsquostrangersquo miserable lsquomiserablersquo
Highly variable (40ndash59post-N position)
52 255 delicieux lsquodeliciousrsquo joyeuxlsquojoyousrsquo puissant lsquopowerfulrsquo
Variable but with anaffinity for post-Nposition (60ndash90post-N position)
41 20 amer lsquobitterrsquo tiede lsquolukewarmrsquoserieux lsquoseriousrsquo
Typically post-N (above90 post-N)
34 17 blanc lsquowhitersquo inconnulsquounknownrsquo sauvage lsquowildrsquo
Exclusively post-N(100 post-Nposition)
34 17 allemand lsquoGermanrsquo dorelsquogoldenrsquo quotidien lsquodailyrsquo
292 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
accurate It is accurate in the sense that the ends of the continuum are
not equivalent in their percentages There are many more adjectives
appearing either only or typically in post-N position than is the case for pre-N
position It is inaccurate in the sense that it ignores the distributional
behavior of 595 percent of the most frequently used adjectives which can
appear in both positions French is not as the rule implies the mirror
opposite of English but instead is more accurately viewed as an lsquoANArsquo
(adjectivendashnounndashadjective) language as opposed to both lsquosystematic ANrsquo
languages like English and German and lsquosystematic NArsquo languages like
Indonesian and Thai (Cinque 1994 99ndash101)
The data also lead us to question the criteria textbooks use for including
certain adjectives in the two lists in (4) and (5) For the so-called prenominal
adjectives a frequency of 90 percent or greater in that position would
establish that adjectives such as excellent fameux lsquo(in)famousrsquo moindre
lsquoslightestrsquo and vrai lsquorealtruersquo need to be added That list itself implies a fixed
lexical class to be memorized though in reality it does nothing more
than provide a handful of frequently attested adjectives (nfrac14 15 [75 percent])
that typically appear just in that position Many more adjectives can
appear in pre-N position as well For the so-called change-of-meaning
adjectives the data show that the list in (5) falls woefully short in terms of
representing those adjectives that actually do regularly appear in both
positions (nfrac14 52)
Even if it were to be established that such rules are accurate with respect
to colloquial French as spoken between native speakers and thus appropriate
for oral proficiency-oriented classrooms such rules obviously fail to capture
the state of affairs in written texts particularly literature to which learners at
later course levels are exposed Yet even students at earlier course levels
occasionally find counter-examples to what they are taught The four cases of
lsquounexpectedrsquo pre-N position in (6) come from the cultural readings of two
first-year and two second-year US textbooks respectively
(6) a La France a un excellent systeme de transports publics(Magnan et al 2002 223)lsquoFrance has an excellent system of public transportrsquo
b Cela explique le grand nombre de magnifiques chateauxdans la region (Valdman and Pons 1997 395)lsquoThis explains the great number of magnificent chateaux inthe regionrsquo
c [U]ne veritable explosion de la technologie (St Onge etal 2003 71) lsquoa veritable explosion of technologyrsquo
d Lrsquohistoire est une douloureuse suite de tragedieshumaines (Oates and Dubois 2003 208)lsquoIts history is a painful series of human tragediesrsquo
One might wish to argue that the examples provided in (6) demonstrate
that inputmdashat least classroom-related inputmdashis not as contradictory to
descriptive norms as argued in this section Examples of the type in (6)
BRUCE ANDERSON 293
however are only occasionally found in textbook reading materials they are
included here to demonstrate that even when making a concerted effort to
write level-appropriate cultural readings textbook authors sometimes break
the very grammar rules they present elsewhere in the textbook suggesting a
true disparity between rules and lsquonaturalisticrsquo usage5 (I return to the issue of
descriptively accurate statements about French adjective position in the
discussion section)
FREQUENCY OF ADJECTIVE USE AND POSITIONIN CLASSROOM INPUT
Methodology
In order to get a general sense of how well the classroom setting as primary
input source reflects either textbook rules or the corpora-based descriptive
norm just summarized frequency counts of adjective type (ie discrete
lexemes) and their position in classroom input were collected by the
researcher from 60 hours of in-class observations of university-level French
courses at a large Midwestern American university Ten hours were spent in
each of six courses spanning the curriculummdashfrom the four semesters
comprising the first- and second-year language skills curriculum (referred to
as Levels 1a 1b 2a and 2b below) to a third-year conversation course
(Level 3) and a fourth-year lecture course on French civilization (Level 4)
Each of these courses was taught almost exclusively in French by a mix of
native and nonnative instructors none of whom were aware of the purpose
of this observation period Class observations were conducted on days when
adjective position was not the focus of explicit instruction as this would have
more than likely biased the frequency counts
The first- and second-year courses consisted of explicit grammar
instruction drills and communicative activities normally involving students
in pair or group work The conversation and civilization courses involved
no explicit grammar instruction the former involved mostly teacherndash
student and studentndashstudent oral communication based on previous course
readings and video segments whereas the latter was almost exclusively
devoted to instructor-delivered lectures Error correction was minimal in all
courses and no case of error correction involving adjective position was
witnessed during the observation period (despite the fact that some errors
were made)
Oral input on adjective position in all courses came from one of three
sources the instructor audio aides (such as cassette and videotape) and
the students themselves Written input in the first- and second-year
courses came from two sources visual aids (such as handouts transparencies
writing on the blackboard) and passages from the textbook that were read
silently or aloud in class At Level 3 written materials included authentic
newspaper and magazine articles in addition to visual aides at Level 4
294 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
a number of works of French literature were read outside of class Note
that written input in these latter two course levels is qualitatively the same
as that used in the corpora compiled by Forsgren (1978) Wilmet (1980) and
Larsson (1993)
At issue during this period of investigation was classroom input that could
be seen andor heard by all students in the classroom (as well as the
researcher) Input from students involved in pair and group work was
therefore excluded from analysis Material to be read outside of class was
also excluded as there was no guarantee that all students had actually
read such material Although the totality of inputmdashincluding sources of input
the learner might have voluntarily sought out outside of classmdashcould
obviously never be exhaustively tallied the exclusion of certain forms of
input was first and foremost a matter of logistics (ie not having the
appropriate funds equipment facilities and time to sound record then
transcribe all classroom input from every participant in every situation of
use) Despite this drawback I see no reason to suspect that student-delivered
input to other students during pair- or group-work would be qualitatively
different (at least in terms of adjective position) from the input delivered by
students that was audible to the entire class this latter source of input
already constitutes 10ndash30 percent of all sources of input tallied as part of the
observation period
Each instance of adjective use during each class session was recorded with
pen and paper by the researcher A coding scheme was devised whereby an
11 17-inch sheet of paper was divided into quadrants labeled according to
the source of the input lsquoinstructorrsquo lsquostudentrsquo lsquoaudiorsquo and lsquowrittenrsquo Each
quadrant was then subdivided according to the position of adjectival
modification lsquopre-Nrsquo lsquopost-Nrsquo lsquopredicatersquo and lsquootherrsquo (to be described
below) Input of the form une bonne idee lsquoa good idearsquo by the instructor was
therefore recorded as an instance of the adjective bonne in the instructor
quadrant pre-N sub-quadrant Hatch marks next to the adjective bonne were
used for any further exemplars of this adjective (including its masculine form
bon) in this position from this source I acknowledge that mere human error
caused by fatigue etc may have resulted in tallies that are less accurate than
if each session were recorded on tape Though the results presented below
should be considered with some caution then occasional errors could not
have reached a significant enough level to detract from the highly salient
patterns of usage that we will find
The resulting sixty sheets each representing one 50-minute class session
were then tallied in order to determine (a) the relative contribution of the
four sources of input (b) the average token (exemplar) count of adjectives
per 50-minute class period (c) the ratio of pre-N to post-Npredicate tokens
(d) the type count (number of discrete lexemes) entering into pre-N and
post-N positions and (e) the frequency of variation in positionmdashthat is the
extent to which a given adjective appeared at least once in both pre-N and
post-N position during the ten hours of observation
BRUCE ANDERSON 295
Results
The instructor not surprisingly was the primary provider of input in the
classroom (See Figure 1 available to online subscribers at http
applijoxfordjournalsorg) This was especially the case in the first-year
courses (Levels 1a 1b) the second- and third-year courses (Levels 2a 2b 3)
showed a greater proportion of input from students and from audio aides
The instructor was by far the predominant source of input in the fourth-year
course (Level 4) given that it involved instructor-delivered lectures on
French civilization with minimal class participation on the part of students
Concerning the average number of times an adjective occurred in the
input during any one 50-minute class period we find a general increase with
level from 42 tokens at Level 1a to 156 tokens per class at Level 4 (See
Figure 2 available to online subscribers at httpapplijoxfordjournalsorg)
Students therefore heard an adjective used anywhere from one to three
times per minute on average
Aggregate data on the frequency with which adjectives appeared in pre-N
post-N and predicate positions or in other constructions (eg used in
isolation in elliptical constructions such as la charmante lsquothe charming [one]rsquo
and following indefinite pronouns as in quelqursquoun de charmant lsquosomeone
charmingrsquo) demonstrate that post-N and predicate positions were generally
equally frequent and together constitute fully 75 percent of the ratio whereas
pre-N position never surpassed 27 percent across course levels Other
constructions were minimal (See Figure 3 available to online subscribers at
httpapplijoxfordjournalsorg)
This roughly 75 percentndash25 percent ratio was then reanalyzed in order to
consider the number of distinct lexemes (lsquotypesrsquo) that make up the
percentages We find for example that only 15 lexemes make up the 14
percent of pre-N adjective use at Level 1a This number gradually increased
with course level but only reached a total of 41 by Level 4 Post-N position
on the other hand admitted a much larger variety of adjectives in classroom
input from 71 at Level 1a to 323 by Level 4 (See Figure 4 available to online
subscribers at httpapplijoxfordjournalsorg) Students therefore got
relatively little input on pre-N position and what input they did get was
made up of a relatively small class of adjectives being used over and over
again and which already figure in the textbook-determined list of pre-N
adjectives in (4) (See Appendix A available to online subscribers at http
applijoxfordjournalsorg for a list of adjectives appearing in pre-N position
by course level)
Finally we find that very few adjectives varied in position during the 10
hours of in-class observation There was no variation attested at Level 1a
and only four to seven adjectives varied in position from Level 1b through to
Level 3 Moreover some of the input provided to learners in these courses
in the form of other studentsrsquo output was nonnativelike6 At the most
advanced undergraduate level learners got somewhat more input involving
296 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
variation in position where 20 adjectives did so (See Figure 5 available to
online subscribers at httpapplijoxfordjournalsorg) Again most (but not
all) of these adjectives already figure into the textbook-determined list of
variable adjectives in (5) (See Appendix B available to online subscribers at
httpapplijoxfordjournalsorg for a list of adjectives varying in position by
course level)
To summarize the observational data just presented the ambient language
of the classroom clearly mimics the pedagogical treatment that American
university-level classroom learners receive from textbooks Although
adjectives are in general a very frequent part of the input the rate of
pre-N position (at roughly 25 percent) is essentially static across levels
pointing to the fact that classroom input across levels is in no way
contradictory to what learners are initially taught Post-N position is basic
(unmarked) and pre-N position is exceptional (marked) The adjectives that
do appear in pre-N position in classroom input are in most cases those that
learners already expect to appear in that position (cf the list in (4) and
Appendix A) The few adjectives that appear in both positions are again in
most cases those that learners expect to do so (cf the list in (5) and
Appendix B)
By the end of the first-year level then both instruction and classroom
input would lead learners to expect that adjectives will appear in post-N
position with the notable exception of the class of adjectives in (4) (those in
(5) having not yet been introduced) By the second-year level it could be
argued that instruction provides examples in the form of so-called change of
meaning adjectives that variation in position is to be associated with
variation in interpretation (and perhaps by implication always to
interpretation) However learnersrsquo explicit knowledge of interpretative
differences as exemplified by the adjectives in (5) typically consists of
arbitrary and unsystematic English glosses By arbitrary I mean that
instruction never provides an adequate explanation for why the pre-N
position of cher should correspond to beloved or for its post-N position to
correspond to expensive7 Though telling students following Bragger and Rice
(1996) that the meaning in post-N position is lsquoconcretersquo and the meaning in
pre-N position is lsquofigurativersquo might (somehow) help it does not account for
other adjectives in (5) like prochain lsquonextrsquo This latter adjective also
demonstrates the fact that English glosses are unsystematic in that prochain
translates as lsquonextrsquo in either position
Before considering our final research question I note that no research on
adjective position in native French speaker discourse is available to date
An anonymous Applied Linguistics reviewer questions the validity of compar-
ing primarily spoken classroom discoursemdashor any other kind of spoken
discoursemdashto written texts without first demonstrating that adjective position
is insensitive to this difference in medium While recognizing the potentially
important distinction in medium and the need for further research in this
area the validity of comparing classroom input to written texts in the present
BRUCE ANDERSON 297
study lies in the fact that such a comparison roughly corresponds to the
curriculum of typical university-level courses an emphasis on listening
comprehension and oral communication in the first- and second-year
language courses leading to an emphasis on reading comprehension
(typically literature-based) and proficiency in writing in the third- and
fourth-year courses Were it to turn out that the degree of variation in
adjective position found in written texts is restricted to just that medium
revisions to pedagogy would necessarily be restricted to writing-intensive
(composition creative composition stylistics) courses at the third- and
fourth-year levels (I return to this issue in the discussion section)
LEARNER INTUITIONS OF ACCEPTABILITY
Thus far we have seen an incongruity between textbook presentations of
adjective position and authentic text sources across a number of written text
genres (research question 1) and that ambient language use involving
adjective position in the classroom is clearly more congruent with and might
therefore be taken to reinforce the explicit rules found in textbooks
(research question 2) We are now in a position to address our third research
questionmdashthe extent to which intuitions of acceptability among classroom
learners are biased by the congruity between pedagogical treatment and
classroom input frequency patterns An empirical study previously reported
in Anderson (2002 2007) can shed some light on this question
Methodology
Anderson (2002 2007) investigated the acquisition of various morphosyn-
tactic properties of French noun phrases by administering an acceptability
judgment task in which participants had to evaluate the appropriateness of
sentences featuring among other things the pre- versus post-N position of
adjectives Each of the adjectives selected were lsquohighly variablersquo with respect
to position (cf Table 1) yet crucially do not figure in textbook lists of
adjectives said to change meaning based on position (cf the list in (10))
These include lourd lsquoheavyrsquo violent lsquoviolentrsquo charmant lsquocharmingrsquo precieux
lsquopreciousrsquo epais lsquothickrsquo etrange lsquostrangersquo and delicieux lsquodeliciousrsquo (See
Appendix C available to online subscribers at httpapplijoxfordjournals
org for test sentences) Study participants included English-speaking learners
of French in the second third and fourth years of undergraduate study
(nfrac14 80) who were attending courses at the same institution at which the
observational data were collected a group of advanced learners (nfrac14 20)
consisting of graduate student instructors at that same institution as well as
secondary education teachers and a group of native French speakers (nfrac14 27)
roughly matched in age and education level
Each adjective appeared in test sentences once prenominally and once
postnominally with each test sentence preceded once by a context
298 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
motivating acceptance and once by a context motivating rejection of one or
the other adjective order for a total of four positionndashinterpretation pairings
per adjective These were randomly distributed across two testing
instruments administered a week apart After reading each context and
paired test sentence study participants were directed to check boxes
labeled lsquofinersquo lsquooddrsquo or lsquocannot decidersquo to the prompt This particular sentence
sounds ___ in this context
As an example each of the sentences in (7) featuring the adjective lourd
lsquoheavyrsquo was paired with each of two contexts
(7) a Pierre doit laisser la lourde valise avec son amib Pierre doit laisser la valise lourde a lrsquoaeroport
lsquoPierre has to leave the heavy suitcase with his friendatthe airportrsquo
In the first context motivating a unique noun-referent interpretation
there was one and only one suitcase which happened to be heavy and
which lsquoPierrersquo had to leave behind with his friend In the second context
motivating a multiple noun-referent interpretation there were a number of
suitcases only one of which was heavy and it was this one that lsquoPierrersquo had
to leave behind at the airport (See Appendix D available to online
subscribers at httpapplijoxfordjournalsorg for the complete version of
each context)
When viewed solely in terms of truth-conditional semantics each of the
four positionndashinterpretation pairings is compatible save for one pre-N
position (7a) in a multiple noun-referent context This is because the singular
definite article la and pre-N adjective lourde together presuppose a unique
suitcase in context (which happens to be heavy) This presupposition
does not hold when the adjective is in post-N position where all that is
required is that there be at least one such suitcase in context The testing
instrument then allows us to examine how classroom learners react to
an lsquoexpectedrsquo word order (ie sentences like (7b) in either context) as well
as to an lsquounexpectedrsquo but semantically and pragmatically appropriate
word order (ie sentences like (7a) but only in the unique noun-referent
context)
Results
As the mean acceptance rates from native French speakers in Table 2 show
truth-conditional semantics was not the only factor determining acceptance
rates Two of the pairings received essentially chance-level rates of
acceptance (pre-N position in a multiple noun-referent context at 49
percent and post-N position in a unique context at 52 percent) whereas the
other two pairings showed much clearer above- or below-chance rates (pre-N
position in a unique noun-referent context at 74 percent and post-N
position in a multiple noun-referent context at 37 percent)
BRUCE ANDERSON 299
As was argued in Anderson (2002 2007) results for each of the four
positionndashinterpretation pairings among the native French speakers are
explainable in terms of a disproportionate interaction between semantics and
pragmatic implicature in particular the Gricean maxim of quantity An
utterance should be no more and no less informative than is required for the
current purposes of the exchange (Grice 1975) The positionndashinterpretation
pairing that was lsquotruersquo in terms of semantics and also the most pragmatically
appropriate was accepted at the highest rate (74 percent) the pairing that
was also lsquotruersquo in terms of semantics but was pragmatically the least
appropriate was accepted at the lowest rate (37 percent)8
Do learner group response patterns show the same bias toward pragmatic
implicature in their acceptance rates The comparison of mean acceptance
rates between native French speakers and learner groups in Table 3 shows
that learners performed in a particular (and particularly non-Frenchlike) way
in which the interaction goes in the opposite direction favoring semantics
over pragmatics
Recall that the only positionndashinterpretation pairing that was semantically
incompatible was pre-N position (7a) in a multiple noun-referent context
given the presupposition of uniqueness of the noun referent All learner
groups did indeed accept this pairing at the lowest rate of the four and with
the exception of the second-year learners (at thorn9 percent) did so more
strongly than the native French speakers at mean acceptance rates that were
11ndash15 percent lower)9 However they also accepted at much lower mean
rates (12 to 23 percent) the most felicitous positionndashinterpretation pairing
for the native French speakers pre-N position in a unique noun-referent
context Recall also that post-N position (7b) was semantically compatible
under either interpretation All learner groups accepted these two positionndash
interpretation pairings at much higher rates than the native French speakers
(thorn23 to thorn31 percent) including the least felicitous positionndashinterpretation
pairing for the native French speakersmdashpost-N position in a multiple noun-
referent context This is precisely what one would expect from the
pedagogical treatment these learners are provided and the congruity of
that treatment with classroom input which under-represents actual usage in
Table 2 Native French speaker mean acceptance rates for the uniquenonunique noun-referent distinction
Prenominal (AN)order ()
Postnominal (NA)order ()
Unique noun referent 74 52
Nonunique noun referent 49 37
Source Adapted from Anderson 2002 2007
300 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
text-based corpora and as we have now seen differs from native speaker
intuitions of overall (semantico-pragmatic) acceptability
Although a bias toward or even strict adherence to explicit pedagogical
rules in evaluating such sentences may not seem all that surprising what does
stand out in the data from Anderson (2002 2007) is that these differential
rates in acceptance between native and nonnative speakers are as true for the
advanced group in that studymdashconsisting of graduate student instructors and
second education teachersmdashas they are for the intermediate (second-year)
learner group This would seem to indicate no increased sensitivity to
contextual (pragmatic) factors in adjective position over the course of studying
French even though the advanced group had been exposed to much more
authentic input through texts written for and by native French speakers (ie
the descriptive norm uncovered in our examination of authentic text corpora)
DISCUSSION
What should be clear from the findings presented here is that rules of French
adjective placement that one typically finds in (at least American) textbooks
though not plainly incorrect are not fully descriptively accurate either at
least with respect to written French across several text genres In particular
such rules portray pre-N and post-N positions as mutually exclusive for
almost all adjectives when in fact they are not
Similar discrepancies are often found between rules for and by native
speakers and what those speakers actually do One need look no further than
the French for an example with colloquial speech often at odds with the
prescriptive bon usage and the admonishments of the Academie francaise One
might therefore also expect to find a discrepancy between textbook rules and
what teachers do in the classroom andor what students find to be
lsquoacceptablersquo in terms of word order As we have seen however this is not
the case These rules are faithfully reflected in classroom language use by
instructors to such an extent that when asked to evaluate the acceptability
of lsquoexpectedrsquo and lsquounexpected but nativelikersquo word orders (and one would
presume in production as well) both instructors and students paid much
less attention to problems in pragmatic implicature than did native speakers
a situation which does not appear to change even in the face of counter-
examples occasionally provided in the textbooks they use (see (6) above) and
most certainly in the written texts that they read at later course levels As
pointed out by an anonymous reviewer this situation can be considered a
testament to the power of classroom instruction (instructor input in
particular) if instruction is changed he or she reasons it is likely that
classroom learners will reflect this change
But should instruction be changed If so who along the continuum from
the theorist-researcher to the teacher-practitioner should inform and guide
this change At which level In what way Such questions are at the very
heart of two on-going debates that I re-examine before proposing some
BRUCE ANDERSON 301
answers the pedagogical versus descriptive norm debate (eg Owen 1993
Widdowson 2000 and the collection of articles in Gass et al 2002) and
the interrelated native-speaker-usage-as-goal debate in foreign language
pedagogy (eg the collection of articles in Blyth 2003)
On the whole the approach to instruction on French adjective
position outlined earlier appears advantageous in its simplicity providing
easily remembered rules of thumb that learners can apply in language
production That this approach to instruction does not as we have seen
straightforwardly follow the descriptive norm is a situation that linguists
utilizing a corpus-based methodology believe can and should be remedied
Conrad (2000) for instance sees in corpus-based research an immediate way
in which linguistics can inform language pedagogy freeing language teachers
from lsquohav[ing] to rely on judgments of grammatical accuracy or on native
speakersrsquo intuition about what sounds naturalrsquo (2000 555) Although the
subsequent study by Biber and Reppen (2002) uses a disclaimer to the effect
that frequency should not be the sole factor in material design they
nevertheless suggest that lsquoa selective revision of pedagogy to reflect actual
use as shown by frequency studies could result in radical changes that
facilitate the learning process for studentsrsquo (2002 199) Owen (1993)
however provides examples of how a total reliance on a corpus does not
necessarily yield better observation and more importantly questions
whether better observation automatically equates to better explanation
(1993 168) (but see Francis and Sinclair 1994 for rebuttal) For his part
Widdowson (2000) objects to the kind of equation just described as a
case of linguistics applied a mistaken belief that what is textually attested
lsquouniquely represents real language and that this reality should define the
foreign language subjectrsquo (2000 8) (but see Stubbs 2001 for rebuttal) What
is needed according to Widdowson is not simply linguistics applied
to pedagogy but a true role for applied linguistics an approach to usage that
also recognizes native speakersrsquo awareness of and opinions about such usage
The need for a truly applied linguistics perspective on native speaker
usage (including frequency patterns) is embodied in a pedagogical norm
originally conceptualized in Valdman (1989) and elaborated upon in
Bardovi-Harlig and Gass (2002) as
[A] combination of language systems and forms selected by linguistsand pedagogues to serve as the immediate language target ortargets that learners seek to acquire during their language studyIn other words pedagogical norms represent a mid-point or seriesof mid-points for learners as they progress toward acquiringnative language norms [thus constituting] a form of languagethat is acceptable to native speakers but easier to learn than the fullnative language system (p 3)
The textbook word-order rules for French adjective position reviewed
earlier would first appear to constitute an appropriate pedagogical norm
302 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
under Bardovi-Harlig and Gassrsquos (2002) definition because the learner is able
to produce a subset of the totality of possible utterances deemed acceptable to
native speakers which may turn out to be even more the case in spoken than
in written French (though the difference in mediums remains to be studied)
The learner data presented here from Anderson (2002 2007) however
provides one piece of evidence that textbook rules and classroom input do not
constitute a sufficient pedagogical norm as learners have stopped in their
progress toward acquiring native speaker (ie descriptively accurate) norms
As the articles in Blyth (2003) attest however it is highly debatable
whether the lsquoNative Speakerrsquo (as an idealized abstraction) should be the
appropriate model for the contemporary learner For Kramsch (2002 2003)
too much of a focus on an authentic native speaker norm reduces
foreign language (FL) study to a constant lsquoapproximation to and conformity
with stereotypical native speakers thus often silencing other FL-specific
forms of language potentialrsquo (2002 60) Language learning in instructional
settings on my understanding of Kramschrsquos (2003) argument should not be
viewed solely in a teleological sensemdashas language acquisition directed
toward a fixed nativelike end-pointmdashbut rather as language appropriation
by learners in the sense that they are constructing a linguistic and even
cultural identity lsquoin the interstices of national languages and on the
margins of monolingual speakersrsquo territoriesrsquo (2003 260) The political nature
of the debate as highlighted by Kramsch becomes all the more apparent
when one additionally considers the status of the foreign language on
the world stage (eg French as an international language in retreat)
and how deeply the normative attitudes of its speakers run (eg Francersquos
three-century-long history of acceptance and maintenance of the standard
(Lodge 1993))
The foregoing synopsis of the debates surrounding native speaker usage as
both the goal of foreign language instruction and the extent to which
pedagogical norms need apply has highlighted the real world complexities
involved in what would otherwise seem to be a simple equation Revise
pedagogy to reflect actual use as shown by frequency studies and this
will result in a more targetlike performance on the part of learners So we
return then to the questions posed earlier (1) Should instruction be
changed (2) If so who along the continuum from the theorist-researcher to
the teacher-practitioner should inform and guide this change (3) At which
level (4) In what ways
With respect to our first question because one can empirically
demonstrate that the pedagogical norm for French adjective position
prevents learners from eventually acquiring native speaker (ie descriptively
accurate) norms instruction should be changed so thatmdashirrespective of
whether the attainment of such a norm is necessarily desirablemdashlearners
have both a choice in selecting one grammatical or lexical form over the other
and an awareness of the meaning potential of each choice (following Kramsch
2002 71)
BRUCE ANDERSON 303
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that (a) participial adjectives are quite rare across the four registers
(b) common attributive adjectives are in fact used the most in conversation
to modify nouns but most nouns actually have no modification at all in that
register and (c) nouns as prenominal modifiers are quite frequent in
newspaper writing and nearly as common as adjectives
Such corpus-based findings Biber and Reppen (2002) suggested should
result in participial forms being given little attention in intermediate to
advanced ESL materials and nouns as prenominal modifiers (and the various
and complex meaning relations they evince) being given more prominent
coverage In other words a descriptive norm for noun modification established
through corpus-based research should serve asmdashor at the very least
significantly inform the elaboration ofmdasha pedagogical norm (See OrsquoConnor Di
Vito 1991 for a similar argument based on frequency data from French) This
seemingly simple equation is not without its critics however (eg Owen 1993
Widdowson 2000) and we re-examine their criticisms in the discussion section
in light of the findings presented here on French adjective position
RESEARCH QUESTIONS
Although the potential for mismatch between native-speaker usage and
pedagogical materials has received some attention in the field of applied
linguistics the present study is unique in also offering a glimpse of the
potential for mismatch between pedagogical materials and ambient language
use in the classroom and as a result the potential for mismatch between
native speaker and L2 learner intuitions of acceptable word order(s) The
following three research questions are therefore addressed
1 To what extent are textbook presentations of French adjective position
congruent with authentic text sources across a number of genres
2 What is the nature of ambient language use involving adjective position
in the L2 French classroom and to what extent is it congruent with
textbook presentations of adjective position
3 To what extent are intuitions of acceptability among classroom learners
biased by any such (in)congruities
Because each of the three research questions represents a mini-study in
and of itself involving a different study methodology and type of data
analysis I address each question in its own section providing first an
overview of the methodology used and then a presentation of results
FRENCH ADJECTIVE POSITION IN TEXTBOOKS ANDAUTHENTIC SOURCES
The complexity of English adjectival modification pointed out in Biber and
Reppen (2002) is compounded in French by the fact that adjectives appear
BRUCE ANDERSON 289
in both adnominal positions prenominally (pre-N) and postnominally
(post-N) following certain lexical and contextual (pragmatic) constraints as
to their interpretation in one position versus the other (see eg Delmonier
1980 for a general review Delbecque 1990 for an account within cognitive
grammar Anderson 2002 for a compositional account within generative
syntax) This state of affairs leads us to question whether the traditional
textbook rules relating to adjectival modification in French reflect
attested native-speaker language use any better than that uncovered by
Biber and Reppen in authentic English-language texts across a number of
genres
English-speaking learners of French at least in the typical American
university setting are provided with three adjective placement rules
which take the form of a general rule with two exceptions Most adjectives
in French are said to be post-N (as in (1)) though a few are pre-N (as in (2))
and a few more are said to change meaning based on their position
(as in (3))
(1) une voiture rougelsquoa red carrsquo
(2) une petite maisonlsquoa small housersquo
(3) a un bijou cher un cher bijoulsquoan expensive jewelrsquo lsquoa cherished jewelrsquob une eglise ancienne une ancienne eglise
lsquoan ancient churchrsquo lsquoa former churchrsquo
First- and second-year textbooks of French published in the USA are
strikingly similar in the statement of such rules and in their order of
presentation though they differ to some degree as to which adjectives are to
be included in the (putative) lexical classes constituting exceptions to the
general post-N rule The textbooks used in the classrooms observed for the
present study are no different In the first-year textbook Chez nous (Valdman
and Pons 1997) students are first told that most adjectives follow the noun
(1997 73) but that the twelve adjectives listed in (4) typically and
exceptionally precede the noun (1997 241)
(4) a jeune vieux lsquoyoung oldrsquob nouveau lsquonewrsquoc petit grand gros lsquosmalllittle bigtall largersquod beau lsquohandsomersquoe joli lsquoprettyrsquof bon mauvais lsquogood badrsquog premier dernier lsquofirst lastrsquo
In the second-year textbook Quant a moi (Bragger and Rice 1996) the basic
post-N rule is reviewed as is the class of adjectives that exceptionally appear
prenominally (1996 33) Note however that this class of exceptions has
eleven members instead of twelve having lost some members from the Chez
290 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
Nous list in (4) (eg bon premier dernier gros) but gaining others (eg long
lsquolongrsquo court lsquoshortrsquo and autre lsquootherrsquo)
Bragger and Rice (1996) then present the second exception to the basic
post-N rule wherein students are told that lsquoa number of adjectives change
meaning depending on whether theyrsquore placed after or before a noun
In general an adjective placed after the noun retains its basic concrete
meaning The same adjective placed before the noun is used in an abstract or
figurative mannerrsquo (1996 34 emphasis in original) The seven adjectives that
apparently fall into this class are listed in (5) The first English gloss captures
the meaning derived from pre-N position the second gloss that derived from
post-N position Note that Bragger and Rice put the adjective grand in both
lists of exceptions In other words grand is apparently exceptionally pre-N
and exceptionally variable at the same time2
(5) ADJECTIVE PRE-N POST-Na ancien lsquoformerrsquo vs lsquoold ancientrsquob cher lsquodear well-lovedrsquo vs lsquoexpensiversquoc dernier lsquolast latest (in a series)rsquo vs lsquolast (before this one)rsquod grand lsquogreatrsquo vs lsquotall largersquoe pauvre lsquopoor (unfortunate)rsquo vs lsquopoor (not rich)rsquof prochain lsquonext (in a series)rsquo vs lsquonext (after this one)rsquog propre lsquoownrsquo vs lsquocleanrsquo
To summarize the pedagogical treatment of French adjective position
commonly found in university-level American textbooks presents post-N
position as the most frequent and typical (frac14unmarked) adjective order
the less frequent and exceptional (frac14marked) pre-N order is either required
when the adjective is a member of a seemingly fixed lexical class such as
the list in (4) or licensed by a particular interpretation that differs from that
obtained in post-N position such as the list in (5) This difference in
interpretation is often though not always captured through distinct English
glosses
Methodology
In order to determine the extent to which the pedagogical treatment just
summarized is congruent with authentic French texts three corpora were
consulted Wilmetrsquos (1980) corpus of literary texts (approximately 10 million
words 3835 adjectives) Forsgrenrsquos (1978) corpus of newspaper texts from Le
Monde and LrsquoExpress (1 million words 829 adjectives) and Larssonrsquos (1993)
corpus of travel catalogues travel guides and other non-literary prose (15
million words 113 adjectives)3 Using Wilmetrsquos corpus as the point of
comparison given that it contains the highest number of discrete lexemes
(3835 adjectives) an arbitrary cut-off point of 25 attestations or more was
used creating a list of 205 adjectives4 The total number of attestations pre-N
attestations and post-N attestations for each of the 205 adjectives across
BRUCE ANDERSON 291
the three corpora was then determined by adding Forsgrenrsquos and Larssonrsquos
data to that found in Wilmet if available
Results
The 205 most frequently attested adjectives in French texts appear to fall
along a continuum at one end of which we find 68 adjectives (or 332
percent) appearing either only in post-N position (nfrac1434) or typically in that
position (nfrac14 34) and at the other end of which we find 15 adjectives (or 73
percent) typically appearing in pre-N position as shown in Table 1 Note that
there are no adjectives appearing only in pre-N position The remaining 122
adjectives (or 595 percent) are regularly attested in both positions with
some having what we might call an affinity for post-N position (nfrac1441) or
pre-N position (nfrac14 29) but others being highly variable (nfrac1452)
Reminiscent of Holmesrsquo (1988) critique of the simplicity versus naturalness
of textbook rules such data raise some problems for the seemingly
straightforward fixed-order rules of adjective position exemplified in (1)ndash(3)
and the associated lists in (4) and (5) A textbook rule to the effect that
adjectives typically appear after the noun in Frenchmdasha word order that
clearly contrasts with what one finds in Englishmdashis only partly descriptively
Table 1 Frequency groupings of adjective position in written French texts
Frequency grouping Type count(nfrac14 205)
Percent oftotal
Examples
Exclusively pre-N (0post-N position)
0 0 mdash
Typically pre-N (below10 post-N position)
15 75 bon lsquogoodrsquo fameuxlsquo(in)famousrsquo petit lsquosmallrsquo
Variable but with anaffinity for pre-Nposition (10ndash39post-N position)
29 14 charmant lsquocharmingrsquo etrangelsquostrangersquo miserable lsquomiserablersquo
Highly variable (40ndash59post-N position)
52 255 delicieux lsquodeliciousrsquo joyeuxlsquojoyousrsquo puissant lsquopowerfulrsquo
Variable but with anaffinity for post-Nposition (60ndash90post-N position)
41 20 amer lsquobitterrsquo tiede lsquolukewarmrsquoserieux lsquoseriousrsquo
Typically post-N (above90 post-N)
34 17 blanc lsquowhitersquo inconnulsquounknownrsquo sauvage lsquowildrsquo
Exclusively post-N(100 post-Nposition)
34 17 allemand lsquoGermanrsquo dorelsquogoldenrsquo quotidien lsquodailyrsquo
292 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
accurate It is accurate in the sense that the ends of the continuum are
not equivalent in their percentages There are many more adjectives
appearing either only or typically in post-N position than is the case for pre-N
position It is inaccurate in the sense that it ignores the distributional
behavior of 595 percent of the most frequently used adjectives which can
appear in both positions French is not as the rule implies the mirror
opposite of English but instead is more accurately viewed as an lsquoANArsquo
(adjectivendashnounndashadjective) language as opposed to both lsquosystematic ANrsquo
languages like English and German and lsquosystematic NArsquo languages like
Indonesian and Thai (Cinque 1994 99ndash101)
The data also lead us to question the criteria textbooks use for including
certain adjectives in the two lists in (4) and (5) For the so-called prenominal
adjectives a frequency of 90 percent or greater in that position would
establish that adjectives such as excellent fameux lsquo(in)famousrsquo moindre
lsquoslightestrsquo and vrai lsquorealtruersquo need to be added That list itself implies a fixed
lexical class to be memorized though in reality it does nothing more
than provide a handful of frequently attested adjectives (nfrac14 15 [75 percent])
that typically appear just in that position Many more adjectives can
appear in pre-N position as well For the so-called change-of-meaning
adjectives the data show that the list in (5) falls woefully short in terms of
representing those adjectives that actually do regularly appear in both
positions (nfrac14 52)
Even if it were to be established that such rules are accurate with respect
to colloquial French as spoken between native speakers and thus appropriate
for oral proficiency-oriented classrooms such rules obviously fail to capture
the state of affairs in written texts particularly literature to which learners at
later course levels are exposed Yet even students at earlier course levels
occasionally find counter-examples to what they are taught The four cases of
lsquounexpectedrsquo pre-N position in (6) come from the cultural readings of two
first-year and two second-year US textbooks respectively
(6) a La France a un excellent systeme de transports publics(Magnan et al 2002 223)lsquoFrance has an excellent system of public transportrsquo
b Cela explique le grand nombre de magnifiques chateauxdans la region (Valdman and Pons 1997 395)lsquoThis explains the great number of magnificent chateaux inthe regionrsquo
c [U]ne veritable explosion de la technologie (St Onge etal 2003 71) lsquoa veritable explosion of technologyrsquo
d Lrsquohistoire est une douloureuse suite de tragedieshumaines (Oates and Dubois 2003 208)lsquoIts history is a painful series of human tragediesrsquo
One might wish to argue that the examples provided in (6) demonstrate
that inputmdashat least classroom-related inputmdashis not as contradictory to
descriptive norms as argued in this section Examples of the type in (6)
BRUCE ANDERSON 293
however are only occasionally found in textbook reading materials they are
included here to demonstrate that even when making a concerted effort to
write level-appropriate cultural readings textbook authors sometimes break
the very grammar rules they present elsewhere in the textbook suggesting a
true disparity between rules and lsquonaturalisticrsquo usage5 (I return to the issue of
descriptively accurate statements about French adjective position in the
discussion section)
FREQUENCY OF ADJECTIVE USE AND POSITIONIN CLASSROOM INPUT
Methodology
In order to get a general sense of how well the classroom setting as primary
input source reflects either textbook rules or the corpora-based descriptive
norm just summarized frequency counts of adjective type (ie discrete
lexemes) and their position in classroom input were collected by the
researcher from 60 hours of in-class observations of university-level French
courses at a large Midwestern American university Ten hours were spent in
each of six courses spanning the curriculummdashfrom the four semesters
comprising the first- and second-year language skills curriculum (referred to
as Levels 1a 1b 2a and 2b below) to a third-year conversation course
(Level 3) and a fourth-year lecture course on French civilization (Level 4)
Each of these courses was taught almost exclusively in French by a mix of
native and nonnative instructors none of whom were aware of the purpose
of this observation period Class observations were conducted on days when
adjective position was not the focus of explicit instruction as this would have
more than likely biased the frequency counts
The first- and second-year courses consisted of explicit grammar
instruction drills and communicative activities normally involving students
in pair or group work The conversation and civilization courses involved
no explicit grammar instruction the former involved mostly teacherndash
student and studentndashstudent oral communication based on previous course
readings and video segments whereas the latter was almost exclusively
devoted to instructor-delivered lectures Error correction was minimal in all
courses and no case of error correction involving adjective position was
witnessed during the observation period (despite the fact that some errors
were made)
Oral input on adjective position in all courses came from one of three
sources the instructor audio aides (such as cassette and videotape) and
the students themselves Written input in the first- and second-year
courses came from two sources visual aids (such as handouts transparencies
writing on the blackboard) and passages from the textbook that were read
silently or aloud in class At Level 3 written materials included authentic
newspaper and magazine articles in addition to visual aides at Level 4
294 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
a number of works of French literature were read outside of class Note
that written input in these latter two course levels is qualitatively the same
as that used in the corpora compiled by Forsgren (1978) Wilmet (1980) and
Larsson (1993)
At issue during this period of investigation was classroom input that could
be seen andor heard by all students in the classroom (as well as the
researcher) Input from students involved in pair and group work was
therefore excluded from analysis Material to be read outside of class was
also excluded as there was no guarantee that all students had actually
read such material Although the totality of inputmdashincluding sources of input
the learner might have voluntarily sought out outside of classmdashcould
obviously never be exhaustively tallied the exclusion of certain forms of
input was first and foremost a matter of logistics (ie not having the
appropriate funds equipment facilities and time to sound record then
transcribe all classroom input from every participant in every situation of
use) Despite this drawback I see no reason to suspect that student-delivered
input to other students during pair- or group-work would be qualitatively
different (at least in terms of adjective position) from the input delivered by
students that was audible to the entire class this latter source of input
already constitutes 10ndash30 percent of all sources of input tallied as part of the
observation period
Each instance of adjective use during each class session was recorded with
pen and paper by the researcher A coding scheme was devised whereby an
11 17-inch sheet of paper was divided into quadrants labeled according to
the source of the input lsquoinstructorrsquo lsquostudentrsquo lsquoaudiorsquo and lsquowrittenrsquo Each
quadrant was then subdivided according to the position of adjectival
modification lsquopre-Nrsquo lsquopost-Nrsquo lsquopredicatersquo and lsquootherrsquo (to be described
below) Input of the form une bonne idee lsquoa good idearsquo by the instructor was
therefore recorded as an instance of the adjective bonne in the instructor
quadrant pre-N sub-quadrant Hatch marks next to the adjective bonne were
used for any further exemplars of this adjective (including its masculine form
bon) in this position from this source I acknowledge that mere human error
caused by fatigue etc may have resulted in tallies that are less accurate than
if each session were recorded on tape Though the results presented below
should be considered with some caution then occasional errors could not
have reached a significant enough level to detract from the highly salient
patterns of usage that we will find
The resulting sixty sheets each representing one 50-minute class session
were then tallied in order to determine (a) the relative contribution of the
four sources of input (b) the average token (exemplar) count of adjectives
per 50-minute class period (c) the ratio of pre-N to post-Npredicate tokens
(d) the type count (number of discrete lexemes) entering into pre-N and
post-N positions and (e) the frequency of variation in positionmdashthat is the
extent to which a given adjective appeared at least once in both pre-N and
post-N position during the ten hours of observation
BRUCE ANDERSON 295
Results
The instructor not surprisingly was the primary provider of input in the
classroom (See Figure 1 available to online subscribers at http
applijoxfordjournalsorg) This was especially the case in the first-year
courses (Levels 1a 1b) the second- and third-year courses (Levels 2a 2b 3)
showed a greater proportion of input from students and from audio aides
The instructor was by far the predominant source of input in the fourth-year
course (Level 4) given that it involved instructor-delivered lectures on
French civilization with minimal class participation on the part of students
Concerning the average number of times an adjective occurred in the
input during any one 50-minute class period we find a general increase with
level from 42 tokens at Level 1a to 156 tokens per class at Level 4 (See
Figure 2 available to online subscribers at httpapplijoxfordjournalsorg)
Students therefore heard an adjective used anywhere from one to three
times per minute on average
Aggregate data on the frequency with which adjectives appeared in pre-N
post-N and predicate positions or in other constructions (eg used in
isolation in elliptical constructions such as la charmante lsquothe charming [one]rsquo
and following indefinite pronouns as in quelqursquoun de charmant lsquosomeone
charmingrsquo) demonstrate that post-N and predicate positions were generally
equally frequent and together constitute fully 75 percent of the ratio whereas
pre-N position never surpassed 27 percent across course levels Other
constructions were minimal (See Figure 3 available to online subscribers at
httpapplijoxfordjournalsorg)
This roughly 75 percentndash25 percent ratio was then reanalyzed in order to
consider the number of distinct lexemes (lsquotypesrsquo) that make up the
percentages We find for example that only 15 lexemes make up the 14
percent of pre-N adjective use at Level 1a This number gradually increased
with course level but only reached a total of 41 by Level 4 Post-N position
on the other hand admitted a much larger variety of adjectives in classroom
input from 71 at Level 1a to 323 by Level 4 (See Figure 4 available to online
subscribers at httpapplijoxfordjournalsorg) Students therefore got
relatively little input on pre-N position and what input they did get was
made up of a relatively small class of adjectives being used over and over
again and which already figure in the textbook-determined list of pre-N
adjectives in (4) (See Appendix A available to online subscribers at http
applijoxfordjournalsorg for a list of adjectives appearing in pre-N position
by course level)
Finally we find that very few adjectives varied in position during the 10
hours of in-class observation There was no variation attested at Level 1a
and only four to seven adjectives varied in position from Level 1b through to
Level 3 Moreover some of the input provided to learners in these courses
in the form of other studentsrsquo output was nonnativelike6 At the most
advanced undergraduate level learners got somewhat more input involving
296 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
variation in position where 20 adjectives did so (See Figure 5 available to
online subscribers at httpapplijoxfordjournalsorg) Again most (but not
all) of these adjectives already figure into the textbook-determined list of
variable adjectives in (5) (See Appendix B available to online subscribers at
httpapplijoxfordjournalsorg for a list of adjectives varying in position by
course level)
To summarize the observational data just presented the ambient language
of the classroom clearly mimics the pedagogical treatment that American
university-level classroom learners receive from textbooks Although
adjectives are in general a very frequent part of the input the rate of
pre-N position (at roughly 25 percent) is essentially static across levels
pointing to the fact that classroom input across levels is in no way
contradictory to what learners are initially taught Post-N position is basic
(unmarked) and pre-N position is exceptional (marked) The adjectives that
do appear in pre-N position in classroom input are in most cases those that
learners already expect to appear in that position (cf the list in (4) and
Appendix A) The few adjectives that appear in both positions are again in
most cases those that learners expect to do so (cf the list in (5) and
Appendix B)
By the end of the first-year level then both instruction and classroom
input would lead learners to expect that adjectives will appear in post-N
position with the notable exception of the class of adjectives in (4) (those in
(5) having not yet been introduced) By the second-year level it could be
argued that instruction provides examples in the form of so-called change of
meaning adjectives that variation in position is to be associated with
variation in interpretation (and perhaps by implication always to
interpretation) However learnersrsquo explicit knowledge of interpretative
differences as exemplified by the adjectives in (5) typically consists of
arbitrary and unsystematic English glosses By arbitrary I mean that
instruction never provides an adequate explanation for why the pre-N
position of cher should correspond to beloved or for its post-N position to
correspond to expensive7 Though telling students following Bragger and Rice
(1996) that the meaning in post-N position is lsquoconcretersquo and the meaning in
pre-N position is lsquofigurativersquo might (somehow) help it does not account for
other adjectives in (5) like prochain lsquonextrsquo This latter adjective also
demonstrates the fact that English glosses are unsystematic in that prochain
translates as lsquonextrsquo in either position
Before considering our final research question I note that no research on
adjective position in native French speaker discourse is available to date
An anonymous Applied Linguistics reviewer questions the validity of compar-
ing primarily spoken classroom discoursemdashor any other kind of spoken
discoursemdashto written texts without first demonstrating that adjective position
is insensitive to this difference in medium While recognizing the potentially
important distinction in medium and the need for further research in this
area the validity of comparing classroom input to written texts in the present
BRUCE ANDERSON 297
study lies in the fact that such a comparison roughly corresponds to the
curriculum of typical university-level courses an emphasis on listening
comprehension and oral communication in the first- and second-year
language courses leading to an emphasis on reading comprehension
(typically literature-based) and proficiency in writing in the third- and
fourth-year courses Were it to turn out that the degree of variation in
adjective position found in written texts is restricted to just that medium
revisions to pedagogy would necessarily be restricted to writing-intensive
(composition creative composition stylistics) courses at the third- and
fourth-year levels (I return to this issue in the discussion section)
LEARNER INTUITIONS OF ACCEPTABILITY
Thus far we have seen an incongruity between textbook presentations of
adjective position and authentic text sources across a number of written text
genres (research question 1) and that ambient language use involving
adjective position in the classroom is clearly more congruent with and might
therefore be taken to reinforce the explicit rules found in textbooks
(research question 2) We are now in a position to address our third research
questionmdashthe extent to which intuitions of acceptability among classroom
learners are biased by the congruity between pedagogical treatment and
classroom input frequency patterns An empirical study previously reported
in Anderson (2002 2007) can shed some light on this question
Methodology
Anderson (2002 2007) investigated the acquisition of various morphosyn-
tactic properties of French noun phrases by administering an acceptability
judgment task in which participants had to evaluate the appropriateness of
sentences featuring among other things the pre- versus post-N position of
adjectives Each of the adjectives selected were lsquohighly variablersquo with respect
to position (cf Table 1) yet crucially do not figure in textbook lists of
adjectives said to change meaning based on position (cf the list in (10))
These include lourd lsquoheavyrsquo violent lsquoviolentrsquo charmant lsquocharmingrsquo precieux
lsquopreciousrsquo epais lsquothickrsquo etrange lsquostrangersquo and delicieux lsquodeliciousrsquo (See
Appendix C available to online subscribers at httpapplijoxfordjournals
org for test sentences) Study participants included English-speaking learners
of French in the second third and fourth years of undergraduate study
(nfrac14 80) who were attending courses at the same institution at which the
observational data were collected a group of advanced learners (nfrac14 20)
consisting of graduate student instructors at that same institution as well as
secondary education teachers and a group of native French speakers (nfrac14 27)
roughly matched in age and education level
Each adjective appeared in test sentences once prenominally and once
postnominally with each test sentence preceded once by a context
298 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
motivating acceptance and once by a context motivating rejection of one or
the other adjective order for a total of four positionndashinterpretation pairings
per adjective These were randomly distributed across two testing
instruments administered a week apart After reading each context and
paired test sentence study participants were directed to check boxes
labeled lsquofinersquo lsquooddrsquo or lsquocannot decidersquo to the prompt This particular sentence
sounds ___ in this context
As an example each of the sentences in (7) featuring the adjective lourd
lsquoheavyrsquo was paired with each of two contexts
(7) a Pierre doit laisser la lourde valise avec son amib Pierre doit laisser la valise lourde a lrsquoaeroport
lsquoPierre has to leave the heavy suitcase with his friendatthe airportrsquo
In the first context motivating a unique noun-referent interpretation
there was one and only one suitcase which happened to be heavy and
which lsquoPierrersquo had to leave behind with his friend In the second context
motivating a multiple noun-referent interpretation there were a number of
suitcases only one of which was heavy and it was this one that lsquoPierrersquo had
to leave behind at the airport (See Appendix D available to online
subscribers at httpapplijoxfordjournalsorg for the complete version of
each context)
When viewed solely in terms of truth-conditional semantics each of the
four positionndashinterpretation pairings is compatible save for one pre-N
position (7a) in a multiple noun-referent context This is because the singular
definite article la and pre-N adjective lourde together presuppose a unique
suitcase in context (which happens to be heavy) This presupposition
does not hold when the adjective is in post-N position where all that is
required is that there be at least one such suitcase in context The testing
instrument then allows us to examine how classroom learners react to
an lsquoexpectedrsquo word order (ie sentences like (7b) in either context) as well
as to an lsquounexpectedrsquo but semantically and pragmatically appropriate
word order (ie sentences like (7a) but only in the unique noun-referent
context)
Results
As the mean acceptance rates from native French speakers in Table 2 show
truth-conditional semantics was not the only factor determining acceptance
rates Two of the pairings received essentially chance-level rates of
acceptance (pre-N position in a multiple noun-referent context at 49
percent and post-N position in a unique context at 52 percent) whereas the
other two pairings showed much clearer above- or below-chance rates (pre-N
position in a unique noun-referent context at 74 percent and post-N
position in a multiple noun-referent context at 37 percent)
BRUCE ANDERSON 299
As was argued in Anderson (2002 2007) results for each of the four
positionndashinterpretation pairings among the native French speakers are
explainable in terms of a disproportionate interaction between semantics and
pragmatic implicature in particular the Gricean maxim of quantity An
utterance should be no more and no less informative than is required for the
current purposes of the exchange (Grice 1975) The positionndashinterpretation
pairing that was lsquotruersquo in terms of semantics and also the most pragmatically
appropriate was accepted at the highest rate (74 percent) the pairing that
was also lsquotruersquo in terms of semantics but was pragmatically the least
appropriate was accepted at the lowest rate (37 percent)8
Do learner group response patterns show the same bias toward pragmatic
implicature in their acceptance rates The comparison of mean acceptance
rates between native French speakers and learner groups in Table 3 shows
that learners performed in a particular (and particularly non-Frenchlike) way
in which the interaction goes in the opposite direction favoring semantics
over pragmatics
Recall that the only positionndashinterpretation pairing that was semantically
incompatible was pre-N position (7a) in a multiple noun-referent context
given the presupposition of uniqueness of the noun referent All learner
groups did indeed accept this pairing at the lowest rate of the four and with
the exception of the second-year learners (at thorn9 percent) did so more
strongly than the native French speakers at mean acceptance rates that were
11ndash15 percent lower)9 However they also accepted at much lower mean
rates (12 to 23 percent) the most felicitous positionndashinterpretation pairing
for the native French speakers pre-N position in a unique noun-referent
context Recall also that post-N position (7b) was semantically compatible
under either interpretation All learner groups accepted these two positionndash
interpretation pairings at much higher rates than the native French speakers
(thorn23 to thorn31 percent) including the least felicitous positionndashinterpretation
pairing for the native French speakersmdashpost-N position in a multiple noun-
referent context This is precisely what one would expect from the
pedagogical treatment these learners are provided and the congruity of
that treatment with classroom input which under-represents actual usage in
Table 2 Native French speaker mean acceptance rates for the uniquenonunique noun-referent distinction
Prenominal (AN)order ()
Postnominal (NA)order ()
Unique noun referent 74 52
Nonunique noun referent 49 37
Source Adapted from Anderson 2002 2007
300 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
text-based corpora and as we have now seen differs from native speaker
intuitions of overall (semantico-pragmatic) acceptability
Although a bias toward or even strict adherence to explicit pedagogical
rules in evaluating such sentences may not seem all that surprising what does
stand out in the data from Anderson (2002 2007) is that these differential
rates in acceptance between native and nonnative speakers are as true for the
advanced group in that studymdashconsisting of graduate student instructors and
second education teachersmdashas they are for the intermediate (second-year)
learner group This would seem to indicate no increased sensitivity to
contextual (pragmatic) factors in adjective position over the course of studying
French even though the advanced group had been exposed to much more
authentic input through texts written for and by native French speakers (ie
the descriptive norm uncovered in our examination of authentic text corpora)
DISCUSSION
What should be clear from the findings presented here is that rules of French
adjective placement that one typically finds in (at least American) textbooks
though not plainly incorrect are not fully descriptively accurate either at
least with respect to written French across several text genres In particular
such rules portray pre-N and post-N positions as mutually exclusive for
almost all adjectives when in fact they are not
Similar discrepancies are often found between rules for and by native
speakers and what those speakers actually do One need look no further than
the French for an example with colloquial speech often at odds with the
prescriptive bon usage and the admonishments of the Academie francaise One
might therefore also expect to find a discrepancy between textbook rules and
what teachers do in the classroom andor what students find to be
lsquoacceptablersquo in terms of word order As we have seen however this is not
the case These rules are faithfully reflected in classroom language use by
instructors to such an extent that when asked to evaluate the acceptability
of lsquoexpectedrsquo and lsquounexpected but nativelikersquo word orders (and one would
presume in production as well) both instructors and students paid much
less attention to problems in pragmatic implicature than did native speakers
a situation which does not appear to change even in the face of counter-
examples occasionally provided in the textbooks they use (see (6) above) and
most certainly in the written texts that they read at later course levels As
pointed out by an anonymous reviewer this situation can be considered a
testament to the power of classroom instruction (instructor input in
particular) if instruction is changed he or she reasons it is likely that
classroom learners will reflect this change
But should instruction be changed If so who along the continuum from
the theorist-researcher to the teacher-practitioner should inform and guide
this change At which level In what way Such questions are at the very
heart of two on-going debates that I re-examine before proposing some
BRUCE ANDERSON 301
answers the pedagogical versus descriptive norm debate (eg Owen 1993
Widdowson 2000 and the collection of articles in Gass et al 2002) and
the interrelated native-speaker-usage-as-goal debate in foreign language
pedagogy (eg the collection of articles in Blyth 2003)
On the whole the approach to instruction on French adjective
position outlined earlier appears advantageous in its simplicity providing
easily remembered rules of thumb that learners can apply in language
production That this approach to instruction does not as we have seen
straightforwardly follow the descriptive norm is a situation that linguists
utilizing a corpus-based methodology believe can and should be remedied
Conrad (2000) for instance sees in corpus-based research an immediate way
in which linguistics can inform language pedagogy freeing language teachers
from lsquohav[ing] to rely on judgments of grammatical accuracy or on native
speakersrsquo intuition about what sounds naturalrsquo (2000 555) Although the
subsequent study by Biber and Reppen (2002) uses a disclaimer to the effect
that frequency should not be the sole factor in material design they
nevertheless suggest that lsquoa selective revision of pedagogy to reflect actual
use as shown by frequency studies could result in radical changes that
facilitate the learning process for studentsrsquo (2002 199) Owen (1993)
however provides examples of how a total reliance on a corpus does not
necessarily yield better observation and more importantly questions
whether better observation automatically equates to better explanation
(1993 168) (but see Francis and Sinclair 1994 for rebuttal) For his part
Widdowson (2000) objects to the kind of equation just described as a
case of linguistics applied a mistaken belief that what is textually attested
lsquouniquely represents real language and that this reality should define the
foreign language subjectrsquo (2000 8) (but see Stubbs 2001 for rebuttal) What
is needed according to Widdowson is not simply linguistics applied
to pedagogy but a true role for applied linguistics an approach to usage that
also recognizes native speakersrsquo awareness of and opinions about such usage
The need for a truly applied linguistics perspective on native speaker
usage (including frequency patterns) is embodied in a pedagogical norm
originally conceptualized in Valdman (1989) and elaborated upon in
Bardovi-Harlig and Gass (2002) as
[A] combination of language systems and forms selected by linguistsand pedagogues to serve as the immediate language target ortargets that learners seek to acquire during their language studyIn other words pedagogical norms represent a mid-point or seriesof mid-points for learners as they progress toward acquiringnative language norms [thus constituting] a form of languagethat is acceptable to native speakers but easier to learn than the fullnative language system (p 3)
The textbook word-order rules for French adjective position reviewed
earlier would first appear to constitute an appropriate pedagogical norm
302 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
under Bardovi-Harlig and Gassrsquos (2002) definition because the learner is able
to produce a subset of the totality of possible utterances deemed acceptable to
native speakers which may turn out to be even more the case in spoken than
in written French (though the difference in mediums remains to be studied)
The learner data presented here from Anderson (2002 2007) however
provides one piece of evidence that textbook rules and classroom input do not
constitute a sufficient pedagogical norm as learners have stopped in their
progress toward acquiring native speaker (ie descriptively accurate) norms
As the articles in Blyth (2003) attest however it is highly debatable
whether the lsquoNative Speakerrsquo (as an idealized abstraction) should be the
appropriate model for the contemporary learner For Kramsch (2002 2003)
too much of a focus on an authentic native speaker norm reduces
foreign language (FL) study to a constant lsquoapproximation to and conformity
with stereotypical native speakers thus often silencing other FL-specific
forms of language potentialrsquo (2002 60) Language learning in instructional
settings on my understanding of Kramschrsquos (2003) argument should not be
viewed solely in a teleological sensemdashas language acquisition directed
toward a fixed nativelike end-pointmdashbut rather as language appropriation
by learners in the sense that they are constructing a linguistic and even
cultural identity lsquoin the interstices of national languages and on the
margins of monolingual speakersrsquo territoriesrsquo (2003 260) The political nature
of the debate as highlighted by Kramsch becomes all the more apparent
when one additionally considers the status of the foreign language on
the world stage (eg French as an international language in retreat)
and how deeply the normative attitudes of its speakers run (eg Francersquos
three-century-long history of acceptance and maintenance of the standard
(Lodge 1993))
The foregoing synopsis of the debates surrounding native speaker usage as
both the goal of foreign language instruction and the extent to which
pedagogical norms need apply has highlighted the real world complexities
involved in what would otherwise seem to be a simple equation Revise
pedagogy to reflect actual use as shown by frequency studies and this
will result in a more targetlike performance on the part of learners So we
return then to the questions posed earlier (1) Should instruction be
changed (2) If so who along the continuum from the theorist-researcher to
the teacher-practitioner should inform and guide this change (3) At which
level (4) In what ways
With respect to our first question because one can empirically
demonstrate that the pedagogical norm for French adjective position
prevents learners from eventually acquiring native speaker (ie descriptively
accurate) norms instruction should be changed so thatmdashirrespective of
whether the attainment of such a norm is necessarily desirablemdashlearners
have both a choice in selecting one grammatical or lexical form over the other
and an awareness of the meaning potential of each choice (following Kramsch
2002 71)
BRUCE ANDERSON 303
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
in both adnominal positions prenominally (pre-N) and postnominally
(post-N) following certain lexical and contextual (pragmatic) constraints as
to their interpretation in one position versus the other (see eg Delmonier
1980 for a general review Delbecque 1990 for an account within cognitive
grammar Anderson 2002 for a compositional account within generative
syntax) This state of affairs leads us to question whether the traditional
textbook rules relating to adjectival modification in French reflect
attested native-speaker language use any better than that uncovered by
Biber and Reppen in authentic English-language texts across a number of
genres
English-speaking learners of French at least in the typical American
university setting are provided with three adjective placement rules
which take the form of a general rule with two exceptions Most adjectives
in French are said to be post-N (as in (1)) though a few are pre-N (as in (2))
and a few more are said to change meaning based on their position
(as in (3))
(1) une voiture rougelsquoa red carrsquo
(2) une petite maisonlsquoa small housersquo
(3) a un bijou cher un cher bijoulsquoan expensive jewelrsquo lsquoa cherished jewelrsquob une eglise ancienne une ancienne eglise
lsquoan ancient churchrsquo lsquoa former churchrsquo
First- and second-year textbooks of French published in the USA are
strikingly similar in the statement of such rules and in their order of
presentation though they differ to some degree as to which adjectives are to
be included in the (putative) lexical classes constituting exceptions to the
general post-N rule The textbooks used in the classrooms observed for the
present study are no different In the first-year textbook Chez nous (Valdman
and Pons 1997) students are first told that most adjectives follow the noun
(1997 73) but that the twelve adjectives listed in (4) typically and
exceptionally precede the noun (1997 241)
(4) a jeune vieux lsquoyoung oldrsquob nouveau lsquonewrsquoc petit grand gros lsquosmalllittle bigtall largersquod beau lsquohandsomersquoe joli lsquoprettyrsquof bon mauvais lsquogood badrsquog premier dernier lsquofirst lastrsquo
In the second-year textbook Quant a moi (Bragger and Rice 1996) the basic
post-N rule is reviewed as is the class of adjectives that exceptionally appear
prenominally (1996 33) Note however that this class of exceptions has
eleven members instead of twelve having lost some members from the Chez
290 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
Nous list in (4) (eg bon premier dernier gros) but gaining others (eg long
lsquolongrsquo court lsquoshortrsquo and autre lsquootherrsquo)
Bragger and Rice (1996) then present the second exception to the basic
post-N rule wherein students are told that lsquoa number of adjectives change
meaning depending on whether theyrsquore placed after or before a noun
In general an adjective placed after the noun retains its basic concrete
meaning The same adjective placed before the noun is used in an abstract or
figurative mannerrsquo (1996 34 emphasis in original) The seven adjectives that
apparently fall into this class are listed in (5) The first English gloss captures
the meaning derived from pre-N position the second gloss that derived from
post-N position Note that Bragger and Rice put the adjective grand in both
lists of exceptions In other words grand is apparently exceptionally pre-N
and exceptionally variable at the same time2
(5) ADJECTIVE PRE-N POST-Na ancien lsquoformerrsquo vs lsquoold ancientrsquob cher lsquodear well-lovedrsquo vs lsquoexpensiversquoc dernier lsquolast latest (in a series)rsquo vs lsquolast (before this one)rsquod grand lsquogreatrsquo vs lsquotall largersquoe pauvre lsquopoor (unfortunate)rsquo vs lsquopoor (not rich)rsquof prochain lsquonext (in a series)rsquo vs lsquonext (after this one)rsquog propre lsquoownrsquo vs lsquocleanrsquo
To summarize the pedagogical treatment of French adjective position
commonly found in university-level American textbooks presents post-N
position as the most frequent and typical (frac14unmarked) adjective order
the less frequent and exceptional (frac14marked) pre-N order is either required
when the adjective is a member of a seemingly fixed lexical class such as
the list in (4) or licensed by a particular interpretation that differs from that
obtained in post-N position such as the list in (5) This difference in
interpretation is often though not always captured through distinct English
glosses
Methodology
In order to determine the extent to which the pedagogical treatment just
summarized is congruent with authentic French texts three corpora were
consulted Wilmetrsquos (1980) corpus of literary texts (approximately 10 million
words 3835 adjectives) Forsgrenrsquos (1978) corpus of newspaper texts from Le
Monde and LrsquoExpress (1 million words 829 adjectives) and Larssonrsquos (1993)
corpus of travel catalogues travel guides and other non-literary prose (15
million words 113 adjectives)3 Using Wilmetrsquos corpus as the point of
comparison given that it contains the highest number of discrete lexemes
(3835 adjectives) an arbitrary cut-off point of 25 attestations or more was
used creating a list of 205 adjectives4 The total number of attestations pre-N
attestations and post-N attestations for each of the 205 adjectives across
BRUCE ANDERSON 291
the three corpora was then determined by adding Forsgrenrsquos and Larssonrsquos
data to that found in Wilmet if available
Results
The 205 most frequently attested adjectives in French texts appear to fall
along a continuum at one end of which we find 68 adjectives (or 332
percent) appearing either only in post-N position (nfrac1434) or typically in that
position (nfrac14 34) and at the other end of which we find 15 adjectives (or 73
percent) typically appearing in pre-N position as shown in Table 1 Note that
there are no adjectives appearing only in pre-N position The remaining 122
adjectives (or 595 percent) are regularly attested in both positions with
some having what we might call an affinity for post-N position (nfrac1441) or
pre-N position (nfrac14 29) but others being highly variable (nfrac1452)
Reminiscent of Holmesrsquo (1988) critique of the simplicity versus naturalness
of textbook rules such data raise some problems for the seemingly
straightforward fixed-order rules of adjective position exemplified in (1)ndash(3)
and the associated lists in (4) and (5) A textbook rule to the effect that
adjectives typically appear after the noun in Frenchmdasha word order that
clearly contrasts with what one finds in Englishmdashis only partly descriptively
Table 1 Frequency groupings of adjective position in written French texts
Frequency grouping Type count(nfrac14 205)
Percent oftotal
Examples
Exclusively pre-N (0post-N position)
0 0 mdash
Typically pre-N (below10 post-N position)
15 75 bon lsquogoodrsquo fameuxlsquo(in)famousrsquo petit lsquosmallrsquo
Variable but with anaffinity for pre-Nposition (10ndash39post-N position)
29 14 charmant lsquocharmingrsquo etrangelsquostrangersquo miserable lsquomiserablersquo
Highly variable (40ndash59post-N position)
52 255 delicieux lsquodeliciousrsquo joyeuxlsquojoyousrsquo puissant lsquopowerfulrsquo
Variable but with anaffinity for post-Nposition (60ndash90post-N position)
41 20 amer lsquobitterrsquo tiede lsquolukewarmrsquoserieux lsquoseriousrsquo
Typically post-N (above90 post-N)
34 17 blanc lsquowhitersquo inconnulsquounknownrsquo sauvage lsquowildrsquo
Exclusively post-N(100 post-Nposition)
34 17 allemand lsquoGermanrsquo dorelsquogoldenrsquo quotidien lsquodailyrsquo
292 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
accurate It is accurate in the sense that the ends of the continuum are
not equivalent in their percentages There are many more adjectives
appearing either only or typically in post-N position than is the case for pre-N
position It is inaccurate in the sense that it ignores the distributional
behavior of 595 percent of the most frequently used adjectives which can
appear in both positions French is not as the rule implies the mirror
opposite of English but instead is more accurately viewed as an lsquoANArsquo
(adjectivendashnounndashadjective) language as opposed to both lsquosystematic ANrsquo
languages like English and German and lsquosystematic NArsquo languages like
Indonesian and Thai (Cinque 1994 99ndash101)
The data also lead us to question the criteria textbooks use for including
certain adjectives in the two lists in (4) and (5) For the so-called prenominal
adjectives a frequency of 90 percent or greater in that position would
establish that adjectives such as excellent fameux lsquo(in)famousrsquo moindre
lsquoslightestrsquo and vrai lsquorealtruersquo need to be added That list itself implies a fixed
lexical class to be memorized though in reality it does nothing more
than provide a handful of frequently attested adjectives (nfrac14 15 [75 percent])
that typically appear just in that position Many more adjectives can
appear in pre-N position as well For the so-called change-of-meaning
adjectives the data show that the list in (5) falls woefully short in terms of
representing those adjectives that actually do regularly appear in both
positions (nfrac14 52)
Even if it were to be established that such rules are accurate with respect
to colloquial French as spoken between native speakers and thus appropriate
for oral proficiency-oriented classrooms such rules obviously fail to capture
the state of affairs in written texts particularly literature to which learners at
later course levels are exposed Yet even students at earlier course levels
occasionally find counter-examples to what they are taught The four cases of
lsquounexpectedrsquo pre-N position in (6) come from the cultural readings of two
first-year and two second-year US textbooks respectively
(6) a La France a un excellent systeme de transports publics(Magnan et al 2002 223)lsquoFrance has an excellent system of public transportrsquo
b Cela explique le grand nombre de magnifiques chateauxdans la region (Valdman and Pons 1997 395)lsquoThis explains the great number of magnificent chateaux inthe regionrsquo
c [U]ne veritable explosion de la technologie (St Onge etal 2003 71) lsquoa veritable explosion of technologyrsquo
d Lrsquohistoire est une douloureuse suite de tragedieshumaines (Oates and Dubois 2003 208)lsquoIts history is a painful series of human tragediesrsquo
One might wish to argue that the examples provided in (6) demonstrate
that inputmdashat least classroom-related inputmdashis not as contradictory to
descriptive norms as argued in this section Examples of the type in (6)
BRUCE ANDERSON 293
however are only occasionally found in textbook reading materials they are
included here to demonstrate that even when making a concerted effort to
write level-appropriate cultural readings textbook authors sometimes break
the very grammar rules they present elsewhere in the textbook suggesting a
true disparity between rules and lsquonaturalisticrsquo usage5 (I return to the issue of
descriptively accurate statements about French adjective position in the
discussion section)
FREQUENCY OF ADJECTIVE USE AND POSITIONIN CLASSROOM INPUT
Methodology
In order to get a general sense of how well the classroom setting as primary
input source reflects either textbook rules or the corpora-based descriptive
norm just summarized frequency counts of adjective type (ie discrete
lexemes) and their position in classroom input were collected by the
researcher from 60 hours of in-class observations of university-level French
courses at a large Midwestern American university Ten hours were spent in
each of six courses spanning the curriculummdashfrom the four semesters
comprising the first- and second-year language skills curriculum (referred to
as Levels 1a 1b 2a and 2b below) to a third-year conversation course
(Level 3) and a fourth-year lecture course on French civilization (Level 4)
Each of these courses was taught almost exclusively in French by a mix of
native and nonnative instructors none of whom were aware of the purpose
of this observation period Class observations were conducted on days when
adjective position was not the focus of explicit instruction as this would have
more than likely biased the frequency counts
The first- and second-year courses consisted of explicit grammar
instruction drills and communicative activities normally involving students
in pair or group work The conversation and civilization courses involved
no explicit grammar instruction the former involved mostly teacherndash
student and studentndashstudent oral communication based on previous course
readings and video segments whereas the latter was almost exclusively
devoted to instructor-delivered lectures Error correction was minimal in all
courses and no case of error correction involving adjective position was
witnessed during the observation period (despite the fact that some errors
were made)
Oral input on adjective position in all courses came from one of three
sources the instructor audio aides (such as cassette and videotape) and
the students themselves Written input in the first- and second-year
courses came from two sources visual aids (such as handouts transparencies
writing on the blackboard) and passages from the textbook that were read
silently or aloud in class At Level 3 written materials included authentic
newspaper and magazine articles in addition to visual aides at Level 4
294 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
a number of works of French literature were read outside of class Note
that written input in these latter two course levels is qualitatively the same
as that used in the corpora compiled by Forsgren (1978) Wilmet (1980) and
Larsson (1993)
At issue during this period of investigation was classroom input that could
be seen andor heard by all students in the classroom (as well as the
researcher) Input from students involved in pair and group work was
therefore excluded from analysis Material to be read outside of class was
also excluded as there was no guarantee that all students had actually
read such material Although the totality of inputmdashincluding sources of input
the learner might have voluntarily sought out outside of classmdashcould
obviously never be exhaustively tallied the exclusion of certain forms of
input was first and foremost a matter of logistics (ie not having the
appropriate funds equipment facilities and time to sound record then
transcribe all classroom input from every participant in every situation of
use) Despite this drawback I see no reason to suspect that student-delivered
input to other students during pair- or group-work would be qualitatively
different (at least in terms of adjective position) from the input delivered by
students that was audible to the entire class this latter source of input
already constitutes 10ndash30 percent of all sources of input tallied as part of the
observation period
Each instance of adjective use during each class session was recorded with
pen and paper by the researcher A coding scheme was devised whereby an
11 17-inch sheet of paper was divided into quadrants labeled according to
the source of the input lsquoinstructorrsquo lsquostudentrsquo lsquoaudiorsquo and lsquowrittenrsquo Each
quadrant was then subdivided according to the position of adjectival
modification lsquopre-Nrsquo lsquopost-Nrsquo lsquopredicatersquo and lsquootherrsquo (to be described
below) Input of the form une bonne idee lsquoa good idearsquo by the instructor was
therefore recorded as an instance of the adjective bonne in the instructor
quadrant pre-N sub-quadrant Hatch marks next to the adjective bonne were
used for any further exemplars of this adjective (including its masculine form
bon) in this position from this source I acknowledge that mere human error
caused by fatigue etc may have resulted in tallies that are less accurate than
if each session were recorded on tape Though the results presented below
should be considered with some caution then occasional errors could not
have reached a significant enough level to detract from the highly salient
patterns of usage that we will find
The resulting sixty sheets each representing one 50-minute class session
were then tallied in order to determine (a) the relative contribution of the
four sources of input (b) the average token (exemplar) count of adjectives
per 50-minute class period (c) the ratio of pre-N to post-Npredicate tokens
(d) the type count (number of discrete lexemes) entering into pre-N and
post-N positions and (e) the frequency of variation in positionmdashthat is the
extent to which a given adjective appeared at least once in both pre-N and
post-N position during the ten hours of observation
BRUCE ANDERSON 295
Results
The instructor not surprisingly was the primary provider of input in the
classroom (See Figure 1 available to online subscribers at http
applijoxfordjournalsorg) This was especially the case in the first-year
courses (Levels 1a 1b) the second- and third-year courses (Levels 2a 2b 3)
showed a greater proportion of input from students and from audio aides
The instructor was by far the predominant source of input in the fourth-year
course (Level 4) given that it involved instructor-delivered lectures on
French civilization with minimal class participation on the part of students
Concerning the average number of times an adjective occurred in the
input during any one 50-minute class period we find a general increase with
level from 42 tokens at Level 1a to 156 tokens per class at Level 4 (See
Figure 2 available to online subscribers at httpapplijoxfordjournalsorg)
Students therefore heard an adjective used anywhere from one to three
times per minute on average
Aggregate data on the frequency with which adjectives appeared in pre-N
post-N and predicate positions or in other constructions (eg used in
isolation in elliptical constructions such as la charmante lsquothe charming [one]rsquo
and following indefinite pronouns as in quelqursquoun de charmant lsquosomeone
charmingrsquo) demonstrate that post-N and predicate positions were generally
equally frequent and together constitute fully 75 percent of the ratio whereas
pre-N position never surpassed 27 percent across course levels Other
constructions were minimal (See Figure 3 available to online subscribers at
httpapplijoxfordjournalsorg)
This roughly 75 percentndash25 percent ratio was then reanalyzed in order to
consider the number of distinct lexemes (lsquotypesrsquo) that make up the
percentages We find for example that only 15 lexemes make up the 14
percent of pre-N adjective use at Level 1a This number gradually increased
with course level but only reached a total of 41 by Level 4 Post-N position
on the other hand admitted a much larger variety of adjectives in classroom
input from 71 at Level 1a to 323 by Level 4 (See Figure 4 available to online
subscribers at httpapplijoxfordjournalsorg) Students therefore got
relatively little input on pre-N position and what input they did get was
made up of a relatively small class of adjectives being used over and over
again and which already figure in the textbook-determined list of pre-N
adjectives in (4) (See Appendix A available to online subscribers at http
applijoxfordjournalsorg for a list of adjectives appearing in pre-N position
by course level)
Finally we find that very few adjectives varied in position during the 10
hours of in-class observation There was no variation attested at Level 1a
and only four to seven adjectives varied in position from Level 1b through to
Level 3 Moreover some of the input provided to learners in these courses
in the form of other studentsrsquo output was nonnativelike6 At the most
advanced undergraduate level learners got somewhat more input involving
296 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
variation in position where 20 adjectives did so (See Figure 5 available to
online subscribers at httpapplijoxfordjournalsorg) Again most (but not
all) of these adjectives already figure into the textbook-determined list of
variable adjectives in (5) (See Appendix B available to online subscribers at
httpapplijoxfordjournalsorg for a list of adjectives varying in position by
course level)
To summarize the observational data just presented the ambient language
of the classroom clearly mimics the pedagogical treatment that American
university-level classroom learners receive from textbooks Although
adjectives are in general a very frequent part of the input the rate of
pre-N position (at roughly 25 percent) is essentially static across levels
pointing to the fact that classroom input across levels is in no way
contradictory to what learners are initially taught Post-N position is basic
(unmarked) and pre-N position is exceptional (marked) The adjectives that
do appear in pre-N position in classroom input are in most cases those that
learners already expect to appear in that position (cf the list in (4) and
Appendix A) The few adjectives that appear in both positions are again in
most cases those that learners expect to do so (cf the list in (5) and
Appendix B)
By the end of the first-year level then both instruction and classroom
input would lead learners to expect that adjectives will appear in post-N
position with the notable exception of the class of adjectives in (4) (those in
(5) having not yet been introduced) By the second-year level it could be
argued that instruction provides examples in the form of so-called change of
meaning adjectives that variation in position is to be associated with
variation in interpretation (and perhaps by implication always to
interpretation) However learnersrsquo explicit knowledge of interpretative
differences as exemplified by the adjectives in (5) typically consists of
arbitrary and unsystematic English glosses By arbitrary I mean that
instruction never provides an adequate explanation for why the pre-N
position of cher should correspond to beloved or for its post-N position to
correspond to expensive7 Though telling students following Bragger and Rice
(1996) that the meaning in post-N position is lsquoconcretersquo and the meaning in
pre-N position is lsquofigurativersquo might (somehow) help it does not account for
other adjectives in (5) like prochain lsquonextrsquo This latter adjective also
demonstrates the fact that English glosses are unsystematic in that prochain
translates as lsquonextrsquo in either position
Before considering our final research question I note that no research on
adjective position in native French speaker discourse is available to date
An anonymous Applied Linguistics reviewer questions the validity of compar-
ing primarily spoken classroom discoursemdashor any other kind of spoken
discoursemdashto written texts without first demonstrating that adjective position
is insensitive to this difference in medium While recognizing the potentially
important distinction in medium and the need for further research in this
area the validity of comparing classroom input to written texts in the present
BRUCE ANDERSON 297
study lies in the fact that such a comparison roughly corresponds to the
curriculum of typical university-level courses an emphasis on listening
comprehension and oral communication in the first- and second-year
language courses leading to an emphasis on reading comprehension
(typically literature-based) and proficiency in writing in the third- and
fourth-year courses Were it to turn out that the degree of variation in
adjective position found in written texts is restricted to just that medium
revisions to pedagogy would necessarily be restricted to writing-intensive
(composition creative composition stylistics) courses at the third- and
fourth-year levels (I return to this issue in the discussion section)
LEARNER INTUITIONS OF ACCEPTABILITY
Thus far we have seen an incongruity between textbook presentations of
adjective position and authentic text sources across a number of written text
genres (research question 1) and that ambient language use involving
adjective position in the classroom is clearly more congruent with and might
therefore be taken to reinforce the explicit rules found in textbooks
(research question 2) We are now in a position to address our third research
questionmdashthe extent to which intuitions of acceptability among classroom
learners are biased by the congruity between pedagogical treatment and
classroom input frequency patterns An empirical study previously reported
in Anderson (2002 2007) can shed some light on this question
Methodology
Anderson (2002 2007) investigated the acquisition of various morphosyn-
tactic properties of French noun phrases by administering an acceptability
judgment task in which participants had to evaluate the appropriateness of
sentences featuring among other things the pre- versus post-N position of
adjectives Each of the adjectives selected were lsquohighly variablersquo with respect
to position (cf Table 1) yet crucially do not figure in textbook lists of
adjectives said to change meaning based on position (cf the list in (10))
These include lourd lsquoheavyrsquo violent lsquoviolentrsquo charmant lsquocharmingrsquo precieux
lsquopreciousrsquo epais lsquothickrsquo etrange lsquostrangersquo and delicieux lsquodeliciousrsquo (See
Appendix C available to online subscribers at httpapplijoxfordjournals
org for test sentences) Study participants included English-speaking learners
of French in the second third and fourth years of undergraduate study
(nfrac14 80) who were attending courses at the same institution at which the
observational data were collected a group of advanced learners (nfrac14 20)
consisting of graduate student instructors at that same institution as well as
secondary education teachers and a group of native French speakers (nfrac14 27)
roughly matched in age and education level
Each adjective appeared in test sentences once prenominally and once
postnominally with each test sentence preceded once by a context
298 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
motivating acceptance and once by a context motivating rejection of one or
the other adjective order for a total of four positionndashinterpretation pairings
per adjective These were randomly distributed across two testing
instruments administered a week apart After reading each context and
paired test sentence study participants were directed to check boxes
labeled lsquofinersquo lsquooddrsquo or lsquocannot decidersquo to the prompt This particular sentence
sounds ___ in this context
As an example each of the sentences in (7) featuring the adjective lourd
lsquoheavyrsquo was paired with each of two contexts
(7) a Pierre doit laisser la lourde valise avec son amib Pierre doit laisser la valise lourde a lrsquoaeroport
lsquoPierre has to leave the heavy suitcase with his friendatthe airportrsquo
In the first context motivating a unique noun-referent interpretation
there was one and only one suitcase which happened to be heavy and
which lsquoPierrersquo had to leave behind with his friend In the second context
motivating a multiple noun-referent interpretation there were a number of
suitcases only one of which was heavy and it was this one that lsquoPierrersquo had
to leave behind at the airport (See Appendix D available to online
subscribers at httpapplijoxfordjournalsorg for the complete version of
each context)
When viewed solely in terms of truth-conditional semantics each of the
four positionndashinterpretation pairings is compatible save for one pre-N
position (7a) in a multiple noun-referent context This is because the singular
definite article la and pre-N adjective lourde together presuppose a unique
suitcase in context (which happens to be heavy) This presupposition
does not hold when the adjective is in post-N position where all that is
required is that there be at least one such suitcase in context The testing
instrument then allows us to examine how classroom learners react to
an lsquoexpectedrsquo word order (ie sentences like (7b) in either context) as well
as to an lsquounexpectedrsquo but semantically and pragmatically appropriate
word order (ie sentences like (7a) but only in the unique noun-referent
context)
Results
As the mean acceptance rates from native French speakers in Table 2 show
truth-conditional semantics was not the only factor determining acceptance
rates Two of the pairings received essentially chance-level rates of
acceptance (pre-N position in a multiple noun-referent context at 49
percent and post-N position in a unique context at 52 percent) whereas the
other two pairings showed much clearer above- or below-chance rates (pre-N
position in a unique noun-referent context at 74 percent and post-N
position in a multiple noun-referent context at 37 percent)
BRUCE ANDERSON 299
As was argued in Anderson (2002 2007) results for each of the four
positionndashinterpretation pairings among the native French speakers are
explainable in terms of a disproportionate interaction between semantics and
pragmatic implicature in particular the Gricean maxim of quantity An
utterance should be no more and no less informative than is required for the
current purposes of the exchange (Grice 1975) The positionndashinterpretation
pairing that was lsquotruersquo in terms of semantics and also the most pragmatically
appropriate was accepted at the highest rate (74 percent) the pairing that
was also lsquotruersquo in terms of semantics but was pragmatically the least
appropriate was accepted at the lowest rate (37 percent)8
Do learner group response patterns show the same bias toward pragmatic
implicature in their acceptance rates The comparison of mean acceptance
rates between native French speakers and learner groups in Table 3 shows
that learners performed in a particular (and particularly non-Frenchlike) way
in which the interaction goes in the opposite direction favoring semantics
over pragmatics
Recall that the only positionndashinterpretation pairing that was semantically
incompatible was pre-N position (7a) in a multiple noun-referent context
given the presupposition of uniqueness of the noun referent All learner
groups did indeed accept this pairing at the lowest rate of the four and with
the exception of the second-year learners (at thorn9 percent) did so more
strongly than the native French speakers at mean acceptance rates that were
11ndash15 percent lower)9 However they also accepted at much lower mean
rates (12 to 23 percent) the most felicitous positionndashinterpretation pairing
for the native French speakers pre-N position in a unique noun-referent
context Recall also that post-N position (7b) was semantically compatible
under either interpretation All learner groups accepted these two positionndash
interpretation pairings at much higher rates than the native French speakers
(thorn23 to thorn31 percent) including the least felicitous positionndashinterpretation
pairing for the native French speakersmdashpost-N position in a multiple noun-
referent context This is precisely what one would expect from the
pedagogical treatment these learners are provided and the congruity of
that treatment with classroom input which under-represents actual usage in
Table 2 Native French speaker mean acceptance rates for the uniquenonunique noun-referent distinction
Prenominal (AN)order ()
Postnominal (NA)order ()
Unique noun referent 74 52
Nonunique noun referent 49 37
Source Adapted from Anderson 2002 2007
300 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
text-based corpora and as we have now seen differs from native speaker
intuitions of overall (semantico-pragmatic) acceptability
Although a bias toward or even strict adherence to explicit pedagogical
rules in evaluating such sentences may not seem all that surprising what does
stand out in the data from Anderson (2002 2007) is that these differential
rates in acceptance between native and nonnative speakers are as true for the
advanced group in that studymdashconsisting of graduate student instructors and
second education teachersmdashas they are for the intermediate (second-year)
learner group This would seem to indicate no increased sensitivity to
contextual (pragmatic) factors in adjective position over the course of studying
French even though the advanced group had been exposed to much more
authentic input through texts written for and by native French speakers (ie
the descriptive norm uncovered in our examination of authentic text corpora)
DISCUSSION
What should be clear from the findings presented here is that rules of French
adjective placement that one typically finds in (at least American) textbooks
though not plainly incorrect are not fully descriptively accurate either at
least with respect to written French across several text genres In particular
such rules portray pre-N and post-N positions as mutually exclusive for
almost all adjectives when in fact they are not
Similar discrepancies are often found between rules for and by native
speakers and what those speakers actually do One need look no further than
the French for an example with colloquial speech often at odds with the
prescriptive bon usage and the admonishments of the Academie francaise One
might therefore also expect to find a discrepancy between textbook rules and
what teachers do in the classroom andor what students find to be
lsquoacceptablersquo in terms of word order As we have seen however this is not
the case These rules are faithfully reflected in classroom language use by
instructors to such an extent that when asked to evaluate the acceptability
of lsquoexpectedrsquo and lsquounexpected but nativelikersquo word orders (and one would
presume in production as well) both instructors and students paid much
less attention to problems in pragmatic implicature than did native speakers
a situation which does not appear to change even in the face of counter-
examples occasionally provided in the textbooks they use (see (6) above) and
most certainly in the written texts that they read at later course levels As
pointed out by an anonymous reviewer this situation can be considered a
testament to the power of classroom instruction (instructor input in
particular) if instruction is changed he or she reasons it is likely that
classroom learners will reflect this change
But should instruction be changed If so who along the continuum from
the theorist-researcher to the teacher-practitioner should inform and guide
this change At which level In what way Such questions are at the very
heart of two on-going debates that I re-examine before proposing some
BRUCE ANDERSON 301
answers the pedagogical versus descriptive norm debate (eg Owen 1993
Widdowson 2000 and the collection of articles in Gass et al 2002) and
the interrelated native-speaker-usage-as-goal debate in foreign language
pedagogy (eg the collection of articles in Blyth 2003)
On the whole the approach to instruction on French adjective
position outlined earlier appears advantageous in its simplicity providing
easily remembered rules of thumb that learners can apply in language
production That this approach to instruction does not as we have seen
straightforwardly follow the descriptive norm is a situation that linguists
utilizing a corpus-based methodology believe can and should be remedied
Conrad (2000) for instance sees in corpus-based research an immediate way
in which linguistics can inform language pedagogy freeing language teachers
from lsquohav[ing] to rely on judgments of grammatical accuracy or on native
speakersrsquo intuition about what sounds naturalrsquo (2000 555) Although the
subsequent study by Biber and Reppen (2002) uses a disclaimer to the effect
that frequency should not be the sole factor in material design they
nevertheless suggest that lsquoa selective revision of pedagogy to reflect actual
use as shown by frequency studies could result in radical changes that
facilitate the learning process for studentsrsquo (2002 199) Owen (1993)
however provides examples of how a total reliance on a corpus does not
necessarily yield better observation and more importantly questions
whether better observation automatically equates to better explanation
(1993 168) (but see Francis and Sinclair 1994 for rebuttal) For his part
Widdowson (2000) objects to the kind of equation just described as a
case of linguistics applied a mistaken belief that what is textually attested
lsquouniquely represents real language and that this reality should define the
foreign language subjectrsquo (2000 8) (but see Stubbs 2001 for rebuttal) What
is needed according to Widdowson is not simply linguistics applied
to pedagogy but a true role for applied linguistics an approach to usage that
also recognizes native speakersrsquo awareness of and opinions about such usage
The need for a truly applied linguistics perspective on native speaker
usage (including frequency patterns) is embodied in a pedagogical norm
originally conceptualized in Valdman (1989) and elaborated upon in
Bardovi-Harlig and Gass (2002) as
[A] combination of language systems and forms selected by linguistsand pedagogues to serve as the immediate language target ortargets that learners seek to acquire during their language studyIn other words pedagogical norms represent a mid-point or seriesof mid-points for learners as they progress toward acquiringnative language norms [thus constituting] a form of languagethat is acceptable to native speakers but easier to learn than the fullnative language system (p 3)
The textbook word-order rules for French adjective position reviewed
earlier would first appear to constitute an appropriate pedagogical norm
302 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
under Bardovi-Harlig and Gassrsquos (2002) definition because the learner is able
to produce a subset of the totality of possible utterances deemed acceptable to
native speakers which may turn out to be even more the case in spoken than
in written French (though the difference in mediums remains to be studied)
The learner data presented here from Anderson (2002 2007) however
provides one piece of evidence that textbook rules and classroom input do not
constitute a sufficient pedagogical norm as learners have stopped in their
progress toward acquiring native speaker (ie descriptively accurate) norms
As the articles in Blyth (2003) attest however it is highly debatable
whether the lsquoNative Speakerrsquo (as an idealized abstraction) should be the
appropriate model for the contemporary learner For Kramsch (2002 2003)
too much of a focus on an authentic native speaker norm reduces
foreign language (FL) study to a constant lsquoapproximation to and conformity
with stereotypical native speakers thus often silencing other FL-specific
forms of language potentialrsquo (2002 60) Language learning in instructional
settings on my understanding of Kramschrsquos (2003) argument should not be
viewed solely in a teleological sensemdashas language acquisition directed
toward a fixed nativelike end-pointmdashbut rather as language appropriation
by learners in the sense that they are constructing a linguistic and even
cultural identity lsquoin the interstices of national languages and on the
margins of monolingual speakersrsquo territoriesrsquo (2003 260) The political nature
of the debate as highlighted by Kramsch becomes all the more apparent
when one additionally considers the status of the foreign language on
the world stage (eg French as an international language in retreat)
and how deeply the normative attitudes of its speakers run (eg Francersquos
three-century-long history of acceptance and maintenance of the standard
(Lodge 1993))
The foregoing synopsis of the debates surrounding native speaker usage as
both the goal of foreign language instruction and the extent to which
pedagogical norms need apply has highlighted the real world complexities
involved in what would otherwise seem to be a simple equation Revise
pedagogy to reflect actual use as shown by frequency studies and this
will result in a more targetlike performance on the part of learners So we
return then to the questions posed earlier (1) Should instruction be
changed (2) If so who along the continuum from the theorist-researcher to
the teacher-practitioner should inform and guide this change (3) At which
level (4) In what ways
With respect to our first question because one can empirically
demonstrate that the pedagogical norm for French adjective position
prevents learners from eventually acquiring native speaker (ie descriptively
accurate) norms instruction should be changed so thatmdashirrespective of
whether the attainment of such a norm is necessarily desirablemdashlearners
have both a choice in selecting one grammatical or lexical form over the other
and an awareness of the meaning potential of each choice (following Kramsch
2002 71)
BRUCE ANDERSON 303
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
Nous list in (4) (eg bon premier dernier gros) but gaining others (eg long
lsquolongrsquo court lsquoshortrsquo and autre lsquootherrsquo)
Bragger and Rice (1996) then present the second exception to the basic
post-N rule wherein students are told that lsquoa number of adjectives change
meaning depending on whether theyrsquore placed after or before a noun
In general an adjective placed after the noun retains its basic concrete
meaning The same adjective placed before the noun is used in an abstract or
figurative mannerrsquo (1996 34 emphasis in original) The seven adjectives that
apparently fall into this class are listed in (5) The first English gloss captures
the meaning derived from pre-N position the second gloss that derived from
post-N position Note that Bragger and Rice put the adjective grand in both
lists of exceptions In other words grand is apparently exceptionally pre-N
and exceptionally variable at the same time2
(5) ADJECTIVE PRE-N POST-Na ancien lsquoformerrsquo vs lsquoold ancientrsquob cher lsquodear well-lovedrsquo vs lsquoexpensiversquoc dernier lsquolast latest (in a series)rsquo vs lsquolast (before this one)rsquod grand lsquogreatrsquo vs lsquotall largersquoe pauvre lsquopoor (unfortunate)rsquo vs lsquopoor (not rich)rsquof prochain lsquonext (in a series)rsquo vs lsquonext (after this one)rsquog propre lsquoownrsquo vs lsquocleanrsquo
To summarize the pedagogical treatment of French adjective position
commonly found in university-level American textbooks presents post-N
position as the most frequent and typical (frac14unmarked) adjective order
the less frequent and exceptional (frac14marked) pre-N order is either required
when the adjective is a member of a seemingly fixed lexical class such as
the list in (4) or licensed by a particular interpretation that differs from that
obtained in post-N position such as the list in (5) This difference in
interpretation is often though not always captured through distinct English
glosses
Methodology
In order to determine the extent to which the pedagogical treatment just
summarized is congruent with authentic French texts three corpora were
consulted Wilmetrsquos (1980) corpus of literary texts (approximately 10 million
words 3835 adjectives) Forsgrenrsquos (1978) corpus of newspaper texts from Le
Monde and LrsquoExpress (1 million words 829 adjectives) and Larssonrsquos (1993)
corpus of travel catalogues travel guides and other non-literary prose (15
million words 113 adjectives)3 Using Wilmetrsquos corpus as the point of
comparison given that it contains the highest number of discrete lexemes
(3835 adjectives) an arbitrary cut-off point of 25 attestations or more was
used creating a list of 205 adjectives4 The total number of attestations pre-N
attestations and post-N attestations for each of the 205 adjectives across
BRUCE ANDERSON 291
the three corpora was then determined by adding Forsgrenrsquos and Larssonrsquos
data to that found in Wilmet if available
Results
The 205 most frequently attested adjectives in French texts appear to fall
along a continuum at one end of which we find 68 adjectives (or 332
percent) appearing either only in post-N position (nfrac1434) or typically in that
position (nfrac14 34) and at the other end of which we find 15 adjectives (or 73
percent) typically appearing in pre-N position as shown in Table 1 Note that
there are no adjectives appearing only in pre-N position The remaining 122
adjectives (or 595 percent) are regularly attested in both positions with
some having what we might call an affinity for post-N position (nfrac1441) or
pre-N position (nfrac14 29) but others being highly variable (nfrac1452)
Reminiscent of Holmesrsquo (1988) critique of the simplicity versus naturalness
of textbook rules such data raise some problems for the seemingly
straightforward fixed-order rules of adjective position exemplified in (1)ndash(3)
and the associated lists in (4) and (5) A textbook rule to the effect that
adjectives typically appear after the noun in Frenchmdasha word order that
clearly contrasts with what one finds in Englishmdashis only partly descriptively
Table 1 Frequency groupings of adjective position in written French texts
Frequency grouping Type count(nfrac14 205)
Percent oftotal
Examples
Exclusively pre-N (0post-N position)
0 0 mdash
Typically pre-N (below10 post-N position)
15 75 bon lsquogoodrsquo fameuxlsquo(in)famousrsquo petit lsquosmallrsquo
Variable but with anaffinity for pre-Nposition (10ndash39post-N position)
29 14 charmant lsquocharmingrsquo etrangelsquostrangersquo miserable lsquomiserablersquo
Highly variable (40ndash59post-N position)
52 255 delicieux lsquodeliciousrsquo joyeuxlsquojoyousrsquo puissant lsquopowerfulrsquo
Variable but with anaffinity for post-Nposition (60ndash90post-N position)
41 20 amer lsquobitterrsquo tiede lsquolukewarmrsquoserieux lsquoseriousrsquo
Typically post-N (above90 post-N)
34 17 blanc lsquowhitersquo inconnulsquounknownrsquo sauvage lsquowildrsquo
Exclusively post-N(100 post-Nposition)
34 17 allemand lsquoGermanrsquo dorelsquogoldenrsquo quotidien lsquodailyrsquo
292 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
accurate It is accurate in the sense that the ends of the continuum are
not equivalent in their percentages There are many more adjectives
appearing either only or typically in post-N position than is the case for pre-N
position It is inaccurate in the sense that it ignores the distributional
behavior of 595 percent of the most frequently used adjectives which can
appear in both positions French is not as the rule implies the mirror
opposite of English but instead is more accurately viewed as an lsquoANArsquo
(adjectivendashnounndashadjective) language as opposed to both lsquosystematic ANrsquo
languages like English and German and lsquosystematic NArsquo languages like
Indonesian and Thai (Cinque 1994 99ndash101)
The data also lead us to question the criteria textbooks use for including
certain adjectives in the two lists in (4) and (5) For the so-called prenominal
adjectives a frequency of 90 percent or greater in that position would
establish that adjectives such as excellent fameux lsquo(in)famousrsquo moindre
lsquoslightestrsquo and vrai lsquorealtruersquo need to be added That list itself implies a fixed
lexical class to be memorized though in reality it does nothing more
than provide a handful of frequently attested adjectives (nfrac14 15 [75 percent])
that typically appear just in that position Many more adjectives can
appear in pre-N position as well For the so-called change-of-meaning
adjectives the data show that the list in (5) falls woefully short in terms of
representing those adjectives that actually do regularly appear in both
positions (nfrac14 52)
Even if it were to be established that such rules are accurate with respect
to colloquial French as spoken between native speakers and thus appropriate
for oral proficiency-oriented classrooms such rules obviously fail to capture
the state of affairs in written texts particularly literature to which learners at
later course levels are exposed Yet even students at earlier course levels
occasionally find counter-examples to what they are taught The four cases of
lsquounexpectedrsquo pre-N position in (6) come from the cultural readings of two
first-year and two second-year US textbooks respectively
(6) a La France a un excellent systeme de transports publics(Magnan et al 2002 223)lsquoFrance has an excellent system of public transportrsquo
b Cela explique le grand nombre de magnifiques chateauxdans la region (Valdman and Pons 1997 395)lsquoThis explains the great number of magnificent chateaux inthe regionrsquo
c [U]ne veritable explosion de la technologie (St Onge etal 2003 71) lsquoa veritable explosion of technologyrsquo
d Lrsquohistoire est une douloureuse suite de tragedieshumaines (Oates and Dubois 2003 208)lsquoIts history is a painful series of human tragediesrsquo
One might wish to argue that the examples provided in (6) demonstrate
that inputmdashat least classroom-related inputmdashis not as contradictory to
descriptive norms as argued in this section Examples of the type in (6)
BRUCE ANDERSON 293
however are only occasionally found in textbook reading materials they are
included here to demonstrate that even when making a concerted effort to
write level-appropriate cultural readings textbook authors sometimes break
the very grammar rules they present elsewhere in the textbook suggesting a
true disparity between rules and lsquonaturalisticrsquo usage5 (I return to the issue of
descriptively accurate statements about French adjective position in the
discussion section)
FREQUENCY OF ADJECTIVE USE AND POSITIONIN CLASSROOM INPUT
Methodology
In order to get a general sense of how well the classroom setting as primary
input source reflects either textbook rules or the corpora-based descriptive
norm just summarized frequency counts of adjective type (ie discrete
lexemes) and their position in classroom input were collected by the
researcher from 60 hours of in-class observations of university-level French
courses at a large Midwestern American university Ten hours were spent in
each of six courses spanning the curriculummdashfrom the four semesters
comprising the first- and second-year language skills curriculum (referred to
as Levels 1a 1b 2a and 2b below) to a third-year conversation course
(Level 3) and a fourth-year lecture course on French civilization (Level 4)
Each of these courses was taught almost exclusively in French by a mix of
native and nonnative instructors none of whom were aware of the purpose
of this observation period Class observations were conducted on days when
adjective position was not the focus of explicit instruction as this would have
more than likely biased the frequency counts
The first- and second-year courses consisted of explicit grammar
instruction drills and communicative activities normally involving students
in pair or group work The conversation and civilization courses involved
no explicit grammar instruction the former involved mostly teacherndash
student and studentndashstudent oral communication based on previous course
readings and video segments whereas the latter was almost exclusively
devoted to instructor-delivered lectures Error correction was minimal in all
courses and no case of error correction involving adjective position was
witnessed during the observation period (despite the fact that some errors
were made)
Oral input on adjective position in all courses came from one of three
sources the instructor audio aides (such as cassette and videotape) and
the students themselves Written input in the first- and second-year
courses came from two sources visual aids (such as handouts transparencies
writing on the blackboard) and passages from the textbook that were read
silently or aloud in class At Level 3 written materials included authentic
newspaper and magazine articles in addition to visual aides at Level 4
294 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
a number of works of French literature were read outside of class Note
that written input in these latter two course levels is qualitatively the same
as that used in the corpora compiled by Forsgren (1978) Wilmet (1980) and
Larsson (1993)
At issue during this period of investigation was classroom input that could
be seen andor heard by all students in the classroom (as well as the
researcher) Input from students involved in pair and group work was
therefore excluded from analysis Material to be read outside of class was
also excluded as there was no guarantee that all students had actually
read such material Although the totality of inputmdashincluding sources of input
the learner might have voluntarily sought out outside of classmdashcould
obviously never be exhaustively tallied the exclusion of certain forms of
input was first and foremost a matter of logistics (ie not having the
appropriate funds equipment facilities and time to sound record then
transcribe all classroom input from every participant in every situation of
use) Despite this drawback I see no reason to suspect that student-delivered
input to other students during pair- or group-work would be qualitatively
different (at least in terms of adjective position) from the input delivered by
students that was audible to the entire class this latter source of input
already constitutes 10ndash30 percent of all sources of input tallied as part of the
observation period
Each instance of adjective use during each class session was recorded with
pen and paper by the researcher A coding scheme was devised whereby an
11 17-inch sheet of paper was divided into quadrants labeled according to
the source of the input lsquoinstructorrsquo lsquostudentrsquo lsquoaudiorsquo and lsquowrittenrsquo Each
quadrant was then subdivided according to the position of adjectival
modification lsquopre-Nrsquo lsquopost-Nrsquo lsquopredicatersquo and lsquootherrsquo (to be described
below) Input of the form une bonne idee lsquoa good idearsquo by the instructor was
therefore recorded as an instance of the adjective bonne in the instructor
quadrant pre-N sub-quadrant Hatch marks next to the adjective bonne were
used for any further exemplars of this adjective (including its masculine form
bon) in this position from this source I acknowledge that mere human error
caused by fatigue etc may have resulted in tallies that are less accurate than
if each session were recorded on tape Though the results presented below
should be considered with some caution then occasional errors could not
have reached a significant enough level to detract from the highly salient
patterns of usage that we will find
The resulting sixty sheets each representing one 50-minute class session
were then tallied in order to determine (a) the relative contribution of the
four sources of input (b) the average token (exemplar) count of adjectives
per 50-minute class period (c) the ratio of pre-N to post-Npredicate tokens
(d) the type count (number of discrete lexemes) entering into pre-N and
post-N positions and (e) the frequency of variation in positionmdashthat is the
extent to which a given adjective appeared at least once in both pre-N and
post-N position during the ten hours of observation
BRUCE ANDERSON 295
Results
The instructor not surprisingly was the primary provider of input in the
classroom (See Figure 1 available to online subscribers at http
applijoxfordjournalsorg) This was especially the case in the first-year
courses (Levels 1a 1b) the second- and third-year courses (Levels 2a 2b 3)
showed a greater proportion of input from students and from audio aides
The instructor was by far the predominant source of input in the fourth-year
course (Level 4) given that it involved instructor-delivered lectures on
French civilization with minimal class participation on the part of students
Concerning the average number of times an adjective occurred in the
input during any one 50-minute class period we find a general increase with
level from 42 tokens at Level 1a to 156 tokens per class at Level 4 (See
Figure 2 available to online subscribers at httpapplijoxfordjournalsorg)
Students therefore heard an adjective used anywhere from one to three
times per minute on average
Aggregate data on the frequency with which adjectives appeared in pre-N
post-N and predicate positions or in other constructions (eg used in
isolation in elliptical constructions such as la charmante lsquothe charming [one]rsquo
and following indefinite pronouns as in quelqursquoun de charmant lsquosomeone
charmingrsquo) demonstrate that post-N and predicate positions were generally
equally frequent and together constitute fully 75 percent of the ratio whereas
pre-N position never surpassed 27 percent across course levels Other
constructions were minimal (See Figure 3 available to online subscribers at
httpapplijoxfordjournalsorg)
This roughly 75 percentndash25 percent ratio was then reanalyzed in order to
consider the number of distinct lexemes (lsquotypesrsquo) that make up the
percentages We find for example that only 15 lexemes make up the 14
percent of pre-N adjective use at Level 1a This number gradually increased
with course level but only reached a total of 41 by Level 4 Post-N position
on the other hand admitted a much larger variety of adjectives in classroom
input from 71 at Level 1a to 323 by Level 4 (See Figure 4 available to online
subscribers at httpapplijoxfordjournalsorg) Students therefore got
relatively little input on pre-N position and what input they did get was
made up of a relatively small class of adjectives being used over and over
again and which already figure in the textbook-determined list of pre-N
adjectives in (4) (See Appendix A available to online subscribers at http
applijoxfordjournalsorg for a list of adjectives appearing in pre-N position
by course level)
Finally we find that very few adjectives varied in position during the 10
hours of in-class observation There was no variation attested at Level 1a
and only four to seven adjectives varied in position from Level 1b through to
Level 3 Moreover some of the input provided to learners in these courses
in the form of other studentsrsquo output was nonnativelike6 At the most
advanced undergraduate level learners got somewhat more input involving
296 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
variation in position where 20 adjectives did so (See Figure 5 available to
online subscribers at httpapplijoxfordjournalsorg) Again most (but not
all) of these adjectives already figure into the textbook-determined list of
variable adjectives in (5) (See Appendix B available to online subscribers at
httpapplijoxfordjournalsorg for a list of adjectives varying in position by
course level)
To summarize the observational data just presented the ambient language
of the classroom clearly mimics the pedagogical treatment that American
university-level classroom learners receive from textbooks Although
adjectives are in general a very frequent part of the input the rate of
pre-N position (at roughly 25 percent) is essentially static across levels
pointing to the fact that classroom input across levels is in no way
contradictory to what learners are initially taught Post-N position is basic
(unmarked) and pre-N position is exceptional (marked) The adjectives that
do appear in pre-N position in classroom input are in most cases those that
learners already expect to appear in that position (cf the list in (4) and
Appendix A) The few adjectives that appear in both positions are again in
most cases those that learners expect to do so (cf the list in (5) and
Appendix B)
By the end of the first-year level then both instruction and classroom
input would lead learners to expect that adjectives will appear in post-N
position with the notable exception of the class of adjectives in (4) (those in
(5) having not yet been introduced) By the second-year level it could be
argued that instruction provides examples in the form of so-called change of
meaning adjectives that variation in position is to be associated with
variation in interpretation (and perhaps by implication always to
interpretation) However learnersrsquo explicit knowledge of interpretative
differences as exemplified by the adjectives in (5) typically consists of
arbitrary and unsystematic English glosses By arbitrary I mean that
instruction never provides an adequate explanation for why the pre-N
position of cher should correspond to beloved or for its post-N position to
correspond to expensive7 Though telling students following Bragger and Rice
(1996) that the meaning in post-N position is lsquoconcretersquo and the meaning in
pre-N position is lsquofigurativersquo might (somehow) help it does not account for
other adjectives in (5) like prochain lsquonextrsquo This latter adjective also
demonstrates the fact that English glosses are unsystematic in that prochain
translates as lsquonextrsquo in either position
Before considering our final research question I note that no research on
adjective position in native French speaker discourse is available to date
An anonymous Applied Linguistics reviewer questions the validity of compar-
ing primarily spoken classroom discoursemdashor any other kind of spoken
discoursemdashto written texts without first demonstrating that adjective position
is insensitive to this difference in medium While recognizing the potentially
important distinction in medium and the need for further research in this
area the validity of comparing classroom input to written texts in the present
BRUCE ANDERSON 297
study lies in the fact that such a comparison roughly corresponds to the
curriculum of typical university-level courses an emphasis on listening
comprehension and oral communication in the first- and second-year
language courses leading to an emphasis on reading comprehension
(typically literature-based) and proficiency in writing in the third- and
fourth-year courses Were it to turn out that the degree of variation in
adjective position found in written texts is restricted to just that medium
revisions to pedagogy would necessarily be restricted to writing-intensive
(composition creative composition stylistics) courses at the third- and
fourth-year levels (I return to this issue in the discussion section)
LEARNER INTUITIONS OF ACCEPTABILITY
Thus far we have seen an incongruity between textbook presentations of
adjective position and authentic text sources across a number of written text
genres (research question 1) and that ambient language use involving
adjective position in the classroom is clearly more congruent with and might
therefore be taken to reinforce the explicit rules found in textbooks
(research question 2) We are now in a position to address our third research
questionmdashthe extent to which intuitions of acceptability among classroom
learners are biased by the congruity between pedagogical treatment and
classroom input frequency patterns An empirical study previously reported
in Anderson (2002 2007) can shed some light on this question
Methodology
Anderson (2002 2007) investigated the acquisition of various morphosyn-
tactic properties of French noun phrases by administering an acceptability
judgment task in which participants had to evaluate the appropriateness of
sentences featuring among other things the pre- versus post-N position of
adjectives Each of the adjectives selected were lsquohighly variablersquo with respect
to position (cf Table 1) yet crucially do not figure in textbook lists of
adjectives said to change meaning based on position (cf the list in (10))
These include lourd lsquoheavyrsquo violent lsquoviolentrsquo charmant lsquocharmingrsquo precieux
lsquopreciousrsquo epais lsquothickrsquo etrange lsquostrangersquo and delicieux lsquodeliciousrsquo (See
Appendix C available to online subscribers at httpapplijoxfordjournals
org for test sentences) Study participants included English-speaking learners
of French in the second third and fourth years of undergraduate study
(nfrac14 80) who were attending courses at the same institution at which the
observational data were collected a group of advanced learners (nfrac14 20)
consisting of graduate student instructors at that same institution as well as
secondary education teachers and a group of native French speakers (nfrac14 27)
roughly matched in age and education level
Each adjective appeared in test sentences once prenominally and once
postnominally with each test sentence preceded once by a context
298 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
motivating acceptance and once by a context motivating rejection of one or
the other adjective order for a total of four positionndashinterpretation pairings
per adjective These were randomly distributed across two testing
instruments administered a week apart After reading each context and
paired test sentence study participants were directed to check boxes
labeled lsquofinersquo lsquooddrsquo or lsquocannot decidersquo to the prompt This particular sentence
sounds ___ in this context
As an example each of the sentences in (7) featuring the adjective lourd
lsquoheavyrsquo was paired with each of two contexts
(7) a Pierre doit laisser la lourde valise avec son amib Pierre doit laisser la valise lourde a lrsquoaeroport
lsquoPierre has to leave the heavy suitcase with his friendatthe airportrsquo
In the first context motivating a unique noun-referent interpretation
there was one and only one suitcase which happened to be heavy and
which lsquoPierrersquo had to leave behind with his friend In the second context
motivating a multiple noun-referent interpretation there were a number of
suitcases only one of which was heavy and it was this one that lsquoPierrersquo had
to leave behind at the airport (See Appendix D available to online
subscribers at httpapplijoxfordjournalsorg for the complete version of
each context)
When viewed solely in terms of truth-conditional semantics each of the
four positionndashinterpretation pairings is compatible save for one pre-N
position (7a) in a multiple noun-referent context This is because the singular
definite article la and pre-N adjective lourde together presuppose a unique
suitcase in context (which happens to be heavy) This presupposition
does not hold when the adjective is in post-N position where all that is
required is that there be at least one such suitcase in context The testing
instrument then allows us to examine how classroom learners react to
an lsquoexpectedrsquo word order (ie sentences like (7b) in either context) as well
as to an lsquounexpectedrsquo but semantically and pragmatically appropriate
word order (ie sentences like (7a) but only in the unique noun-referent
context)
Results
As the mean acceptance rates from native French speakers in Table 2 show
truth-conditional semantics was not the only factor determining acceptance
rates Two of the pairings received essentially chance-level rates of
acceptance (pre-N position in a multiple noun-referent context at 49
percent and post-N position in a unique context at 52 percent) whereas the
other two pairings showed much clearer above- or below-chance rates (pre-N
position in a unique noun-referent context at 74 percent and post-N
position in a multiple noun-referent context at 37 percent)
BRUCE ANDERSON 299
As was argued in Anderson (2002 2007) results for each of the four
positionndashinterpretation pairings among the native French speakers are
explainable in terms of a disproportionate interaction between semantics and
pragmatic implicature in particular the Gricean maxim of quantity An
utterance should be no more and no less informative than is required for the
current purposes of the exchange (Grice 1975) The positionndashinterpretation
pairing that was lsquotruersquo in terms of semantics and also the most pragmatically
appropriate was accepted at the highest rate (74 percent) the pairing that
was also lsquotruersquo in terms of semantics but was pragmatically the least
appropriate was accepted at the lowest rate (37 percent)8
Do learner group response patterns show the same bias toward pragmatic
implicature in their acceptance rates The comparison of mean acceptance
rates between native French speakers and learner groups in Table 3 shows
that learners performed in a particular (and particularly non-Frenchlike) way
in which the interaction goes in the opposite direction favoring semantics
over pragmatics
Recall that the only positionndashinterpretation pairing that was semantically
incompatible was pre-N position (7a) in a multiple noun-referent context
given the presupposition of uniqueness of the noun referent All learner
groups did indeed accept this pairing at the lowest rate of the four and with
the exception of the second-year learners (at thorn9 percent) did so more
strongly than the native French speakers at mean acceptance rates that were
11ndash15 percent lower)9 However they also accepted at much lower mean
rates (12 to 23 percent) the most felicitous positionndashinterpretation pairing
for the native French speakers pre-N position in a unique noun-referent
context Recall also that post-N position (7b) was semantically compatible
under either interpretation All learner groups accepted these two positionndash
interpretation pairings at much higher rates than the native French speakers
(thorn23 to thorn31 percent) including the least felicitous positionndashinterpretation
pairing for the native French speakersmdashpost-N position in a multiple noun-
referent context This is precisely what one would expect from the
pedagogical treatment these learners are provided and the congruity of
that treatment with classroom input which under-represents actual usage in
Table 2 Native French speaker mean acceptance rates for the uniquenonunique noun-referent distinction
Prenominal (AN)order ()
Postnominal (NA)order ()
Unique noun referent 74 52
Nonunique noun referent 49 37
Source Adapted from Anderson 2002 2007
300 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
text-based corpora and as we have now seen differs from native speaker
intuitions of overall (semantico-pragmatic) acceptability
Although a bias toward or even strict adherence to explicit pedagogical
rules in evaluating such sentences may not seem all that surprising what does
stand out in the data from Anderson (2002 2007) is that these differential
rates in acceptance between native and nonnative speakers are as true for the
advanced group in that studymdashconsisting of graduate student instructors and
second education teachersmdashas they are for the intermediate (second-year)
learner group This would seem to indicate no increased sensitivity to
contextual (pragmatic) factors in adjective position over the course of studying
French even though the advanced group had been exposed to much more
authentic input through texts written for and by native French speakers (ie
the descriptive norm uncovered in our examination of authentic text corpora)
DISCUSSION
What should be clear from the findings presented here is that rules of French
adjective placement that one typically finds in (at least American) textbooks
though not plainly incorrect are not fully descriptively accurate either at
least with respect to written French across several text genres In particular
such rules portray pre-N and post-N positions as mutually exclusive for
almost all adjectives when in fact they are not
Similar discrepancies are often found between rules for and by native
speakers and what those speakers actually do One need look no further than
the French for an example with colloquial speech often at odds with the
prescriptive bon usage and the admonishments of the Academie francaise One
might therefore also expect to find a discrepancy between textbook rules and
what teachers do in the classroom andor what students find to be
lsquoacceptablersquo in terms of word order As we have seen however this is not
the case These rules are faithfully reflected in classroom language use by
instructors to such an extent that when asked to evaluate the acceptability
of lsquoexpectedrsquo and lsquounexpected but nativelikersquo word orders (and one would
presume in production as well) both instructors and students paid much
less attention to problems in pragmatic implicature than did native speakers
a situation which does not appear to change even in the face of counter-
examples occasionally provided in the textbooks they use (see (6) above) and
most certainly in the written texts that they read at later course levels As
pointed out by an anonymous reviewer this situation can be considered a
testament to the power of classroom instruction (instructor input in
particular) if instruction is changed he or she reasons it is likely that
classroom learners will reflect this change
But should instruction be changed If so who along the continuum from
the theorist-researcher to the teacher-practitioner should inform and guide
this change At which level In what way Such questions are at the very
heart of two on-going debates that I re-examine before proposing some
BRUCE ANDERSON 301
answers the pedagogical versus descriptive norm debate (eg Owen 1993
Widdowson 2000 and the collection of articles in Gass et al 2002) and
the interrelated native-speaker-usage-as-goal debate in foreign language
pedagogy (eg the collection of articles in Blyth 2003)
On the whole the approach to instruction on French adjective
position outlined earlier appears advantageous in its simplicity providing
easily remembered rules of thumb that learners can apply in language
production That this approach to instruction does not as we have seen
straightforwardly follow the descriptive norm is a situation that linguists
utilizing a corpus-based methodology believe can and should be remedied
Conrad (2000) for instance sees in corpus-based research an immediate way
in which linguistics can inform language pedagogy freeing language teachers
from lsquohav[ing] to rely on judgments of grammatical accuracy or on native
speakersrsquo intuition about what sounds naturalrsquo (2000 555) Although the
subsequent study by Biber and Reppen (2002) uses a disclaimer to the effect
that frequency should not be the sole factor in material design they
nevertheless suggest that lsquoa selective revision of pedagogy to reflect actual
use as shown by frequency studies could result in radical changes that
facilitate the learning process for studentsrsquo (2002 199) Owen (1993)
however provides examples of how a total reliance on a corpus does not
necessarily yield better observation and more importantly questions
whether better observation automatically equates to better explanation
(1993 168) (but see Francis and Sinclair 1994 for rebuttal) For his part
Widdowson (2000) objects to the kind of equation just described as a
case of linguistics applied a mistaken belief that what is textually attested
lsquouniquely represents real language and that this reality should define the
foreign language subjectrsquo (2000 8) (but see Stubbs 2001 for rebuttal) What
is needed according to Widdowson is not simply linguistics applied
to pedagogy but a true role for applied linguistics an approach to usage that
also recognizes native speakersrsquo awareness of and opinions about such usage
The need for a truly applied linguistics perspective on native speaker
usage (including frequency patterns) is embodied in a pedagogical norm
originally conceptualized in Valdman (1989) and elaborated upon in
Bardovi-Harlig and Gass (2002) as
[A] combination of language systems and forms selected by linguistsand pedagogues to serve as the immediate language target ortargets that learners seek to acquire during their language studyIn other words pedagogical norms represent a mid-point or seriesof mid-points for learners as they progress toward acquiringnative language norms [thus constituting] a form of languagethat is acceptable to native speakers but easier to learn than the fullnative language system (p 3)
The textbook word-order rules for French adjective position reviewed
earlier would first appear to constitute an appropriate pedagogical norm
302 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
under Bardovi-Harlig and Gassrsquos (2002) definition because the learner is able
to produce a subset of the totality of possible utterances deemed acceptable to
native speakers which may turn out to be even more the case in spoken than
in written French (though the difference in mediums remains to be studied)
The learner data presented here from Anderson (2002 2007) however
provides one piece of evidence that textbook rules and classroom input do not
constitute a sufficient pedagogical norm as learners have stopped in their
progress toward acquiring native speaker (ie descriptively accurate) norms
As the articles in Blyth (2003) attest however it is highly debatable
whether the lsquoNative Speakerrsquo (as an idealized abstraction) should be the
appropriate model for the contemporary learner For Kramsch (2002 2003)
too much of a focus on an authentic native speaker norm reduces
foreign language (FL) study to a constant lsquoapproximation to and conformity
with stereotypical native speakers thus often silencing other FL-specific
forms of language potentialrsquo (2002 60) Language learning in instructional
settings on my understanding of Kramschrsquos (2003) argument should not be
viewed solely in a teleological sensemdashas language acquisition directed
toward a fixed nativelike end-pointmdashbut rather as language appropriation
by learners in the sense that they are constructing a linguistic and even
cultural identity lsquoin the interstices of national languages and on the
margins of monolingual speakersrsquo territoriesrsquo (2003 260) The political nature
of the debate as highlighted by Kramsch becomes all the more apparent
when one additionally considers the status of the foreign language on
the world stage (eg French as an international language in retreat)
and how deeply the normative attitudes of its speakers run (eg Francersquos
three-century-long history of acceptance and maintenance of the standard
(Lodge 1993))
The foregoing synopsis of the debates surrounding native speaker usage as
both the goal of foreign language instruction and the extent to which
pedagogical norms need apply has highlighted the real world complexities
involved in what would otherwise seem to be a simple equation Revise
pedagogy to reflect actual use as shown by frequency studies and this
will result in a more targetlike performance on the part of learners So we
return then to the questions posed earlier (1) Should instruction be
changed (2) If so who along the continuum from the theorist-researcher to
the teacher-practitioner should inform and guide this change (3) At which
level (4) In what ways
With respect to our first question because one can empirically
demonstrate that the pedagogical norm for French adjective position
prevents learners from eventually acquiring native speaker (ie descriptively
accurate) norms instruction should be changed so thatmdashirrespective of
whether the attainment of such a norm is necessarily desirablemdashlearners
have both a choice in selecting one grammatical or lexical form over the other
and an awareness of the meaning potential of each choice (following Kramsch
2002 71)
BRUCE ANDERSON 303
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
the three corpora was then determined by adding Forsgrenrsquos and Larssonrsquos
data to that found in Wilmet if available
Results
The 205 most frequently attested adjectives in French texts appear to fall
along a continuum at one end of which we find 68 adjectives (or 332
percent) appearing either only in post-N position (nfrac1434) or typically in that
position (nfrac14 34) and at the other end of which we find 15 adjectives (or 73
percent) typically appearing in pre-N position as shown in Table 1 Note that
there are no adjectives appearing only in pre-N position The remaining 122
adjectives (or 595 percent) are regularly attested in both positions with
some having what we might call an affinity for post-N position (nfrac1441) or
pre-N position (nfrac14 29) but others being highly variable (nfrac1452)
Reminiscent of Holmesrsquo (1988) critique of the simplicity versus naturalness
of textbook rules such data raise some problems for the seemingly
straightforward fixed-order rules of adjective position exemplified in (1)ndash(3)
and the associated lists in (4) and (5) A textbook rule to the effect that
adjectives typically appear after the noun in Frenchmdasha word order that
clearly contrasts with what one finds in Englishmdashis only partly descriptively
Table 1 Frequency groupings of adjective position in written French texts
Frequency grouping Type count(nfrac14 205)
Percent oftotal
Examples
Exclusively pre-N (0post-N position)
0 0 mdash
Typically pre-N (below10 post-N position)
15 75 bon lsquogoodrsquo fameuxlsquo(in)famousrsquo petit lsquosmallrsquo
Variable but with anaffinity for pre-Nposition (10ndash39post-N position)
29 14 charmant lsquocharmingrsquo etrangelsquostrangersquo miserable lsquomiserablersquo
Highly variable (40ndash59post-N position)
52 255 delicieux lsquodeliciousrsquo joyeuxlsquojoyousrsquo puissant lsquopowerfulrsquo
Variable but with anaffinity for post-Nposition (60ndash90post-N position)
41 20 amer lsquobitterrsquo tiede lsquolukewarmrsquoserieux lsquoseriousrsquo
Typically post-N (above90 post-N)
34 17 blanc lsquowhitersquo inconnulsquounknownrsquo sauvage lsquowildrsquo
Exclusively post-N(100 post-Nposition)
34 17 allemand lsquoGermanrsquo dorelsquogoldenrsquo quotidien lsquodailyrsquo
292 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
accurate It is accurate in the sense that the ends of the continuum are
not equivalent in their percentages There are many more adjectives
appearing either only or typically in post-N position than is the case for pre-N
position It is inaccurate in the sense that it ignores the distributional
behavior of 595 percent of the most frequently used adjectives which can
appear in both positions French is not as the rule implies the mirror
opposite of English but instead is more accurately viewed as an lsquoANArsquo
(adjectivendashnounndashadjective) language as opposed to both lsquosystematic ANrsquo
languages like English and German and lsquosystematic NArsquo languages like
Indonesian and Thai (Cinque 1994 99ndash101)
The data also lead us to question the criteria textbooks use for including
certain adjectives in the two lists in (4) and (5) For the so-called prenominal
adjectives a frequency of 90 percent or greater in that position would
establish that adjectives such as excellent fameux lsquo(in)famousrsquo moindre
lsquoslightestrsquo and vrai lsquorealtruersquo need to be added That list itself implies a fixed
lexical class to be memorized though in reality it does nothing more
than provide a handful of frequently attested adjectives (nfrac14 15 [75 percent])
that typically appear just in that position Many more adjectives can
appear in pre-N position as well For the so-called change-of-meaning
adjectives the data show that the list in (5) falls woefully short in terms of
representing those adjectives that actually do regularly appear in both
positions (nfrac14 52)
Even if it were to be established that such rules are accurate with respect
to colloquial French as spoken between native speakers and thus appropriate
for oral proficiency-oriented classrooms such rules obviously fail to capture
the state of affairs in written texts particularly literature to which learners at
later course levels are exposed Yet even students at earlier course levels
occasionally find counter-examples to what they are taught The four cases of
lsquounexpectedrsquo pre-N position in (6) come from the cultural readings of two
first-year and two second-year US textbooks respectively
(6) a La France a un excellent systeme de transports publics(Magnan et al 2002 223)lsquoFrance has an excellent system of public transportrsquo
b Cela explique le grand nombre de magnifiques chateauxdans la region (Valdman and Pons 1997 395)lsquoThis explains the great number of magnificent chateaux inthe regionrsquo
c [U]ne veritable explosion de la technologie (St Onge etal 2003 71) lsquoa veritable explosion of technologyrsquo
d Lrsquohistoire est une douloureuse suite de tragedieshumaines (Oates and Dubois 2003 208)lsquoIts history is a painful series of human tragediesrsquo
One might wish to argue that the examples provided in (6) demonstrate
that inputmdashat least classroom-related inputmdashis not as contradictory to
descriptive norms as argued in this section Examples of the type in (6)
BRUCE ANDERSON 293
however are only occasionally found in textbook reading materials they are
included here to demonstrate that even when making a concerted effort to
write level-appropriate cultural readings textbook authors sometimes break
the very grammar rules they present elsewhere in the textbook suggesting a
true disparity between rules and lsquonaturalisticrsquo usage5 (I return to the issue of
descriptively accurate statements about French adjective position in the
discussion section)
FREQUENCY OF ADJECTIVE USE AND POSITIONIN CLASSROOM INPUT
Methodology
In order to get a general sense of how well the classroom setting as primary
input source reflects either textbook rules or the corpora-based descriptive
norm just summarized frequency counts of adjective type (ie discrete
lexemes) and their position in classroom input were collected by the
researcher from 60 hours of in-class observations of university-level French
courses at a large Midwestern American university Ten hours were spent in
each of six courses spanning the curriculummdashfrom the four semesters
comprising the first- and second-year language skills curriculum (referred to
as Levels 1a 1b 2a and 2b below) to a third-year conversation course
(Level 3) and a fourth-year lecture course on French civilization (Level 4)
Each of these courses was taught almost exclusively in French by a mix of
native and nonnative instructors none of whom were aware of the purpose
of this observation period Class observations were conducted on days when
adjective position was not the focus of explicit instruction as this would have
more than likely biased the frequency counts
The first- and second-year courses consisted of explicit grammar
instruction drills and communicative activities normally involving students
in pair or group work The conversation and civilization courses involved
no explicit grammar instruction the former involved mostly teacherndash
student and studentndashstudent oral communication based on previous course
readings and video segments whereas the latter was almost exclusively
devoted to instructor-delivered lectures Error correction was minimal in all
courses and no case of error correction involving adjective position was
witnessed during the observation period (despite the fact that some errors
were made)
Oral input on adjective position in all courses came from one of three
sources the instructor audio aides (such as cassette and videotape) and
the students themselves Written input in the first- and second-year
courses came from two sources visual aids (such as handouts transparencies
writing on the blackboard) and passages from the textbook that were read
silently or aloud in class At Level 3 written materials included authentic
newspaper and magazine articles in addition to visual aides at Level 4
294 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
a number of works of French literature were read outside of class Note
that written input in these latter two course levels is qualitatively the same
as that used in the corpora compiled by Forsgren (1978) Wilmet (1980) and
Larsson (1993)
At issue during this period of investigation was classroom input that could
be seen andor heard by all students in the classroom (as well as the
researcher) Input from students involved in pair and group work was
therefore excluded from analysis Material to be read outside of class was
also excluded as there was no guarantee that all students had actually
read such material Although the totality of inputmdashincluding sources of input
the learner might have voluntarily sought out outside of classmdashcould
obviously never be exhaustively tallied the exclusion of certain forms of
input was first and foremost a matter of logistics (ie not having the
appropriate funds equipment facilities and time to sound record then
transcribe all classroom input from every participant in every situation of
use) Despite this drawback I see no reason to suspect that student-delivered
input to other students during pair- or group-work would be qualitatively
different (at least in terms of adjective position) from the input delivered by
students that was audible to the entire class this latter source of input
already constitutes 10ndash30 percent of all sources of input tallied as part of the
observation period
Each instance of adjective use during each class session was recorded with
pen and paper by the researcher A coding scheme was devised whereby an
11 17-inch sheet of paper was divided into quadrants labeled according to
the source of the input lsquoinstructorrsquo lsquostudentrsquo lsquoaudiorsquo and lsquowrittenrsquo Each
quadrant was then subdivided according to the position of adjectival
modification lsquopre-Nrsquo lsquopost-Nrsquo lsquopredicatersquo and lsquootherrsquo (to be described
below) Input of the form une bonne idee lsquoa good idearsquo by the instructor was
therefore recorded as an instance of the adjective bonne in the instructor
quadrant pre-N sub-quadrant Hatch marks next to the adjective bonne were
used for any further exemplars of this adjective (including its masculine form
bon) in this position from this source I acknowledge that mere human error
caused by fatigue etc may have resulted in tallies that are less accurate than
if each session were recorded on tape Though the results presented below
should be considered with some caution then occasional errors could not
have reached a significant enough level to detract from the highly salient
patterns of usage that we will find
The resulting sixty sheets each representing one 50-minute class session
were then tallied in order to determine (a) the relative contribution of the
four sources of input (b) the average token (exemplar) count of adjectives
per 50-minute class period (c) the ratio of pre-N to post-Npredicate tokens
(d) the type count (number of discrete lexemes) entering into pre-N and
post-N positions and (e) the frequency of variation in positionmdashthat is the
extent to which a given adjective appeared at least once in both pre-N and
post-N position during the ten hours of observation
BRUCE ANDERSON 295
Results
The instructor not surprisingly was the primary provider of input in the
classroom (See Figure 1 available to online subscribers at http
applijoxfordjournalsorg) This was especially the case in the first-year
courses (Levels 1a 1b) the second- and third-year courses (Levels 2a 2b 3)
showed a greater proportion of input from students and from audio aides
The instructor was by far the predominant source of input in the fourth-year
course (Level 4) given that it involved instructor-delivered lectures on
French civilization with minimal class participation on the part of students
Concerning the average number of times an adjective occurred in the
input during any one 50-minute class period we find a general increase with
level from 42 tokens at Level 1a to 156 tokens per class at Level 4 (See
Figure 2 available to online subscribers at httpapplijoxfordjournalsorg)
Students therefore heard an adjective used anywhere from one to three
times per minute on average
Aggregate data on the frequency with which adjectives appeared in pre-N
post-N and predicate positions or in other constructions (eg used in
isolation in elliptical constructions such as la charmante lsquothe charming [one]rsquo
and following indefinite pronouns as in quelqursquoun de charmant lsquosomeone
charmingrsquo) demonstrate that post-N and predicate positions were generally
equally frequent and together constitute fully 75 percent of the ratio whereas
pre-N position never surpassed 27 percent across course levels Other
constructions were minimal (See Figure 3 available to online subscribers at
httpapplijoxfordjournalsorg)
This roughly 75 percentndash25 percent ratio was then reanalyzed in order to
consider the number of distinct lexemes (lsquotypesrsquo) that make up the
percentages We find for example that only 15 lexemes make up the 14
percent of pre-N adjective use at Level 1a This number gradually increased
with course level but only reached a total of 41 by Level 4 Post-N position
on the other hand admitted a much larger variety of adjectives in classroom
input from 71 at Level 1a to 323 by Level 4 (See Figure 4 available to online
subscribers at httpapplijoxfordjournalsorg) Students therefore got
relatively little input on pre-N position and what input they did get was
made up of a relatively small class of adjectives being used over and over
again and which already figure in the textbook-determined list of pre-N
adjectives in (4) (See Appendix A available to online subscribers at http
applijoxfordjournalsorg for a list of adjectives appearing in pre-N position
by course level)
Finally we find that very few adjectives varied in position during the 10
hours of in-class observation There was no variation attested at Level 1a
and only four to seven adjectives varied in position from Level 1b through to
Level 3 Moreover some of the input provided to learners in these courses
in the form of other studentsrsquo output was nonnativelike6 At the most
advanced undergraduate level learners got somewhat more input involving
296 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
variation in position where 20 adjectives did so (See Figure 5 available to
online subscribers at httpapplijoxfordjournalsorg) Again most (but not
all) of these adjectives already figure into the textbook-determined list of
variable adjectives in (5) (See Appendix B available to online subscribers at
httpapplijoxfordjournalsorg for a list of adjectives varying in position by
course level)
To summarize the observational data just presented the ambient language
of the classroom clearly mimics the pedagogical treatment that American
university-level classroom learners receive from textbooks Although
adjectives are in general a very frequent part of the input the rate of
pre-N position (at roughly 25 percent) is essentially static across levels
pointing to the fact that classroom input across levels is in no way
contradictory to what learners are initially taught Post-N position is basic
(unmarked) and pre-N position is exceptional (marked) The adjectives that
do appear in pre-N position in classroom input are in most cases those that
learners already expect to appear in that position (cf the list in (4) and
Appendix A) The few adjectives that appear in both positions are again in
most cases those that learners expect to do so (cf the list in (5) and
Appendix B)
By the end of the first-year level then both instruction and classroom
input would lead learners to expect that adjectives will appear in post-N
position with the notable exception of the class of adjectives in (4) (those in
(5) having not yet been introduced) By the second-year level it could be
argued that instruction provides examples in the form of so-called change of
meaning adjectives that variation in position is to be associated with
variation in interpretation (and perhaps by implication always to
interpretation) However learnersrsquo explicit knowledge of interpretative
differences as exemplified by the adjectives in (5) typically consists of
arbitrary and unsystematic English glosses By arbitrary I mean that
instruction never provides an adequate explanation for why the pre-N
position of cher should correspond to beloved or for its post-N position to
correspond to expensive7 Though telling students following Bragger and Rice
(1996) that the meaning in post-N position is lsquoconcretersquo and the meaning in
pre-N position is lsquofigurativersquo might (somehow) help it does not account for
other adjectives in (5) like prochain lsquonextrsquo This latter adjective also
demonstrates the fact that English glosses are unsystematic in that prochain
translates as lsquonextrsquo in either position
Before considering our final research question I note that no research on
adjective position in native French speaker discourse is available to date
An anonymous Applied Linguistics reviewer questions the validity of compar-
ing primarily spoken classroom discoursemdashor any other kind of spoken
discoursemdashto written texts without first demonstrating that adjective position
is insensitive to this difference in medium While recognizing the potentially
important distinction in medium and the need for further research in this
area the validity of comparing classroom input to written texts in the present
BRUCE ANDERSON 297
study lies in the fact that such a comparison roughly corresponds to the
curriculum of typical university-level courses an emphasis on listening
comprehension and oral communication in the first- and second-year
language courses leading to an emphasis on reading comprehension
(typically literature-based) and proficiency in writing in the third- and
fourth-year courses Were it to turn out that the degree of variation in
adjective position found in written texts is restricted to just that medium
revisions to pedagogy would necessarily be restricted to writing-intensive
(composition creative composition stylistics) courses at the third- and
fourth-year levels (I return to this issue in the discussion section)
LEARNER INTUITIONS OF ACCEPTABILITY
Thus far we have seen an incongruity between textbook presentations of
adjective position and authentic text sources across a number of written text
genres (research question 1) and that ambient language use involving
adjective position in the classroom is clearly more congruent with and might
therefore be taken to reinforce the explicit rules found in textbooks
(research question 2) We are now in a position to address our third research
questionmdashthe extent to which intuitions of acceptability among classroom
learners are biased by the congruity between pedagogical treatment and
classroom input frequency patterns An empirical study previously reported
in Anderson (2002 2007) can shed some light on this question
Methodology
Anderson (2002 2007) investigated the acquisition of various morphosyn-
tactic properties of French noun phrases by administering an acceptability
judgment task in which participants had to evaluate the appropriateness of
sentences featuring among other things the pre- versus post-N position of
adjectives Each of the adjectives selected were lsquohighly variablersquo with respect
to position (cf Table 1) yet crucially do not figure in textbook lists of
adjectives said to change meaning based on position (cf the list in (10))
These include lourd lsquoheavyrsquo violent lsquoviolentrsquo charmant lsquocharmingrsquo precieux
lsquopreciousrsquo epais lsquothickrsquo etrange lsquostrangersquo and delicieux lsquodeliciousrsquo (See
Appendix C available to online subscribers at httpapplijoxfordjournals
org for test sentences) Study participants included English-speaking learners
of French in the second third and fourth years of undergraduate study
(nfrac14 80) who were attending courses at the same institution at which the
observational data were collected a group of advanced learners (nfrac14 20)
consisting of graduate student instructors at that same institution as well as
secondary education teachers and a group of native French speakers (nfrac14 27)
roughly matched in age and education level
Each adjective appeared in test sentences once prenominally and once
postnominally with each test sentence preceded once by a context
298 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
motivating acceptance and once by a context motivating rejection of one or
the other adjective order for a total of four positionndashinterpretation pairings
per adjective These were randomly distributed across two testing
instruments administered a week apart After reading each context and
paired test sentence study participants were directed to check boxes
labeled lsquofinersquo lsquooddrsquo or lsquocannot decidersquo to the prompt This particular sentence
sounds ___ in this context
As an example each of the sentences in (7) featuring the adjective lourd
lsquoheavyrsquo was paired with each of two contexts
(7) a Pierre doit laisser la lourde valise avec son amib Pierre doit laisser la valise lourde a lrsquoaeroport
lsquoPierre has to leave the heavy suitcase with his friendatthe airportrsquo
In the first context motivating a unique noun-referent interpretation
there was one and only one suitcase which happened to be heavy and
which lsquoPierrersquo had to leave behind with his friend In the second context
motivating a multiple noun-referent interpretation there were a number of
suitcases only one of which was heavy and it was this one that lsquoPierrersquo had
to leave behind at the airport (See Appendix D available to online
subscribers at httpapplijoxfordjournalsorg for the complete version of
each context)
When viewed solely in terms of truth-conditional semantics each of the
four positionndashinterpretation pairings is compatible save for one pre-N
position (7a) in a multiple noun-referent context This is because the singular
definite article la and pre-N adjective lourde together presuppose a unique
suitcase in context (which happens to be heavy) This presupposition
does not hold when the adjective is in post-N position where all that is
required is that there be at least one such suitcase in context The testing
instrument then allows us to examine how classroom learners react to
an lsquoexpectedrsquo word order (ie sentences like (7b) in either context) as well
as to an lsquounexpectedrsquo but semantically and pragmatically appropriate
word order (ie sentences like (7a) but only in the unique noun-referent
context)
Results
As the mean acceptance rates from native French speakers in Table 2 show
truth-conditional semantics was not the only factor determining acceptance
rates Two of the pairings received essentially chance-level rates of
acceptance (pre-N position in a multiple noun-referent context at 49
percent and post-N position in a unique context at 52 percent) whereas the
other two pairings showed much clearer above- or below-chance rates (pre-N
position in a unique noun-referent context at 74 percent and post-N
position in a multiple noun-referent context at 37 percent)
BRUCE ANDERSON 299
As was argued in Anderson (2002 2007) results for each of the four
positionndashinterpretation pairings among the native French speakers are
explainable in terms of a disproportionate interaction between semantics and
pragmatic implicature in particular the Gricean maxim of quantity An
utterance should be no more and no less informative than is required for the
current purposes of the exchange (Grice 1975) The positionndashinterpretation
pairing that was lsquotruersquo in terms of semantics and also the most pragmatically
appropriate was accepted at the highest rate (74 percent) the pairing that
was also lsquotruersquo in terms of semantics but was pragmatically the least
appropriate was accepted at the lowest rate (37 percent)8
Do learner group response patterns show the same bias toward pragmatic
implicature in their acceptance rates The comparison of mean acceptance
rates between native French speakers and learner groups in Table 3 shows
that learners performed in a particular (and particularly non-Frenchlike) way
in which the interaction goes in the opposite direction favoring semantics
over pragmatics
Recall that the only positionndashinterpretation pairing that was semantically
incompatible was pre-N position (7a) in a multiple noun-referent context
given the presupposition of uniqueness of the noun referent All learner
groups did indeed accept this pairing at the lowest rate of the four and with
the exception of the second-year learners (at thorn9 percent) did so more
strongly than the native French speakers at mean acceptance rates that were
11ndash15 percent lower)9 However they also accepted at much lower mean
rates (12 to 23 percent) the most felicitous positionndashinterpretation pairing
for the native French speakers pre-N position in a unique noun-referent
context Recall also that post-N position (7b) was semantically compatible
under either interpretation All learner groups accepted these two positionndash
interpretation pairings at much higher rates than the native French speakers
(thorn23 to thorn31 percent) including the least felicitous positionndashinterpretation
pairing for the native French speakersmdashpost-N position in a multiple noun-
referent context This is precisely what one would expect from the
pedagogical treatment these learners are provided and the congruity of
that treatment with classroom input which under-represents actual usage in
Table 2 Native French speaker mean acceptance rates for the uniquenonunique noun-referent distinction
Prenominal (AN)order ()
Postnominal (NA)order ()
Unique noun referent 74 52
Nonunique noun referent 49 37
Source Adapted from Anderson 2002 2007
300 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
text-based corpora and as we have now seen differs from native speaker
intuitions of overall (semantico-pragmatic) acceptability
Although a bias toward or even strict adherence to explicit pedagogical
rules in evaluating such sentences may not seem all that surprising what does
stand out in the data from Anderson (2002 2007) is that these differential
rates in acceptance between native and nonnative speakers are as true for the
advanced group in that studymdashconsisting of graduate student instructors and
second education teachersmdashas they are for the intermediate (second-year)
learner group This would seem to indicate no increased sensitivity to
contextual (pragmatic) factors in adjective position over the course of studying
French even though the advanced group had been exposed to much more
authentic input through texts written for and by native French speakers (ie
the descriptive norm uncovered in our examination of authentic text corpora)
DISCUSSION
What should be clear from the findings presented here is that rules of French
adjective placement that one typically finds in (at least American) textbooks
though not plainly incorrect are not fully descriptively accurate either at
least with respect to written French across several text genres In particular
such rules portray pre-N and post-N positions as mutually exclusive for
almost all adjectives when in fact they are not
Similar discrepancies are often found between rules for and by native
speakers and what those speakers actually do One need look no further than
the French for an example with colloquial speech often at odds with the
prescriptive bon usage and the admonishments of the Academie francaise One
might therefore also expect to find a discrepancy between textbook rules and
what teachers do in the classroom andor what students find to be
lsquoacceptablersquo in terms of word order As we have seen however this is not
the case These rules are faithfully reflected in classroom language use by
instructors to such an extent that when asked to evaluate the acceptability
of lsquoexpectedrsquo and lsquounexpected but nativelikersquo word orders (and one would
presume in production as well) both instructors and students paid much
less attention to problems in pragmatic implicature than did native speakers
a situation which does not appear to change even in the face of counter-
examples occasionally provided in the textbooks they use (see (6) above) and
most certainly in the written texts that they read at later course levels As
pointed out by an anonymous reviewer this situation can be considered a
testament to the power of classroom instruction (instructor input in
particular) if instruction is changed he or she reasons it is likely that
classroom learners will reflect this change
But should instruction be changed If so who along the continuum from
the theorist-researcher to the teacher-practitioner should inform and guide
this change At which level In what way Such questions are at the very
heart of two on-going debates that I re-examine before proposing some
BRUCE ANDERSON 301
answers the pedagogical versus descriptive norm debate (eg Owen 1993
Widdowson 2000 and the collection of articles in Gass et al 2002) and
the interrelated native-speaker-usage-as-goal debate in foreign language
pedagogy (eg the collection of articles in Blyth 2003)
On the whole the approach to instruction on French adjective
position outlined earlier appears advantageous in its simplicity providing
easily remembered rules of thumb that learners can apply in language
production That this approach to instruction does not as we have seen
straightforwardly follow the descriptive norm is a situation that linguists
utilizing a corpus-based methodology believe can and should be remedied
Conrad (2000) for instance sees in corpus-based research an immediate way
in which linguistics can inform language pedagogy freeing language teachers
from lsquohav[ing] to rely on judgments of grammatical accuracy or on native
speakersrsquo intuition about what sounds naturalrsquo (2000 555) Although the
subsequent study by Biber and Reppen (2002) uses a disclaimer to the effect
that frequency should not be the sole factor in material design they
nevertheless suggest that lsquoa selective revision of pedagogy to reflect actual
use as shown by frequency studies could result in radical changes that
facilitate the learning process for studentsrsquo (2002 199) Owen (1993)
however provides examples of how a total reliance on a corpus does not
necessarily yield better observation and more importantly questions
whether better observation automatically equates to better explanation
(1993 168) (but see Francis and Sinclair 1994 for rebuttal) For his part
Widdowson (2000) objects to the kind of equation just described as a
case of linguistics applied a mistaken belief that what is textually attested
lsquouniquely represents real language and that this reality should define the
foreign language subjectrsquo (2000 8) (but see Stubbs 2001 for rebuttal) What
is needed according to Widdowson is not simply linguistics applied
to pedagogy but a true role for applied linguistics an approach to usage that
also recognizes native speakersrsquo awareness of and opinions about such usage
The need for a truly applied linguistics perspective on native speaker
usage (including frequency patterns) is embodied in a pedagogical norm
originally conceptualized in Valdman (1989) and elaborated upon in
Bardovi-Harlig and Gass (2002) as
[A] combination of language systems and forms selected by linguistsand pedagogues to serve as the immediate language target ortargets that learners seek to acquire during their language studyIn other words pedagogical norms represent a mid-point or seriesof mid-points for learners as they progress toward acquiringnative language norms [thus constituting] a form of languagethat is acceptable to native speakers but easier to learn than the fullnative language system (p 3)
The textbook word-order rules for French adjective position reviewed
earlier would first appear to constitute an appropriate pedagogical norm
302 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
under Bardovi-Harlig and Gassrsquos (2002) definition because the learner is able
to produce a subset of the totality of possible utterances deemed acceptable to
native speakers which may turn out to be even more the case in spoken than
in written French (though the difference in mediums remains to be studied)
The learner data presented here from Anderson (2002 2007) however
provides one piece of evidence that textbook rules and classroom input do not
constitute a sufficient pedagogical norm as learners have stopped in their
progress toward acquiring native speaker (ie descriptively accurate) norms
As the articles in Blyth (2003) attest however it is highly debatable
whether the lsquoNative Speakerrsquo (as an idealized abstraction) should be the
appropriate model for the contemporary learner For Kramsch (2002 2003)
too much of a focus on an authentic native speaker norm reduces
foreign language (FL) study to a constant lsquoapproximation to and conformity
with stereotypical native speakers thus often silencing other FL-specific
forms of language potentialrsquo (2002 60) Language learning in instructional
settings on my understanding of Kramschrsquos (2003) argument should not be
viewed solely in a teleological sensemdashas language acquisition directed
toward a fixed nativelike end-pointmdashbut rather as language appropriation
by learners in the sense that they are constructing a linguistic and even
cultural identity lsquoin the interstices of national languages and on the
margins of monolingual speakersrsquo territoriesrsquo (2003 260) The political nature
of the debate as highlighted by Kramsch becomes all the more apparent
when one additionally considers the status of the foreign language on
the world stage (eg French as an international language in retreat)
and how deeply the normative attitudes of its speakers run (eg Francersquos
three-century-long history of acceptance and maintenance of the standard
(Lodge 1993))
The foregoing synopsis of the debates surrounding native speaker usage as
both the goal of foreign language instruction and the extent to which
pedagogical norms need apply has highlighted the real world complexities
involved in what would otherwise seem to be a simple equation Revise
pedagogy to reflect actual use as shown by frequency studies and this
will result in a more targetlike performance on the part of learners So we
return then to the questions posed earlier (1) Should instruction be
changed (2) If so who along the continuum from the theorist-researcher to
the teacher-practitioner should inform and guide this change (3) At which
level (4) In what ways
With respect to our first question because one can empirically
demonstrate that the pedagogical norm for French adjective position
prevents learners from eventually acquiring native speaker (ie descriptively
accurate) norms instruction should be changed so thatmdashirrespective of
whether the attainment of such a norm is necessarily desirablemdashlearners
have both a choice in selecting one grammatical or lexical form over the other
and an awareness of the meaning potential of each choice (following Kramsch
2002 71)
BRUCE ANDERSON 303
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
accurate It is accurate in the sense that the ends of the continuum are
not equivalent in their percentages There are many more adjectives
appearing either only or typically in post-N position than is the case for pre-N
position It is inaccurate in the sense that it ignores the distributional
behavior of 595 percent of the most frequently used adjectives which can
appear in both positions French is not as the rule implies the mirror
opposite of English but instead is more accurately viewed as an lsquoANArsquo
(adjectivendashnounndashadjective) language as opposed to both lsquosystematic ANrsquo
languages like English and German and lsquosystematic NArsquo languages like
Indonesian and Thai (Cinque 1994 99ndash101)
The data also lead us to question the criteria textbooks use for including
certain adjectives in the two lists in (4) and (5) For the so-called prenominal
adjectives a frequency of 90 percent or greater in that position would
establish that adjectives such as excellent fameux lsquo(in)famousrsquo moindre
lsquoslightestrsquo and vrai lsquorealtruersquo need to be added That list itself implies a fixed
lexical class to be memorized though in reality it does nothing more
than provide a handful of frequently attested adjectives (nfrac14 15 [75 percent])
that typically appear just in that position Many more adjectives can
appear in pre-N position as well For the so-called change-of-meaning
adjectives the data show that the list in (5) falls woefully short in terms of
representing those adjectives that actually do regularly appear in both
positions (nfrac14 52)
Even if it were to be established that such rules are accurate with respect
to colloquial French as spoken between native speakers and thus appropriate
for oral proficiency-oriented classrooms such rules obviously fail to capture
the state of affairs in written texts particularly literature to which learners at
later course levels are exposed Yet even students at earlier course levels
occasionally find counter-examples to what they are taught The four cases of
lsquounexpectedrsquo pre-N position in (6) come from the cultural readings of two
first-year and two second-year US textbooks respectively
(6) a La France a un excellent systeme de transports publics(Magnan et al 2002 223)lsquoFrance has an excellent system of public transportrsquo
b Cela explique le grand nombre de magnifiques chateauxdans la region (Valdman and Pons 1997 395)lsquoThis explains the great number of magnificent chateaux inthe regionrsquo
c [U]ne veritable explosion de la technologie (St Onge etal 2003 71) lsquoa veritable explosion of technologyrsquo
d Lrsquohistoire est une douloureuse suite de tragedieshumaines (Oates and Dubois 2003 208)lsquoIts history is a painful series of human tragediesrsquo
One might wish to argue that the examples provided in (6) demonstrate
that inputmdashat least classroom-related inputmdashis not as contradictory to
descriptive norms as argued in this section Examples of the type in (6)
BRUCE ANDERSON 293
however are only occasionally found in textbook reading materials they are
included here to demonstrate that even when making a concerted effort to
write level-appropriate cultural readings textbook authors sometimes break
the very grammar rules they present elsewhere in the textbook suggesting a
true disparity between rules and lsquonaturalisticrsquo usage5 (I return to the issue of
descriptively accurate statements about French adjective position in the
discussion section)
FREQUENCY OF ADJECTIVE USE AND POSITIONIN CLASSROOM INPUT
Methodology
In order to get a general sense of how well the classroom setting as primary
input source reflects either textbook rules or the corpora-based descriptive
norm just summarized frequency counts of adjective type (ie discrete
lexemes) and their position in classroom input were collected by the
researcher from 60 hours of in-class observations of university-level French
courses at a large Midwestern American university Ten hours were spent in
each of six courses spanning the curriculummdashfrom the four semesters
comprising the first- and second-year language skills curriculum (referred to
as Levels 1a 1b 2a and 2b below) to a third-year conversation course
(Level 3) and a fourth-year lecture course on French civilization (Level 4)
Each of these courses was taught almost exclusively in French by a mix of
native and nonnative instructors none of whom were aware of the purpose
of this observation period Class observations were conducted on days when
adjective position was not the focus of explicit instruction as this would have
more than likely biased the frequency counts
The first- and second-year courses consisted of explicit grammar
instruction drills and communicative activities normally involving students
in pair or group work The conversation and civilization courses involved
no explicit grammar instruction the former involved mostly teacherndash
student and studentndashstudent oral communication based on previous course
readings and video segments whereas the latter was almost exclusively
devoted to instructor-delivered lectures Error correction was minimal in all
courses and no case of error correction involving adjective position was
witnessed during the observation period (despite the fact that some errors
were made)
Oral input on adjective position in all courses came from one of three
sources the instructor audio aides (such as cassette and videotape) and
the students themselves Written input in the first- and second-year
courses came from two sources visual aids (such as handouts transparencies
writing on the blackboard) and passages from the textbook that were read
silently or aloud in class At Level 3 written materials included authentic
newspaper and magazine articles in addition to visual aides at Level 4
294 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
a number of works of French literature were read outside of class Note
that written input in these latter two course levels is qualitatively the same
as that used in the corpora compiled by Forsgren (1978) Wilmet (1980) and
Larsson (1993)
At issue during this period of investigation was classroom input that could
be seen andor heard by all students in the classroom (as well as the
researcher) Input from students involved in pair and group work was
therefore excluded from analysis Material to be read outside of class was
also excluded as there was no guarantee that all students had actually
read such material Although the totality of inputmdashincluding sources of input
the learner might have voluntarily sought out outside of classmdashcould
obviously never be exhaustively tallied the exclusion of certain forms of
input was first and foremost a matter of logistics (ie not having the
appropriate funds equipment facilities and time to sound record then
transcribe all classroom input from every participant in every situation of
use) Despite this drawback I see no reason to suspect that student-delivered
input to other students during pair- or group-work would be qualitatively
different (at least in terms of adjective position) from the input delivered by
students that was audible to the entire class this latter source of input
already constitutes 10ndash30 percent of all sources of input tallied as part of the
observation period
Each instance of adjective use during each class session was recorded with
pen and paper by the researcher A coding scheme was devised whereby an
11 17-inch sheet of paper was divided into quadrants labeled according to
the source of the input lsquoinstructorrsquo lsquostudentrsquo lsquoaudiorsquo and lsquowrittenrsquo Each
quadrant was then subdivided according to the position of adjectival
modification lsquopre-Nrsquo lsquopost-Nrsquo lsquopredicatersquo and lsquootherrsquo (to be described
below) Input of the form une bonne idee lsquoa good idearsquo by the instructor was
therefore recorded as an instance of the adjective bonne in the instructor
quadrant pre-N sub-quadrant Hatch marks next to the adjective bonne were
used for any further exemplars of this adjective (including its masculine form
bon) in this position from this source I acknowledge that mere human error
caused by fatigue etc may have resulted in tallies that are less accurate than
if each session were recorded on tape Though the results presented below
should be considered with some caution then occasional errors could not
have reached a significant enough level to detract from the highly salient
patterns of usage that we will find
The resulting sixty sheets each representing one 50-minute class session
were then tallied in order to determine (a) the relative contribution of the
four sources of input (b) the average token (exemplar) count of adjectives
per 50-minute class period (c) the ratio of pre-N to post-Npredicate tokens
(d) the type count (number of discrete lexemes) entering into pre-N and
post-N positions and (e) the frequency of variation in positionmdashthat is the
extent to which a given adjective appeared at least once in both pre-N and
post-N position during the ten hours of observation
BRUCE ANDERSON 295
Results
The instructor not surprisingly was the primary provider of input in the
classroom (See Figure 1 available to online subscribers at http
applijoxfordjournalsorg) This was especially the case in the first-year
courses (Levels 1a 1b) the second- and third-year courses (Levels 2a 2b 3)
showed a greater proportion of input from students and from audio aides
The instructor was by far the predominant source of input in the fourth-year
course (Level 4) given that it involved instructor-delivered lectures on
French civilization with minimal class participation on the part of students
Concerning the average number of times an adjective occurred in the
input during any one 50-minute class period we find a general increase with
level from 42 tokens at Level 1a to 156 tokens per class at Level 4 (See
Figure 2 available to online subscribers at httpapplijoxfordjournalsorg)
Students therefore heard an adjective used anywhere from one to three
times per minute on average
Aggregate data on the frequency with which adjectives appeared in pre-N
post-N and predicate positions or in other constructions (eg used in
isolation in elliptical constructions such as la charmante lsquothe charming [one]rsquo
and following indefinite pronouns as in quelqursquoun de charmant lsquosomeone
charmingrsquo) demonstrate that post-N and predicate positions were generally
equally frequent and together constitute fully 75 percent of the ratio whereas
pre-N position never surpassed 27 percent across course levels Other
constructions were minimal (See Figure 3 available to online subscribers at
httpapplijoxfordjournalsorg)
This roughly 75 percentndash25 percent ratio was then reanalyzed in order to
consider the number of distinct lexemes (lsquotypesrsquo) that make up the
percentages We find for example that only 15 lexemes make up the 14
percent of pre-N adjective use at Level 1a This number gradually increased
with course level but only reached a total of 41 by Level 4 Post-N position
on the other hand admitted a much larger variety of adjectives in classroom
input from 71 at Level 1a to 323 by Level 4 (See Figure 4 available to online
subscribers at httpapplijoxfordjournalsorg) Students therefore got
relatively little input on pre-N position and what input they did get was
made up of a relatively small class of adjectives being used over and over
again and which already figure in the textbook-determined list of pre-N
adjectives in (4) (See Appendix A available to online subscribers at http
applijoxfordjournalsorg for a list of adjectives appearing in pre-N position
by course level)
Finally we find that very few adjectives varied in position during the 10
hours of in-class observation There was no variation attested at Level 1a
and only four to seven adjectives varied in position from Level 1b through to
Level 3 Moreover some of the input provided to learners in these courses
in the form of other studentsrsquo output was nonnativelike6 At the most
advanced undergraduate level learners got somewhat more input involving
296 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
variation in position where 20 adjectives did so (See Figure 5 available to
online subscribers at httpapplijoxfordjournalsorg) Again most (but not
all) of these adjectives already figure into the textbook-determined list of
variable adjectives in (5) (See Appendix B available to online subscribers at
httpapplijoxfordjournalsorg for a list of adjectives varying in position by
course level)
To summarize the observational data just presented the ambient language
of the classroom clearly mimics the pedagogical treatment that American
university-level classroom learners receive from textbooks Although
adjectives are in general a very frequent part of the input the rate of
pre-N position (at roughly 25 percent) is essentially static across levels
pointing to the fact that classroom input across levels is in no way
contradictory to what learners are initially taught Post-N position is basic
(unmarked) and pre-N position is exceptional (marked) The adjectives that
do appear in pre-N position in classroom input are in most cases those that
learners already expect to appear in that position (cf the list in (4) and
Appendix A) The few adjectives that appear in both positions are again in
most cases those that learners expect to do so (cf the list in (5) and
Appendix B)
By the end of the first-year level then both instruction and classroom
input would lead learners to expect that adjectives will appear in post-N
position with the notable exception of the class of adjectives in (4) (those in
(5) having not yet been introduced) By the second-year level it could be
argued that instruction provides examples in the form of so-called change of
meaning adjectives that variation in position is to be associated with
variation in interpretation (and perhaps by implication always to
interpretation) However learnersrsquo explicit knowledge of interpretative
differences as exemplified by the adjectives in (5) typically consists of
arbitrary and unsystematic English glosses By arbitrary I mean that
instruction never provides an adequate explanation for why the pre-N
position of cher should correspond to beloved or for its post-N position to
correspond to expensive7 Though telling students following Bragger and Rice
(1996) that the meaning in post-N position is lsquoconcretersquo and the meaning in
pre-N position is lsquofigurativersquo might (somehow) help it does not account for
other adjectives in (5) like prochain lsquonextrsquo This latter adjective also
demonstrates the fact that English glosses are unsystematic in that prochain
translates as lsquonextrsquo in either position
Before considering our final research question I note that no research on
adjective position in native French speaker discourse is available to date
An anonymous Applied Linguistics reviewer questions the validity of compar-
ing primarily spoken classroom discoursemdashor any other kind of spoken
discoursemdashto written texts without first demonstrating that adjective position
is insensitive to this difference in medium While recognizing the potentially
important distinction in medium and the need for further research in this
area the validity of comparing classroom input to written texts in the present
BRUCE ANDERSON 297
study lies in the fact that such a comparison roughly corresponds to the
curriculum of typical university-level courses an emphasis on listening
comprehension and oral communication in the first- and second-year
language courses leading to an emphasis on reading comprehension
(typically literature-based) and proficiency in writing in the third- and
fourth-year courses Were it to turn out that the degree of variation in
adjective position found in written texts is restricted to just that medium
revisions to pedagogy would necessarily be restricted to writing-intensive
(composition creative composition stylistics) courses at the third- and
fourth-year levels (I return to this issue in the discussion section)
LEARNER INTUITIONS OF ACCEPTABILITY
Thus far we have seen an incongruity between textbook presentations of
adjective position and authentic text sources across a number of written text
genres (research question 1) and that ambient language use involving
adjective position in the classroom is clearly more congruent with and might
therefore be taken to reinforce the explicit rules found in textbooks
(research question 2) We are now in a position to address our third research
questionmdashthe extent to which intuitions of acceptability among classroom
learners are biased by the congruity between pedagogical treatment and
classroom input frequency patterns An empirical study previously reported
in Anderson (2002 2007) can shed some light on this question
Methodology
Anderson (2002 2007) investigated the acquisition of various morphosyn-
tactic properties of French noun phrases by administering an acceptability
judgment task in which participants had to evaluate the appropriateness of
sentences featuring among other things the pre- versus post-N position of
adjectives Each of the adjectives selected were lsquohighly variablersquo with respect
to position (cf Table 1) yet crucially do not figure in textbook lists of
adjectives said to change meaning based on position (cf the list in (10))
These include lourd lsquoheavyrsquo violent lsquoviolentrsquo charmant lsquocharmingrsquo precieux
lsquopreciousrsquo epais lsquothickrsquo etrange lsquostrangersquo and delicieux lsquodeliciousrsquo (See
Appendix C available to online subscribers at httpapplijoxfordjournals
org for test sentences) Study participants included English-speaking learners
of French in the second third and fourth years of undergraduate study
(nfrac14 80) who were attending courses at the same institution at which the
observational data were collected a group of advanced learners (nfrac14 20)
consisting of graduate student instructors at that same institution as well as
secondary education teachers and a group of native French speakers (nfrac14 27)
roughly matched in age and education level
Each adjective appeared in test sentences once prenominally and once
postnominally with each test sentence preceded once by a context
298 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
motivating acceptance and once by a context motivating rejection of one or
the other adjective order for a total of four positionndashinterpretation pairings
per adjective These were randomly distributed across two testing
instruments administered a week apart After reading each context and
paired test sentence study participants were directed to check boxes
labeled lsquofinersquo lsquooddrsquo or lsquocannot decidersquo to the prompt This particular sentence
sounds ___ in this context
As an example each of the sentences in (7) featuring the adjective lourd
lsquoheavyrsquo was paired with each of two contexts
(7) a Pierre doit laisser la lourde valise avec son amib Pierre doit laisser la valise lourde a lrsquoaeroport
lsquoPierre has to leave the heavy suitcase with his friendatthe airportrsquo
In the first context motivating a unique noun-referent interpretation
there was one and only one suitcase which happened to be heavy and
which lsquoPierrersquo had to leave behind with his friend In the second context
motivating a multiple noun-referent interpretation there were a number of
suitcases only one of which was heavy and it was this one that lsquoPierrersquo had
to leave behind at the airport (See Appendix D available to online
subscribers at httpapplijoxfordjournalsorg for the complete version of
each context)
When viewed solely in terms of truth-conditional semantics each of the
four positionndashinterpretation pairings is compatible save for one pre-N
position (7a) in a multiple noun-referent context This is because the singular
definite article la and pre-N adjective lourde together presuppose a unique
suitcase in context (which happens to be heavy) This presupposition
does not hold when the adjective is in post-N position where all that is
required is that there be at least one such suitcase in context The testing
instrument then allows us to examine how classroom learners react to
an lsquoexpectedrsquo word order (ie sentences like (7b) in either context) as well
as to an lsquounexpectedrsquo but semantically and pragmatically appropriate
word order (ie sentences like (7a) but only in the unique noun-referent
context)
Results
As the mean acceptance rates from native French speakers in Table 2 show
truth-conditional semantics was not the only factor determining acceptance
rates Two of the pairings received essentially chance-level rates of
acceptance (pre-N position in a multiple noun-referent context at 49
percent and post-N position in a unique context at 52 percent) whereas the
other two pairings showed much clearer above- or below-chance rates (pre-N
position in a unique noun-referent context at 74 percent and post-N
position in a multiple noun-referent context at 37 percent)
BRUCE ANDERSON 299
As was argued in Anderson (2002 2007) results for each of the four
positionndashinterpretation pairings among the native French speakers are
explainable in terms of a disproportionate interaction between semantics and
pragmatic implicature in particular the Gricean maxim of quantity An
utterance should be no more and no less informative than is required for the
current purposes of the exchange (Grice 1975) The positionndashinterpretation
pairing that was lsquotruersquo in terms of semantics and also the most pragmatically
appropriate was accepted at the highest rate (74 percent) the pairing that
was also lsquotruersquo in terms of semantics but was pragmatically the least
appropriate was accepted at the lowest rate (37 percent)8
Do learner group response patterns show the same bias toward pragmatic
implicature in their acceptance rates The comparison of mean acceptance
rates between native French speakers and learner groups in Table 3 shows
that learners performed in a particular (and particularly non-Frenchlike) way
in which the interaction goes in the opposite direction favoring semantics
over pragmatics
Recall that the only positionndashinterpretation pairing that was semantically
incompatible was pre-N position (7a) in a multiple noun-referent context
given the presupposition of uniqueness of the noun referent All learner
groups did indeed accept this pairing at the lowest rate of the four and with
the exception of the second-year learners (at thorn9 percent) did so more
strongly than the native French speakers at mean acceptance rates that were
11ndash15 percent lower)9 However they also accepted at much lower mean
rates (12 to 23 percent) the most felicitous positionndashinterpretation pairing
for the native French speakers pre-N position in a unique noun-referent
context Recall also that post-N position (7b) was semantically compatible
under either interpretation All learner groups accepted these two positionndash
interpretation pairings at much higher rates than the native French speakers
(thorn23 to thorn31 percent) including the least felicitous positionndashinterpretation
pairing for the native French speakersmdashpost-N position in a multiple noun-
referent context This is precisely what one would expect from the
pedagogical treatment these learners are provided and the congruity of
that treatment with classroom input which under-represents actual usage in
Table 2 Native French speaker mean acceptance rates for the uniquenonunique noun-referent distinction
Prenominal (AN)order ()
Postnominal (NA)order ()
Unique noun referent 74 52
Nonunique noun referent 49 37
Source Adapted from Anderson 2002 2007
300 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
text-based corpora and as we have now seen differs from native speaker
intuitions of overall (semantico-pragmatic) acceptability
Although a bias toward or even strict adherence to explicit pedagogical
rules in evaluating such sentences may not seem all that surprising what does
stand out in the data from Anderson (2002 2007) is that these differential
rates in acceptance between native and nonnative speakers are as true for the
advanced group in that studymdashconsisting of graduate student instructors and
second education teachersmdashas they are for the intermediate (second-year)
learner group This would seem to indicate no increased sensitivity to
contextual (pragmatic) factors in adjective position over the course of studying
French even though the advanced group had been exposed to much more
authentic input through texts written for and by native French speakers (ie
the descriptive norm uncovered in our examination of authentic text corpora)
DISCUSSION
What should be clear from the findings presented here is that rules of French
adjective placement that one typically finds in (at least American) textbooks
though not plainly incorrect are not fully descriptively accurate either at
least with respect to written French across several text genres In particular
such rules portray pre-N and post-N positions as mutually exclusive for
almost all adjectives when in fact they are not
Similar discrepancies are often found between rules for and by native
speakers and what those speakers actually do One need look no further than
the French for an example with colloquial speech often at odds with the
prescriptive bon usage and the admonishments of the Academie francaise One
might therefore also expect to find a discrepancy between textbook rules and
what teachers do in the classroom andor what students find to be
lsquoacceptablersquo in terms of word order As we have seen however this is not
the case These rules are faithfully reflected in classroom language use by
instructors to such an extent that when asked to evaluate the acceptability
of lsquoexpectedrsquo and lsquounexpected but nativelikersquo word orders (and one would
presume in production as well) both instructors and students paid much
less attention to problems in pragmatic implicature than did native speakers
a situation which does not appear to change even in the face of counter-
examples occasionally provided in the textbooks they use (see (6) above) and
most certainly in the written texts that they read at later course levels As
pointed out by an anonymous reviewer this situation can be considered a
testament to the power of classroom instruction (instructor input in
particular) if instruction is changed he or she reasons it is likely that
classroom learners will reflect this change
But should instruction be changed If so who along the continuum from
the theorist-researcher to the teacher-practitioner should inform and guide
this change At which level In what way Such questions are at the very
heart of two on-going debates that I re-examine before proposing some
BRUCE ANDERSON 301
answers the pedagogical versus descriptive norm debate (eg Owen 1993
Widdowson 2000 and the collection of articles in Gass et al 2002) and
the interrelated native-speaker-usage-as-goal debate in foreign language
pedagogy (eg the collection of articles in Blyth 2003)
On the whole the approach to instruction on French adjective
position outlined earlier appears advantageous in its simplicity providing
easily remembered rules of thumb that learners can apply in language
production That this approach to instruction does not as we have seen
straightforwardly follow the descriptive norm is a situation that linguists
utilizing a corpus-based methodology believe can and should be remedied
Conrad (2000) for instance sees in corpus-based research an immediate way
in which linguistics can inform language pedagogy freeing language teachers
from lsquohav[ing] to rely on judgments of grammatical accuracy or on native
speakersrsquo intuition about what sounds naturalrsquo (2000 555) Although the
subsequent study by Biber and Reppen (2002) uses a disclaimer to the effect
that frequency should not be the sole factor in material design they
nevertheless suggest that lsquoa selective revision of pedagogy to reflect actual
use as shown by frequency studies could result in radical changes that
facilitate the learning process for studentsrsquo (2002 199) Owen (1993)
however provides examples of how a total reliance on a corpus does not
necessarily yield better observation and more importantly questions
whether better observation automatically equates to better explanation
(1993 168) (but see Francis and Sinclair 1994 for rebuttal) For his part
Widdowson (2000) objects to the kind of equation just described as a
case of linguistics applied a mistaken belief that what is textually attested
lsquouniquely represents real language and that this reality should define the
foreign language subjectrsquo (2000 8) (but see Stubbs 2001 for rebuttal) What
is needed according to Widdowson is not simply linguistics applied
to pedagogy but a true role for applied linguistics an approach to usage that
also recognizes native speakersrsquo awareness of and opinions about such usage
The need for a truly applied linguistics perspective on native speaker
usage (including frequency patterns) is embodied in a pedagogical norm
originally conceptualized in Valdman (1989) and elaborated upon in
Bardovi-Harlig and Gass (2002) as
[A] combination of language systems and forms selected by linguistsand pedagogues to serve as the immediate language target ortargets that learners seek to acquire during their language studyIn other words pedagogical norms represent a mid-point or seriesof mid-points for learners as they progress toward acquiringnative language norms [thus constituting] a form of languagethat is acceptable to native speakers but easier to learn than the fullnative language system (p 3)
The textbook word-order rules for French adjective position reviewed
earlier would first appear to constitute an appropriate pedagogical norm
302 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
under Bardovi-Harlig and Gassrsquos (2002) definition because the learner is able
to produce a subset of the totality of possible utterances deemed acceptable to
native speakers which may turn out to be even more the case in spoken than
in written French (though the difference in mediums remains to be studied)
The learner data presented here from Anderson (2002 2007) however
provides one piece of evidence that textbook rules and classroom input do not
constitute a sufficient pedagogical norm as learners have stopped in their
progress toward acquiring native speaker (ie descriptively accurate) norms
As the articles in Blyth (2003) attest however it is highly debatable
whether the lsquoNative Speakerrsquo (as an idealized abstraction) should be the
appropriate model for the contemporary learner For Kramsch (2002 2003)
too much of a focus on an authentic native speaker norm reduces
foreign language (FL) study to a constant lsquoapproximation to and conformity
with stereotypical native speakers thus often silencing other FL-specific
forms of language potentialrsquo (2002 60) Language learning in instructional
settings on my understanding of Kramschrsquos (2003) argument should not be
viewed solely in a teleological sensemdashas language acquisition directed
toward a fixed nativelike end-pointmdashbut rather as language appropriation
by learners in the sense that they are constructing a linguistic and even
cultural identity lsquoin the interstices of national languages and on the
margins of monolingual speakersrsquo territoriesrsquo (2003 260) The political nature
of the debate as highlighted by Kramsch becomes all the more apparent
when one additionally considers the status of the foreign language on
the world stage (eg French as an international language in retreat)
and how deeply the normative attitudes of its speakers run (eg Francersquos
three-century-long history of acceptance and maintenance of the standard
(Lodge 1993))
The foregoing synopsis of the debates surrounding native speaker usage as
both the goal of foreign language instruction and the extent to which
pedagogical norms need apply has highlighted the real world complexities
involved in what would otherwise seem to be a simple equation Revise
pedagogy to reflect actual use as shown by frequency studies and this
will result in a more targetlike performance on the part of learners So we
return then to the questions posed earlier (1) Should instruction be
changed (2) If so who along the continuum from the theorist-researcher to
the teacher-practitioner should inform and guide this change (3) At which
level (4) In what ways
With respect to our first question because one can empirically
demonstrate that the pedagogical norm for French adjective position
prevents learners from eventually acquiring native speaker (ie descriptively
accurate) norms instruction should be changed so thatmdashirrespective of
whether the attainment of such a norm is necessarily desirablemdashlearners
have both a choice in selecting one grammatical or lexical form over the other
and an awareness of the meaning potential of each choice (following Kramsch
2002 71)
BRUCE ANDERSON 303
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
however are only occasionally found in textbook reading materials they are
included here to demonstrate that even when making a concerted effort to
write level-appropriate cultural readings textbook authors sometimes break
the very grammar rules they present elsewhere in the textbook suggesting a
true disparity between rules and lsquonaturalisticrsquo usage5 (I return to the issue of
descriptively accurate statements about French adjective position in the
discussion section)
FREQUENCY OF ADJECTIVE USE AND POSITIONIN CLASSROOM INPUT
Methodology
In order to get a general sense of how well the classroom setting as primary
input source reflects either textbook rules or the corpora-based descriptive
norm just summarized frequency counts of adjective type (ie discrete
lexemes) and their position in classroom input were collected by the
researcher from 60 hours of in-class observations of university-level French
courses at a large Midwestern American university Ten hours were spent in
each of six courses spanning the curriculummdashfrom the four semesters
comprising the first- and second-year language skills curriculum (referred to
as Levels 1a 1b 2a and 2b below) to a third-year conversation course
(Level 3) and a fourth-year lecture course on French civilization (Level 4)
Each of these courses was taught almost exclusively in French by a mix of
native and nonnative instructors none of whom were aware of the purpose
of this observation period Class observations were conducted on days when
adjective position was not the focus of explicit instruction as this would have
more than likely biased the frequency counts
The first- and second-year courses consisted of explicit grammar
instruction drills and communicative activities normally involving students
in pair or group work The conversation and civilization courses involved
no explicit grammar instruction the former involved mostly teacherndash
student and studentndashstudent oral communication based on previous course
readings and video segments whereas the latter was almost exclusively
devoted to instructor-delivered lectures Error correction was minimal in all
courses and no case of error correction involving adjective position was
witnessed during the observation period (despite the fact that some errors
were made)
Oral input on adjective position in all courses came from one of three
sources the instructor audio aides (such as cassette and videotape) and
the students themselves Written input in the first- and second-year
courses came from two sources visual aids (such as handouts transparencies
writing on the blackboard) and passages from the textbook that were read
silently or aloud in class At Level 3 written materials included authentic
newspaper and magazine articles in addition to visual aides at Level 4
294 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
a number of works of French literature were read outside of class Note
that written input in these latter two course levels is qualitatively the same
as that used in the corpora compiled by Forsgren (1978) Wilmet (1980) and
Larsson (1993)
At issue during this period of investigation was classroom input that could
be seen andor heard by all students in the classroom (as well as the
researcher) Input from students involved in pair and group work was
therefore excluded from analysis Material to be read outside of class was
also excluded as there was no guarantee that all students had actually
read such material Although the totality of inputmdashincluding sources of input
the learner might have voluntarily sought out outside of classmdashcould
obviously never be exhaustively tallied the exclusion of certain forms of
input was first and foremost a matter of logistics (ie not having the
appropriate funds equipment facilities and time to sound record then
transcribe all classroom input from every participant in every situation of
use) Despite this drawback I see no reason to suspect that student-delivered
input to other students during pair- or group-work would be qualitatively
different (at least in terms of adjective position) from the input delivered by
students that was audible to the entire class this latter source of input
already constitutes 10ndash30 percent of all sources of input tallied as part of the
observation period
Each instance of adjective use during each class session was recorded with
pen and paper by the researcher A coding scheme was devised whereby an
11 17-inch sheet of paper was divided into quadrants labeled according to
the source of the input lsquoinstructorrsquo lsquostudentrsquo lsquoaudiorsquo and lsquowrittenrsquo Each
quadrant was then subdivided according to the position of adjectival
modification lsquopre-Nrsquo lsquopost-Nrsquo lsquopredicatersquo and lsquootherrsquo (to be described
below) Input of the form une bonne idee lsquoa good idearsquo by the instructor was
therefore recorded as an instance of the adjective bonne in the instructor
quadrant pre-N sub-quadrant Hatch marks next to the adjective bonne were
used for any further exemplars of this adjective (including its masculine form
bon) in this position from this source I acknowledge that mere human error
caused by fatigue etc may have resulted in tallies that are less accurate than
if each session were recorded on tape Though the results presented below
should be considered with some caution then occasional errors could not
have reached a significant enough level to detract from the highly salient
patterns of usage that we will find
The resulting sixty sheets each representing one 50-minute class session
were then tallied in order to determine (a) the relative contribution of the
four sources of input (b) the average token (exemplar) count of adjectives
per 50-minute class period (c) the ratio of pre-N to post-Npredicate tokens
(d) the type count (number of discrete lexemes) entering into pre-N and
post-N positions and (e) the frequency of variation in positionmdashthat is the
extent to which a given adjective appeared at least once in both pre-N and
post-N position during the ten hours of observation
BRUCE ANDERSON 295
Results
The instructor not surprisingly was the primary provider of input in the
classroom (See Figure 1 available to online subscribers at http
applijoxfordjournalsorg) This was especially the case in the first-year
courses (Levels 1a 1b) the second- and third-year courses (Levels 2a 2b 3)
showed a greater proportion of input from students and from audio aides
The instructor was by far the predominant source of input in the fourth-year
course (Level 4) given that it involved instructor-delivered lectures on
French civilization with minimal class participation on the part of students
Concerning the average number of times an adjective occurred in the
input during any one 50-minute class period we find a general increase with
level from 42 tokens at Level 1a to 156 tokens per class at Level 4 (See
Figure 2 available to online subscribers at httpapplijoxfordjournalsorg)
Students therefore heard an adjective used anywhere from one to three
times per minute on average
Aggregate data on the frequency with which adjectives appeared in pre-N
post-N and predicate positions or in other constructions (eg used in
isolation in elliptical constructions such as la charmante lsquothe charming [one]rsquo
and following indefinite pronouns as in quelqursquoun de charmant lsquosomeone
charmingrsquo) demonstrate that post-N and predicate positions were generally
equally frequent and together constitute fully 75 percent of the ratio whereas
pre-N position never surpassed 27 percent across course levels Other
constructions were minimal (See Figure 3 available to online subscribers at
httpapplijoxfordjournalsorg)
This roughly 75 percentndash25 percent ratio was then reanalyzed in order to
consider the number of distinct lexemes (lsquotypesrsquo) that make up the
percentages We find for example that only 15 lexemes make up the 14
percent of pre-N adjective use at Level 1a This number gradually increased
with course level but only reached a total of 41 by Level 4 Post-N position
on the other hand admitted a much larger variety of adjectives in classroom
input from 71 at Level 1a to 323 by Level 4 (See Figure 4 available to online
subscribers at httpapplijoxfordjournalsorg) Students therefore got
relatively little input on pre-N position and what input they did get was
made up of a relatively small class of adjectives being used over and over
again and which already figure in the textbook-determined list of pre-N
adjectives in (4) (See Appendix A available to online subscribers at http
applijoxfordjournalsorg for a list of adjectives appearing in pre-N position
by course level)
Finally we find that very few adjectives varied in position during the 10
hours of in-class observation There was no variation attested at Level 1a
and only four to seven adjectives varied in position from Level 1b through to
Level 3 Moreover some of the input provided to learners in these courses
in the form of other studentsrsquo output was nonnativelike6 At the most
advanced undergraduate level learners got somewhat more input involving
296 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
variation in position where 20 adjectives did so (See Figure 5 available to
online subscribers at httpapplijoxfordjournalsorg) Again most (but not
all) of these adjectives already figure into the textbook-determined list of
variable adjectives in (5) (See Appendix B available to online subscribers at
httpapplijoxfordjournalsorg for a list of adjectives varying in position by
course level)
To summarize the observational data just presented the ambient language
of the classroom clearly mimics the pedagogical treatment that American
university-level classroom learners receive from textbooks Although
adjectives are in general a very frequent part of the input the rate of
pre-N position (at roughly 25 percent) is essentially static across levels
pointing to the fact that classroom input across levels is in no way
contradictory to what learners are initially taught Post-N position is basic
(unmarked) and pre-N position is exceptional (marked) The adjectives that
do appear in pre-N position in classroom input are in most cases those that
learners already expect to appear in that position (cf the list in (4) and
Appendix A) The few adjectives that appear in both positions are again in
most cases those that learners expect to do so (cf the list in (5) and
Appendix B)
By the end of the first-year level then both instruction and classroom
input would lead learners to expect that adjectives will appear in post-N
position with the notable exception of the class of adjectives in (4) (those in
(5) having not yet been introduced) By the second-year level it could be
argued that instruction provides examples in the form of so-called change of
meaning adjectives that variation in position is to be associated with
variation in interpretation (and perhaps by implication always to
interpretation) However learnersrsquo explicit knowledge of interpretative
differences as exemplified by the adjectives in (5) typically consists of
arbitrary and unsystematic English glosses By arbitrary I mean that
instruction never provides an adequate explanation for why the pre-N
position of cher should correspond to beloved or for its post-N position to
correspond to expensive7 Though telling students following Bragger and Rice
(1996) that the meaning in post-N position is lsquoconcretersquo and the meaning in
pre-N position is lsquofigurativersquo might (somehow) help it does not account for
other adjectives in (5) like prochain lsquonextrsquo This latter adjective also
demonstrates the fact that English glosses are unsystematic in that prochain
translates as lsquonextrsquo in either position
Before considering our final research question I note that no research on
adjective position in native French speaker discourse is available to date
An anonymous Applied Linguistics reviewer questions the validity of compar-
ing primarily spoken classroom discoursemdashor any other kind of spoken
discoursemdashto written texts without first demonstrating that adjective position
is insensitive to this difference in medium While recognizing the potentially
important distinction in medium and the need for further research in this
area the validity of comparing classroom input to written texts in the present
BRUCE ANDERSON 297
study lies in the fact that such a comparison roughly corresponds to the
curriculum of typical university-level courses an emphasis on listening
comprehension and oral communication in the first- and second-year
language courses leading to an emphasis on reading comprehension
(typically literature-based) and proficiency in writing in the third- and
fourth-year courses Were it to turn out that the degree of variation in
adjective position found in written texts is restricted to just that medium
revisions to pedagogy would necessarily be restricted to writing-intensive
(composition creative composition stylistics) courses at the third- and
fourth-year levels (I return to this issue in the discussion section)
LEARNER INTUITIONS OF ACCEPTABILITY
Thus far we have seen an incongruity between textbook presentations of
adjective position and authentic text sources across a number of written text
genres (research question 1) and that ambient language use involving
adjective position in the classroom is clearly more congruent with and might
therefore be taken to reinforce the explicit rules found in textbooks
(research question 2) We are now in a position to address our third research
questionmdashthe extent to which intuitions of acceptability among classroom
learners are biased by the congruity between pedagogical treatment and
classroom input frequency patterns An empirical study previously reported
in Anderson (2002 2007) can shed some light on this question
Methodology
Anderson (2002 2007) investigated the acquisition of various morphosyn-
tactic properties of French noun phrases by administering an acceptability
judgment task in which participants had to evaluate the appropriateness of
sentences featuring among other things the pre- versus post-N position of
adjectives Each of the adjectives selected were lsquohighly variablersquo with respect
to position (cf Table 1) yet crucially do not figure in textbook lists of
adjectives said to change meaning based on position (cf the list in (10))
These include lourd lsquoheavyrsquo violent lsquoviolentrsquo charmant lsquocharmingrsquo precieux
lsquopreciousrsquo epais lsquothickrsquo etrange lsquostrangersquo and delicieux lsquodeliciousrsquo (See
Appendix C available to online subscribers at httpapplijoxfordjournals
org for test sentences) Study participants included English-speaking learners
of French in the second third and fourth years of undergraduate study
(nfrac14 80) who were attending courses at the same institution at which the
observational data were collected a group of advanced learners (nfrac14 20)
consisting of graduate student instructors at that same institution as well as
secondary education teachers and a group of native French speakers (nfrac14 27)
roughly matched in age and education level
Each adjective appeared in test sentences once prenominally and once
postnominally with each test sentence preceded once by a context
298 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
motivating acceptance and once by a context motivating rejection of one or
the other adjective order for a total of four positionndashinterpretation pairings
per adjective These were randomly distributed across two testing
instruments administered a week apart After reading each context and
paired test sentence study participants were directed to check boxes
labeled lsquofinersquo lsquooddrsquo or lsquocannot decidersquo to the prompt This particular sentence
sounds ___ in this context
As an example each of the sentences in (7) featuring the adjective lourd
lsquoheavyrsquo was paired with each of two contexts
(7) a Pierre doit laisser la lourde valise avec son amib Pierre doit laisser la valise lourde a lrsquoaeroport
lsquoPierre has to leave the heavy suitcase with his friendatthe airportrsquo
In the first context motivating a unique noun-referent interpretation
there was one and only one suitcase which happened to be heavy and
which lsquoPierrersquo had to leave behind with his friend In the second context
motivating a multiple noun-referent interpretation there were a number of
suitcases only one of which was heavy and it was this one that lsquoPierrersquo had
to leave behind at the airport (See Appendix D available to online
subscribers at httpapplijoxfordjournalsorg for the complete version of
each context)
When viewed solely in terms of truth-conditional semantics each of the
four positionndashinterpretation pairings is compatible save for one pre-N
position (7a) in a multiple noun-referent context This is because the singular
definite article la and pre-N adjective lourde together presuppose a unique
suitcase in context (which happens to be heavy) This presupposition
does not hold when the adjective is in post-N position where all that is
required is that there be at least one such suitcase in context The testing
instrument then allows us to examine how classroom learners react to
an lsquoexpectedrsquo word order (ie sentences like (7b) in either context) as well
as to an lsquounexpectedrsquo but semantically and pragmatically appropriate
word order (ie sentences like (7a) but only in the unique noun-referent
context)
Results
As the mean acceptance rates from native French speakers in Table 2 show
truth-conditional semantics was not the only factor determining acceptance
rates Two of the pairings received essentially chance-level rates of
acceptance (pre-N position in a multiple noun-referent context at 49
percent and post-N position in a unique context at 52 percent) whereas the
other two pairings showed much clearer above- or below-chance rates (pre-N
position in a unique noun-referent context at 74 percent and post-N
position in a multiple noun-referent context at 37 percent)
BRUCE ANDERSON 299
As was argued in Anderson (2002 2007) results for each of the four
positionndashinterpretation pairings among the native French speakers are
explainable in terms of a disproportionate interaction between semantics and
pragmatic implicature in particular the Gricean maxim of quantity An
utterance should be no more and no less informative than is required for the
current purposes of the exchange (Grice 1975) The positionndashinterpretation
pairing that was lsquotruersquo in terms of semantics and also the most pragmatically
appropriate was accepted at the highest rate (74 percent) the pairing that
was also lsquotruersquo in terms of semantics but was pragmatically the least
appropriate was accepted at the lowest rate (37 percent)8
Do learner group response patterns show the same bias toward pragmatic
implicature in their acceptance rates The comparison of mean acceptance
rates between native French speakers and learner groups in Table 3 shows
that learners performed in a particular (and particularly non-Frenchlike) way
in which the interaction goes in the opposite direction favoring semantics
over pragmatics
Recall that the only positionndashinterpretation pairing that was semantically
incompatible was pre-N position (7a) in a multiple noun-referent context
given the presupposition of uniqueness of the noun referent All learner
groups did indeed accept this pairing at the lowest rate of the four and with
the exception of the second-year learners (at thorn9 percent) did so more
strongly than the native French speakers at mean acceptance rates that were
11ndash15 percent lower)9 However they also accepted at much lower mean
rates (12 to 23 percent) the most felicitous positionndashinterpretation pairing
for the native French speakers pre-N position in a unique noun-referent
context Recall also that post-N position (7b) was semantically compatible
under either interpretation All learner groups accepted these two positionndash
interpretation pairings at much higher rates than the native French speakers
(thorn23 to thorn31 percent) including the least felicitous positionndashinterpretation
pairing for the native French speakersmdashpost-N position in a multiple noun-
referent context This is precisely what one would expect from the
pedagogical treatment these learners are provided and the congruity of
that treatment with classroom input which under-represents actual usage in
Table 2 Native French speaker mean acceptance rates for the uniquenonunique noun-referent distinction
Prenominal (AN)order ()
Postnominal (NA)order ()
Unique noun referent 74 52
Nonunique noun referent 49 37
Source Adapted from Anderson 2002 2007
300 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
text-based corpora and as we have now seen differs from native speaker
intuitions of overall (semantico-pragmatic) acceptability
Although a bias toward or even strict adherence to explicit pedagogical
rules in evaluating such sentences may not seem all that surprising what does
stand out in the data from Anderson (2002 2007) is that these differential
rates in acceptance between native and nonnative speakers are as true for the
advanced group in that studymdashconsisting of graduate student instructors and
second education teachersmdashas they are for the intermediate (second-year)
learner group This would seem to indicate no increased sensitivity to
contextual (pragmatic) factors in adjective position over the course of studying
French even though the advanced group had been exposed to much more
authentic input through texts written for and by native French speakers (ie
the descriptive norm uncovered in our examination of authentic text corpora)
DISCUSSION
What should be clear from the findings presented here is that rules of French
adjective placement that one typically finds in (at least American) textbooks
though not plainly incorrect are not fully descriptively accurate either at
least with respect to written French across several text genres In particular
such rules portray pre-N and post-N positions as mutually exclusive for
almost all adjectives when in fact they are not
Similar discrepancies are often found between rules for and by native
speakers and what those speakers actually do One need look no further than
the French for an example with colloquial speech often at odds with the
prescriptive bon usage and the admonishments of the Academie francaise One
might therefore also expect to find a discrepancy between textbook rules and
what teachers do in the classroom andor what students find to be
lsquoacceptablersquo in terms of word order As we have seen however this is not
the case These rules are faithfully reflected in classroom language use by
instructors to such an extent that when asked to evaluate the acceptability
of lsquoexpectedrsquo and lsquounexpected but nativelikersquo word orders (and one would
presume in production as well) both instructors and students paid much
less attention to problems in pragmatic implicature than did native speakers
a situation which does not appear to change even in the face of counter-
examples occasionally provided in the textbooks they use (see (6) above) and
most certainly in the written texts that they read at later course levels As
pointed out by an anonymous reviewer this situation can be considered a
testament to the power of classroom instruction (instructor input in
particular) if instruction is changed he or she reasons it is likely that
classroom learners will reflect this change
But should instruction be changed If so who along the continuum from
the theorist-researcher to the teacher-practitioner should inform and guide
this change At which level In what way Such questions are at the very
heart of two on-going debates that I re-examine before proposing some
BRUCE ANDERSON 301
answers the pedagogical versus descriptive norm debate (eg Owen 1993
Widdowson 2000 and the collection of articles in Gass et al 2002) and
the interrelated native-speaker-usage-as-goal debate in foreign language
pedagogy (eg the collection of articles in Blyth 2003)
On the whole the approach to instruction on French adjective
position outlined earlier appears advantageous in its simplicity providing
easily remembered rules of thumb that learners can apply in language
production That this approach to instruction does not as we have seen
straightforwardly follow the descriptive norm is a situation that linguists
utilizing a corpus-based methodology believe can and should be remedied
Conrad (2000) for instance sees in corpus-based research an immediate way
in which linguistics can inform language pedagogy freeing language teachers
from lsquohav[ing] to rely on judgments of grammatical accuracy or on native
speakersrsquo intuition about what sounds naturalrsquo (2000 555) Although the
subsequent study by Biber and Reppen (2002) uses a disclaimer to the effect
that frequency should not be the sole factor in material design they
nevertheless suggest that lsquoa selective revision of pedagogy to reflect actual
use as shown by frequency studies could result in radical changes that
facilitate the learning process for studentsrsquo (2002 199) Owen (1993)
however provides examples of how a total reliance on a corpus does not
necessarily yield better observation and more importantly questions
whether better observation automatically equates to better explanation
(1993 168) (but see Francis and Sinclair 1994 for rebuttal) For his part
Widdowson (2000) objects to the kind of equation just described as a
case of linguistics applied a mistaken belief that what is textually attested
lsquouniquely represents real language and that this reality should define the
foreign language subjectrsquo (2000 8) (but see Stubbs 2001 for rebuttal) What
is needed according to Widdowson is not simply linguistics applied
to pedagogy but a true role for applied linguistics an approach to usage that
also recognizes native speakersrsquo awareness of and opinions about such usage
The need for a truly applied linguistics perspective on native speaker
usage (including frequency patterns) is embodied in a pedagogical norm
originally conceptualized in Valdman (1989) and elaborated upon in
Bardovi-Harlig and Gass (2002) as
[A] combination of language systems and forms selected by linguistsand pedagogues to serve as the immediate language target ortargets that learners seek to acquire during their language studyIn other words pedagogical norms represent a mid-point or seriesof mid-points for learners as they progress toward acquiringnative language norms [thus constituting] a form of languagethat is acceptable to native speakers but easier to learn than the fullnative language system (p 3)
The textbook word-order rules for French adjective position reviewed
earlier would first appear to constitute an appropriate pedagogical norm
302 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
under Bardovi-Harlig and Gassrsquos (2002) definition because the learner is able
to produce a subset of the totality of possible utterances deemed acceptable to
native speakers which may turn out to be even more the case in spoken than
in written French (though the difference in mediums remains to be studied)
The learner data presented here from Anderson (2002 2007) however
provides one piece of evidence that textbook rules and classroom input do not
constitute a sufficient pedagogical norm as learners have stopped in their
progress toward acquiring native speaker (ie descriptively accurate) norms
As the articles in Blyth (2003) attest however it is highly debatable
whether the lsquoNative Speakerrsquo (as an idealized abstraction) should be the
appropriate model for the contemporary learner For Kramsch (2002 2003)
too much of a focus on an authentic native speaker norm reduces
foreign language (FL) study to a constant lsquoapproximation to and conformity
with stereotypical native speakers thus often silencing other FL-specific
forms of language potentialrsquo (2002 60) Language learning in instructional
settings on my understanding of Kramschrsquos (2003) argument should not be
viewed solely in a teleological sensemdashas language acquisition directed
toward a fixed nativelike end-pointmdashbut rather as language appropriation
by learners in the sense that they are constructing a linguistic and even
cultural identity lsquoin the interstices of national languages and on the
margins of monolingual speakersrsquo territoriesrsquo (2003 260) The political nature
of the debate as highlighted by Kramsch becomes all the more apparent
when one additionally considers the status of the foreign language on
the world stage (eg French as an international language in retreat)
and how deeply the normative attitudes of its speakers run (eg Francersquos
three-century-long history of acceptance and maintenance of the standard
(Lodge 1993))
The foregoing synopsis of the debates surrounding native speaker usage as
both the goal of foreign language instruction and the extent to which
pedagogical norms need apply has highlighted the real world complexities
involved in what would otherwise seem to be a simple equation Revise
pedagogy to reflect actual use as shown by frequency studies and this
will result in a more targetlike performance on the part of learners So we
return then to the questions posed earlier (1) Should instruction be
changed (2) If so who along the continuum from the theorist-researcher to
the teacher-practitioner should inform and guide this change (3) At which
level (4) In what ways
With respect to our first question because one can empirically
demonstrate that the pedagogical norm for French adjective position
prevents learners from eventually acquiring native speaker (ie descriptively
accurate) norms instruction should be changed so thatmdashirrespective of
whether the attainment of such a norm is necessarily desirablemdashlearners
have both a choice in selecting one grammatical or lexical form over the other
and an awareness of the meaning potential of each choice (following Kramsch
2002 71)
BRUCE ANDERSON 303
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
a number of works of French literature were read outside of class Note
that written input in these latter two course levels is qualitatively the same
as that used in the corpora compiled by Forsgren (1978) Wilmet (1980) and
Larsson (1993)
At issue during this period of investigation was classroom input that could
be seen andor heard by all students in the classroom (as well as the
researcher) Input from students involved in pair and group work was
therefore excluded from analysis Material to be read outside of class was
also excluded as there was no guarantee that all students had actually
read such material Although the totality of inputmdashincluding sources of input
the learner might have voluntarily sought out outside of classmdashcould
obviously never be exhaustively tallied the exclusion of certain forms of
input was first and foremost a matter of logistics (ie not having the
appropriate funds equipment facilities and time to sound record then
transcribe all classroom input from every participant in every situation of
use) Despite this drawback I see no reason to suspect that student-delivered
input to other students during pair- or group-work would be qualitatively
different (at least in terms of adjective position) from the input delivered by
students that was audible to the entire class this latter source of input
already constitutes 10ndash30 percent of all sources of input tallied as part of the
observation period
Each instance of adjective use during each class session was recorded with
pen and paper by the researcher A coding scheme was devised whereby an
11 17-inch sheet of paper was divided into quadrants labeled according to
the source of the input lsquoinstructorrsquo lsquostudentrsquo lsquoaudiorsquo and lsquowrittenrsquo Each
quadrant was then subdivided according to the position of adjectival
modification lsquopre-Nrsquo lsquopost-Nrsquo lsquopredicatersquo and lsquootherrsquo (to be described
below) Input of the form une bonne idee lsquoa good idearsquo by the instructor was
therefore recorded as an instance of the adjective bonne in the instructor
quadrant pre-N sub-quadrant Hatch marks next to the adjective bonne were
used for any further exemplars of this adjective (including its masculine form
bon) in this position from this source I acknowledge that mere human error
caused by fatigue etc may have resulted in tallies that are less accurate than
if each session were recorded on tape Though the results presented below
should be considered with some caution then occasional errors could not
have reached a significant enough level to detract from the highly salient
patterns of usage that we will find
The resulting sixty sheets each representing one 50-minute class session
were then tallied in order to determine (a) the relative contribution of the
four sources of input (b) the average token (exemplar) count of adjectives
per 50-minute class period (c) the ratio of pre-N to post-Npredicate tokens
(d) the type count (number of discrete lexemes) entering into pre-N and
post-N positions and (e) the frequency of variation in positionmdashthat is the
extent to which a given adjective appeared at least once in both pre-N and
post-N position during the ten hours of observation
BRUCE ANDERSON 295
Results
The instructor not surprisingly was the primary provider of input in the
classroom (See Figure 1 available to online subscribers at http
applijoxfordjournalsorg) This was especially the case in the first-year
courses (Levels 1a 1b) the second- and third-year courses (Levels 2a 2b 3)
showed a greater proportion of input from students and from audio aides
The instructor was by far the predominant source of input in the fourth-year
course (Level 4) given that it involved instructor-delivered lectures on
French civilization with minimal class participation on the part of students
Concerning the average number of times an adjective occurred in the
input during any one 50-minute class period we find a general increase with
level from 42 tokens at Level 1a to 156 tokens per class at Level 4 (See
Figure 2 available to online subscribers at httpapplijoxfordjournalsorg)
Students therefore heard an adjective used anywhere from one to three
times per minute on average
Aggregate data on the frequency with which adjectives appeared in pre-N
post-N and predicate positions or in other constructions (eg used in
isolation in elliptical constructions such as la charmante lsquothe charming [one]rsquo
and following indefinite pronouns as in quelqursquoun de charmant lsquosomeone
charmingrsquo) demonstrate that post-N and predicate positions were generally
equally frequent and together constitute fully 75 percent of the ratio whereas
pre-N position never surpassed 27 percent across course levels Other
constructions were minimal (See Figure 3 available to online subscribers at
httpapplijoxfordjournalsorg)
This roughly 75 percentndash25 percent ratio was then reanalyzed in order to
consider the number of distinct lexemes (lsquotypesrsquo) that make up the
percentages We find for example that only 15 lexemes make up the 14
percent of pre-N adjective use at Level 1a This number gradually increased
with course level but only reached a total of 41 by Level 4 Post-N position
on the other hand admitted a much larger variety of adjectives in classroom
input from 71 at Level 1a to 323 by Level 4 (See Figure 4 available to online
subscribers at httpapplijoxfordjournalsorg) Students therefore got
relatively little input on pre-N position and what input they did get was
made up of a relatively small class of adjectives being used over and over
again and which already figure in the textbook-determined list of pre-N
adjectives in (4) (See Appendix A available to online subscribers at http
applijoxfordjournalsorg for a list of adjectives appearing in pre-N position
by course level)
Finally we find that very few adjectives varied in position during the 10
hours of in-class observation There was no variation attested at Level 1a
and only four to seven adjectives varied in position from Level 1b through to
Level 3 Moreover some of the input provided to learners in these courses
in the form of other studentsrsquo output was nonnativelike6 At the most
advanced undergraduate level learners got somewhat more input involving
296 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
variation in position where 20 adjectives did so (See Figure 5 available to
online subscribers at httpapplijoxfordjournalsorg) Again most (but not
all) of these adjectives already figure into the textbook-determined list of
variable adjectives in (5) (See Appendix B available to online subscribers at
httpapplijoxfordjournalsorg for a list of adjectives varying in position by
course level)
To summarize the observational data just presented the ambient language
of the classroom clearly mimics the pedagogical treatment that American
university-level classroom learners receive from textbooks Although
adjectives are in general a very frequent part of the input the rate of
pre-N position (at roughly 25 percent) is essentially static across levels
pointing to the fact that classroom input across levels is in no way
contradictory to what learners are initially taught Post-N position is basic
(unmarked) and pre-N position is exceptional (marked) The adjectives that
do appear in pre-N position in classroom input are in most cases those that
learners already expect to appear in that position (cf the list in (4) and
Appendix A) The few adjectives that appear in both positions are again in
most cases those that learners expect to do so (cf the list in (5) and
Appendix B)
By the end of the first-year level then both instruction and classroom
input would lead learners to expect that adjectives will appear in post-N
position with the notable exception of the class of adjectives in (4) (those in
(5) having not yet been introduced) By the second-year level it could be
argued that instruction provides examples in the form of so-called change of
meaning adjectives that variation in position is to be associated with
variation in interpretation (and perhaps by implication always to
interpretation) However learnersrsquo explicit knowledge of interpretative
differences as exemplified by the adjectives in (5) typically consists of
arbitrary and unsystematic English glosses By arbitrary I mean that
instruction never provides an adequate explanation for why the pre-N
position of cher should correspond to beloved or for its post-N position to
correspond to expensive7 Though telling students following Bragger and Rice
(1996) that the meaning in post-N position is lsquoconcretersquo and the meaning in
pre-N position is lsquofigurativersquo might (somehow) help it does not account for
other adjectives in (5) like prochain lsquonextrsquo This latter adjective also
demonstrates the fact that English glosses are unsystematic in that prochain
translates as lsquonextrsquo in either position
Before considering our final research question I note that no research on
adjective position in native French speaker discourse is available to date
An anonymous Applied Linguistics reviewer questions the validity of compar-
ing primarily spoken classroom discoursemdashor any other kind of spoken
discoursemdashto written texts without first demonstrating that adjective position
is insensitive to this difference in medium While recognizing the potentially
important distinction in medium and the need for further research in this
area the validity of comparing classroom input to written texts in the present
BRUCE ANDERSON 297
study lies in the fact that such a comparison roughly corresponds to the
curriculum of typical university-level courses an emphasis on listening
comprehension and oral communication in the first- and second-year
language courses leading to an emphasis on reading comprehension
(typically literature-based) and proficiency in writing in the third- and
fourth-year courses Were it to turn out that the degree of variation in
adjective position found in written texts is restricted to just that medium
revisions to pedagogy would necessarily be restricted to writing-intensive
(composition creative composition stylistics) courses at the third- and
fourth-year levels (I return to this issue in the discussion section)
LEARNER INTUITIONS OF ACCEPTABILITY
Thus far we have seen an incongruity between textbook presentations of
adjective position and authentic text sources across a number of written text
genres (research question 1) and that ambient language use involving
adjective position in the classroom is clearly more congruent with and might
therefore be taken to reinforce the explicit rules found in textbooks
(research question 2) We are now in a position to address our third research
questionmdashthe extent to which intuitions of acceptability among classroom
learners are biased by the congruity between pedagogical treatment and
classroom input frequency patterns An empirical study previously reported
in Anderson (2002 2007) can shed some light on this question
Methodology
Anderson (2002 2007) investigated the acquisition of various morphosyn-
tactic properties of French noun phrases by administering an acceptability
judgment task in which participants had to evaluate the appropriateness of
sentences featuring among other things the pre- versus post-N position of
adjectives Each of the adjectives selected were lsquohighly variablersquo with respect
to position (cf Table 1) yet crucially do not figure in textbook lists of
adjectives said to change meaning based on position (cf the list in (10))
These include lourd lsquoheavyrsquo violent lsquoviolentrsquo charmant lsquocharmingrsquo precieux
lsquopreciousrsquo epais lsquothickrsquo etrange lsquostrangersquo and delicieux lsquodeliciousrsquo (See
Appendix C available to online subscribers at httpapplijoxfordjournals
org for test sentences) Study participants included English-speaking learners
of French in the second third and fourth years of undergraduate study
(nfrac14 80) who were attending courses at the same institution at which the
observational data were collected a group of advanced learners (nfrac14 20)
consisting of graduate student instructors at that same institution as well as
secondary education teachers and a group of native French speakers (nfrac14 27)
roughly matched in age and education level
Each adjective appeared in test sentences once prenominally and once
postnominally with each test sentence preceded once by a context
298 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
motivating acceptance and once by a context motivating rejection of one or
the other adjective order for a total of four positionndashinterpretation pairings
per adjective These were randomly distributed across two testing
instruments administered a week apart After reading each context and
paired test sentence study participants were directed to check boxes
labeled lsquofinersquo lsquooddrsquo or lsquocannot decidersquo to the prompt This particular sentence
sounds ___ in this context
As an example each of the sentences in (7) featuring the adjective lourd
lsquoheavyrsquo was paired with each of two contexts
(7) a Pierre doit laisser la lourde valise avec son amib Pierre doit laisser la valise lourde a lrsquoaeroport
lsquoPierre has to leave the heavy suitcase with his friendatthe airportrsquo
In the first context motivating a unique noun-referent interpretation
there was one and only one suitcase which happened to be heavy and
which lsquoPierrersquo had to leave behind with his friend In the second context
motivating a multiple noun-referent interpretation there were a number of
suitcases only one of which was heavy and it was this one that lsquoPierrersquo had
to leave behind at the airport (See Appendix D available to online
subscribers at httpapplijoxfordjournalsorg for the complete version of
each context)
When viewed solely in terms of truth-conditional semantics each of the
four positionndashinterpretation pairings is compatible save for one pre-N
position (7a) in a multiple noun-referent context This is because the singular
definite article la and pre-N adjective lourde together presuppose a unique
suitcase in context (which happens to be heavy) This presupposition
does not hold when the adjective is in post-N position where all that is
required is that there be at least one such suitcase in context The testing
instrument then allows us to examine how classroom learners react to
an lsquoexpectedrsquo word order (ie sentences like (7b) in either context) as well
as to an lsquounexpectedrsquo but semantically and pragmatically appropriate
word order (ie sentences like (7a) but only in the unique noun-referent
context)
Results
As the mean acceptance rates from native French speakers in Table 2 show
truth-conditional semantics was not the only factor determining acceptance
rates Two of the pairings received essentially chance-level rates of
acceptance (pre-N position in a multiple noun-referent context at 49
percent and post-N position in a unique context at 52 percent) whereas the
other two pairings showed much clearer above- or below-chance rates (pre-N
position in a unique noun-referent context at 74 percent and post-N
position in a multiple noun-referent context at 37 percent)
BRUCE ANDERSON 299
As was argued in Anderson (2002 2007) results for each of the four
positionndashinterpretation pairings among the native French speakers are
explainable in terms of a disproportionate interaction between semantics and
pragmatic implicature in particular the Gricean maxim of quantity An
utterance should be no more and no less informative than is required for the
current purposes of the exchange (Grice 1975) The positionndashinterpretation
pairing that was lsquotruersquo in terms of semantics and also the most pragmatically
appropriate was accepted at the highest rate (74 percent) the pairing that
was also lsquotruersquo in terms of semantics but was pragmatically the least
appropriate was accepted at the lowest rate (37 percent)8
Do learner group response patterns show the same bias toward pragmatic
implicature in their acceptance rates The comparison of mean acceptance
rates between native French speakers and learner groups in Table 3 shows
that learners performed in a particular (and particularly non-Frenchlike) way
in which the interaction goes in the opposite direction favoring semantics
over pragmatics
Recall that the only positionndashinterpretation pairing that was semantically
incompatible was pre-N position (7a) in a multiple noun-referent context
given the presupposition of uniqueness of the noun referent All learner
groups did indeed accept this pairing at the lowest rate of the four and with
the exception of the second-year learners (at thorn9 percent) did so more
strongly than the native French speakers at mean acceptance rates that were
11ndash15 percent lower)9 However they also accepted at much lower mean
rates (12 to 23 percent) the most felicitous positionndashinterpretation pairing
for the native French speakers pre-N position in a unique noun-referent
context Recall also that post-N position (7b) was semantically compatible
under either interpretation All learner groups accepted these two positionndash
interpretation pairings at much higher rates than the native French speakers
(thorn23 to thorn31 percent) including the least felicitous positionndashinterpretation
pairing for the native French speakersmdashpost-N position in a multiple noun-
referent context This is precisely what one would expect from the
pedagogical treatment these learners are provided and the congruity of
that treatment with classroom input which under-represents actual usage in
Table 2 Native French speaker mean acceptance rates for the uniquenonunique noun-referent distinction
Prenominal (AN)order ()
Postnominal (NA)order ()
Unique noun referent 74 52
Nonunique noun referent 49 37
Source Adapted from Anderson 2002 2007
300 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
text-based corpora and as we have now seen differs from native speaker
intuitions of overall (semantico-pragmatic) acceptability
Although a bias toward or even strict adherence to explicit pedagogical
rules in evaluating such sentences may not seem all that surprising what does
stand out in the data from Anderson (2002 2007) is that these differential
rates in acceptance between native and nonnative speakers are as true for the
advanced group in that studymdashconsisting of graduate student instructors and
second education teachersmdashas they are for the intermediate (second-year)
learner group This would seem to indicate no increased sensitivity to
contextual (pragmatic) factors in adjective position over the course of studying
French even though the advanced group had been exposed to much more
authentic input through texts written for and by native French speakers (ie
the descriptive norm uncovered in our examination of authentic text corpora)
DISCUSSION
What should be clear from the findings presented here is that rules of French
adjective placement that one typically finds in (at least American) textbooks
though not plainly incorrect are not fully descriptively accurate either at
least with respect to written French across several text genres In particular
such rules portray pre-N and post-N positions as mutually exclusive for
almost all adjectives when in fact they are not
Similar discrepancies are often found between rules for and by native
speakers and what those speakers actually do One need look no further than
the French for an example with colloquial speech often at odds with the
prescriptive bon usage and the admonishments of the Academie francaise One
might therefore also expect to find a discrepancy between textbook rules and
what teachers do in the classroom andor what students find to be
lsquoacceptablersquo in terms of word order As we have seen however this is not
the case These rules are faithfully reflected in classroom language use by
instructors to such an extent that when asked to evaluate the acceptability
of lsquoexpectedrsquo and lsquounexpected but nativelikersquo word orders (and one would
presume in production as well) both instructors and students paid much
less attention to problems in pragmatic implicature than did native speakers
a situation which does not appear to change even in the face of counter-
examples occasionally provided in the textbooks they use (see (6) above) and
most certainly in the written texts that they read at later course levels As
pointed out by an anonymous reviewer this situation can be considered a
testament to the power of classroom instruction (instructor input in
particular) if instruction is changed he or she reasons it is likely that
classroom learners will reflect this change
But should instruction be changed If so who along the continuum from
the theorist-researcher to the teacher-practitioner should inform and guide
this change At which level In what way Such questions are at the very
heart of two on-going debates that I re-examine before proposing some
BRUCE ANDERSON 301
answers the pedagogical versus descriptive norm debate (eg Owen 1993
Widdowson 2000 and the collection of articles in Gass et al 2002) and
the interrelated native-speaker-usage-as-goal debate in foreign language
pedagogy (eg the collection of articles in Blyth 2003)
On the whole the approach to instruction on French adjective
position outlined earlier appears advantageous in its simplicity providing
easily remembered rules of thumb that learners can apply in language
production That this approach to instruction does not as we have seen
straightforwardly follow the descriptive norm is a situation that linguists
utilizing a corpus-based methodology believe can and should be remedied
Conrad (2000) for instance sees in corpus-based research an immediate way
in which linguistics can inform language pedagogy freeing language teachers
from lsquohav[ing] to rely on judgments of grammatical accuracy or on native
speakersrsquo intuition about what sounds naturalrsquo (2000 555) Although the
subsequent study by Biber and Reppen (2002) uses a disclaimer to the effect
that frequency should not be the sole factor in material design they
nevertheless suggest that lsquoa selective revision of pedagogy to reflect actual
use as shown by frequency studies could result in radical changes that
facilitate the learning process for studentsrsquo (2002 199) Owen (1993)
however provides examples of how a total reliance on a corpus does not
necessarily yield better observation and more importantly questions
whether better observation automatically equates to better explanation
(1993 168) (but see Francis and Sinclair 1994 for rebuttal) For his part
Widdowson (2000) objects to the kind of equation just described as a
case of linguistics applied a mistaken belief that what is textually attested
lsquouniquely represents real language and that this reality should define the
foreign language subjectrsquo (2000 8) (but see Stubbs 2001 for rebuttal) What
is needed according to Widdowson is not simply linguistics applied
to pedagogy but a true role for applied linguistics an approach to usage that
also recognizes native speakersrsquo awareness of and opinions about such usage
The need for a truly applied linguistics perspective on native speaker
usage (including frequency patterns) is embodied in a pedagogical norm
originally conceptualized in Valdman (1989) and elaborated upon in
Bardovi-Harlig and Gass (2002) as
[A] combination of language systems and forms selected by linguistsand pedagogues to serve as the immediate language target ortargets that learners seek to acquire during their language studyIn other words pedagogical norms represent a mid-point or seriesof mid-points for learners as they progress toward acquiringnative language norms [thus constituting] a form of languagethat is acceptable to native speakers but easier to learn than the fullnative language system (p 3)
The textbook word-order rules for French adjective position reviewed
earlier would first appear to constitute an appropriate pedagogical norm
302 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
under Bardovi-Harlig and Gassrsquos (2002) definition because the learner is able
to produce a subset of the totality of possible utterances deemed acceptable to
native speakers which may turn out to be even more the case in spoken than
in written French (though the difference in mediums remains to be studied)
The learner data presented here from Anderson (2002 2007) however
provides one piece of evidence that textbook rules and classroom input do not
constitute a sufficient pedagogical norm as learners have stopped in their
progress toward acquiring native speaker (ie descriptively accurate) norms
As the articles in Blyth (2003) attest however it is highly debatable
whether the lsquoNative Speakerrsquo (as an idealized abstraction) should be the
appropriate model for the contemporary learner For Kramsch (2002 2003)
too much of a focus on an authentic native speaker norm reduces
foreign language (FL) study to a constant lsquoapproximation to and conformity
with stereotypical native speakers thus often silencing other FL-specific
forms of language potentialrsquo (2002 60) Language learning in instructional
settings on my understanding of Kramschrsquos (2003) argument should not be
viewed solely in a teleological sensemdashas language acquisition directed
toward a fixed nativelike end-pointmdashbut rather as language appropriation
by learners in the sense that they are constructing a linguistic and even
cultural identity lsquoin the interstices of national languages and on the
margins of monolingual speakersrsquo territoriesrsquo (2003 260) The political nature
of the debate as highlighted by Kramsch becomes all the more apparent
when one additionally considers the status of the foreign language on
the world stage (eg French as an international language in retreat)
and how deeply the normative attitudes of its speakers run (eg Francersquos
three-century-long history of acceptance and maintenance of the standard
(Lodge 1993))
The foregoing synopsis of the debates surrounding native speaker usage as
both the goal of foreign language instruction and the extent to which
pedagogical norms need apply has highlighted the real world complexities
involved in what would otherwise seem to be a simple equation Revise
pedagogy to reflect actual use as shown by frequency studies and this
will result in a more targetlike performance on the part of learners So we
return then to the questions posed earlier (1) Should instruction be
changed (2) If so who along the continuum from the theorist-researcher to
the teacher-practitioner should inform and guide this change (3) At which
level (4) In what ways
With respect to our first question because one can empirically
demonstrate that the pedagogical norm for French adjective position
prevents learners from eventually acquiring native speaker (ie descriptively
accurate) norms instruction should be changed so thatmdashirrespective of
whether the attainment of such a norm is necessarily desirablemdashlearners
have both a choice in selecting one grammatical or lexical form over the other
and an awareness of the meaning potential of each choice (following Kramsch
2002 71)
BRUCE ANDERSON 303
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
Results
The instructor not surprisingly was the primary provider of input in the
classroom (See Figure 1 available to online subscribers at http
applijoxfordjournalsorg) This was especially the case in the first-year
courses (Levels 1a 1b) the second- and third-year courses (Levels 2a 2b 3)
showed a greater proportion of input from students and from audio aides
The instructor was by far the predominant source of input in the fourth-year
course (Level 4) given that it involved instructor-delivered lectures on
French civilization with minimal class participation on the part of students
Concerning the average number of times an adjective occurred in the
input during any one 50-minute class period we find a general increase with
level from 42 tokens at Level 1a to 156 tokens per class at Level 4 (See
Figure 2 available to online subscribers at httpapplijoxfordjournalsorg)
Students therefore heard an adjective used anywhere from one to three
times per minute on average
Aggregate data on the frequency with which adjectives appeared in pre-N
post-N and predicate positions or in other constructions (eg used in
isolation in elliptical constructions such as la charmante lsquothe charming [one]rsquo
and following indefinite pronouns as in quelqursquoun de charmant lsquosomeone
charmingrsquo) demonstrate that post-N and predicate positions were generally
equally frequent and together constitute fully 75 percent of the ratio whereas
pre-N position never surpassed 27 percent across course levels Other
constructions were minimal (See Figure 3 available to online subscribers at
httpapplijoxfordjournalsorg)
This roughly 75 percentndash25 percent ratio was then reanalyzed in order to
consider the number of distinct lexemes (lsquotypesrsquo) that make up the
percentages We find for example that only 15 lexemes make up the 14
percent of pre-N adjective use at Level 1a This number gradually increased
with course level but only reached a total of 41 by Level 4 Post-N position
on the other hand admitted a much larger variety of adjectives in classroom
input from 71 at Level 1a to 323 by Level 4 (See Figure 4 available to online
subscribers at httpapplijoxfordjournalsorg) Students therefore got
relatively little input on pre-N position and what input they did get was
made up of a relatively small class of adjectives being used over and over
again and which already figure in the textbook-determined list of pre-N
adjectives in (4) (See Appendix A available to online subscribers at http
applijoxfordjournalsorg for a list of adjectives appearing in pre-N position
by course level)
Finally we find that very few adjectives varied in position during the 10
hours of in-class observation There was no variation attested at Level 1a
and only four to seven adjectives varied in position from Level 1b through to
Level 3 Moreover some of the input provided to learners in these courses
in the form of other studentsrsquo output was nonnativelike6 At the most
advanced undergraduate level learners got somewhat more input involving
296 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
variation in position where 20 adjectives did so (See Figure 5 available to
online subscribers at httpapplijoxfordjournalsorg) Again most (but not
all) of these adjectives already figure into the textbook-determined list of
variable adjectives in (5) (See Appendix B available to online subscribers at
httpapplijoxfordjournalsorg for a list of adjectives varying in position by
course level)
To summarize the observational data just presented the ambient language
of the classroom clearly mimics the pedagogical treatment that American
university-level classroom learners receive from textbooks Although
adjectives are in general a very frequent part of the input the rate of
pre-N position (at roughly 25 percent) is essentially static across levels
pointing to the fact that classroom input across levels is in no way
contradictory to what learners are initially taught Post-N position is basic
(unmarked) and pre-N position is exceptional (marked) The adjectives that
do appear in pre-N position in classroom input are in most cases those that
learners already expect to appear in that position (cf the list in (4) and
Appendix A) The few adjectives that appear in both positions are again in
most cases those that learners expect to do so (cf the list in (5) and
Appendix B)
By the end of the first-year level then both instruction and classroom
input would lead learners to expect that adjectives will appear in post-N
position with the notable exception of the class of adjectives in (4) (those in
(5) having not yet been introduced) By the second-year level it could be
argued that instruction provides examples in the form of so-called change of
meaning adjectives that variation in position is to be associated with
variation in interpretation (and perhaps by implication always to
interpretation) However learnersrsquo explicit knowledge of interpretative
differences as exemplified by the adjectives in (5) typically consists of
arbitrary and unsystematic English glosses By arbitrary I mean that
instruction never provides an adequate explanation for why the pre-N
position of cher should correspond to beloved or for its post-N position to
correspond to expensive7 Though telling students following Bragger and Rice
(1996) that the meaning in post-N position is lsquoconcretersquo and the meaning in
pre-N position is lsquofigurativersquo might (somehow) help it does not account for
other adjectives in (5) like prochain lsquonextrsquo This latter adjective also
demonstrates the fact that English glosses are unsystematic in that prochain
translates as lsquonextrsquo in either position
Before considering our final research question I note that no research on
adjective position in native French speaker discourse is available to date
An anonymous Applied Linguistics reviewer questions the validity of compar-
ing primarily spoken classroom discoursemdashor any other kind of spoken
discoursemdashto written texts without first demonstrating that adjective position
is insensitive to this difference in medium While recognizing the potentially
important distinction in medium and the need for further research in this
area the validity of comparing classroom input to written texts in the present
BRUCE ANDERSON 297
study lies in the fact that such a comparison roughly corresponds to the
curriculum of typical university-level courses an emphasis on listening
comprehension and oral communication in the first- and second-year
language courses leading to an emphasis on reading comprehension
(typically literature-based) and proficiency in writing in the third- and
fourth-year courses Were it to turn out that the degree of variation in
adjective position found in written texts is restricted to just that medium
revisions to pedagogy would necessarily be restricted to writing-intensive
(composition creative composition stylistics) courses at the third- and
fourth-year levels (I return to this issue in the discussion section)
LEARNER INTUITIONS OF ACCEPTABILITY
Thus far we have seen an incongruity between textbook presentations of
adjective position and authentic text sources across a number of written text
genres (research question 1) and that ambient language use involving
adjective position in the classroom is clearly more congruent with and might
therefore be taken to reinforce the explicit rules found in textbooks
(research question 2) We are now in a position to address our third research
questionmdashthe extent to which intuitions of acceptability among classroom
learners are biased by the congruity between pedagogical treatment and
classroom input frequency patterns An empirical study previously reported
in Anderson (2002 2007) can shed some light on this question
Methodology
Anderson (2002 2007) investigated the acquisition of various morphosyn-
tactic properties of French noun phrases by administering an acceptability
judgment task in which participants had to evaluate the appropriateness of
sentences featuring among other things the pre- versus post-N position of
adjectives Each of the adjectives selected were lsquohighly variablersquo with respect
to position (cf Table 1) yet crucially do not figure in textbook lists of
adjectives said to change meaning based on position (cf the list in (10))
These include lourd lsquoheavyrsquo violent lsquoviolentrsquo charmant lsquocharmingrsquo precieux
lsquopreciousrsquo epais lsquothickrsquo etrange lsquostrangersquo and delicieux lsquodeliciousrsquo (See
Appendix C available to online subscribers at httpapplijoxfordjournals
org for test sentences) Study participants included English-speaking learners
of French in the second third and fourth years of undergraduate study
(nfrac14 80) who were attending courses at the same institution at which the
observational data were collected a group of advanced learners (nfrac14 20)
consisting of graduate student instructors at that same institution as well as
secondary education teachers and a group of native French speakers (nfrac14 27)
roughly matched in age and education level
Each adjective appeared in test sentences once prenominally and once
postnominally with each test sentence preceded once by a context
298 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
motivating acceptance and once by a context motivating rejection of one or
the other adjective order for a total of four positionndashinterpretation pairings
per adjective These were randomly distributed across two testing
instruments administered a week apart After reading each context and
paired test sentence study participants were directed to check boxes
labeled lsquofinersquo lsquooddrsquo or lsquocannot decidersquo to the prompt This particular sentence
sounds ___ in this context
As an example each of the sentences in (7) featuring the adjective lourd
lsquoheavyrsquo was paired with each of two contexts
(7) a Pierre doit laisser la lourde valise avec son amib Pierre doit laisser la valise lourde a lrsquoaeroport
lsquoPierre has to leave the heavy suitcase with his friendatthe airportrsquo
In the first context motivating a unique noun-referent interpretation
there was one and only one suitcase which happened to be heavy and
which lsquoPierrersquo had to leave behind with his friend In the second context
motivating a multiple noun-referent interpretation there were a number of
suitcases only one of which was heavy and it was this one that lsquoPierrersquo had
to leave behind at the airport (See Appendix D available to online
subscribers at httpapplijoxfordjournalsorg for the complete version of
each context)
When viewed solely in terms of truth-conditional semantics each of the
four positionndashinterpretation pairings is compatible save for one pre-N
position (7a) in a multiple noun-referent context This is because the singular
definite article la and pre-N adjective lourde together presuppose a unique
suitcase in context (which happens to be heavy) This presupposition
does not hold when the adjective is in post-N position where all that is
required is that there be at least one such suitcase in context The testing
instrument then allows us to examine how classroom learners react to
an lsquoexpectedrsquo word order (ie sentences like (7b) in either context) as well
as to an lsquounexpectedrsquo but semantically and pragmatically appropriate
word order (ie sentences like (7a) but only in the unique noun-referent
context)
Results
As the mean acceptance rates from native French speakers in Table 2 show
truth-conditional semantics was not the only factor determining acceptance
rates Two of the pairings received essentially chance-level rates of
acceptance (pre-N position in a multiple noun-referent context at 49
percent and post-N position in a unique context at 52 percent) whereas the
other two pairings showed much clearer above- or below-chance rates (pre-N
position in a unique noun-referent context at 74 percent and post-N
position in a multiple noun-referent context at 37 percent)
BRUCE ANDERSON 299
As was argued in Anderson (2002 2007) results for each of the four
positionndashinterpretation pairings among the native French speakers are
explainable in terms of a disproportionate interaction between semantics and
pragmatic implicature in particular the Gricean maxim of quantity An
utterance should be no more and no less informative than is required for the
current purposes of the exchange (Grice 1975) The positionndashinterpretation
pairing that was lsquotruersquo in terms of semantics and also the most pragmatically
appropriate was accepted at the highest rate (74 percent) the pairing that
was also lsquotruersquo in terms of semantics but was pragmatically the least
appropriate was accepted at the lowest rate (37 percent)8
Do learner group response patterns show the same bias toward pragmatic
implicature in their acceptance rates The comparison of mean acceptance
rates between native French speakers and learner groups in Table 3 shows
that learners performed in a particular (and particularly non-Frenchlike) way
in which the interaction goes in the opposite direction favoring semantics
over pragmatics
Recall that the only positionndashinterpretation pairing that was semantically
incompatible was pre-N position (7a) in a multiple noun-referent context
given the presupposition of uniqueness of the noun referent All learner
groups did indeed accept this pairing at the lowest rate of the four and with
the exception of the second-year learners (at thorn9 percent) did so more
strongly than the native French speakers at mean acceptance rates that were
11ndash15 percent lower)9 However they also accepted at much lower mean
rates (12 to 23 percent) the most felicitous positionndashinterpretation pairing
for the native French speakers pre-N position in a unique noun-referent
context Recall also that post-N position (7b) was semantically compatible
under either interpretation All learner groups accepted these two positionndash
interpretation pairings at much higher rates than the native French speakers
(thorn23 to thorn31 percent) including the least felicitous positionndashinterpretation
pairing for the native French speakersmdashpost-N position in a multiple noun-
referent context This is precisely what one would expect from the
pedagogical treatment these learners are provided and the congruity of
that treatment with classroom input which under-represents actual usage in
Table 2 Native French speaker mean acceptance rates for the uniquenonunique noun-referent distinction
Prenominal (AN)order ()
Postnominal (NA)order ()
Unique noun referent 74 52
Nonunique noun referent 49 37
Source Adapted from Anderson 2002 2007
300 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
text-based corpora and as we have now seen differs from native speaker
intuitions of overall (semantico-pragmatic) acceptability
Although a bias toward or even strict adherence to explicit pedagogical
rules in evaluating such sentences may not seem all that surprising what does
stand out in the data from Anderson (2002 2007) is that these differential
rates in acceptance between native and nonnative speakers are as true for the
advanced group in that studymdashconsisting of graduate student instructors and
second education teachersmdashas they are for the intermediate (second-year)
learner group This would seem to indicate no increased sensitivity to
contextual (pragmatic) factors in adjective position over the course of studying
French even though the advanced group had been exposed to much more
authentic input through texts written for and by native French speakers (ie
the descriptive norm uncovered in our examination of authentic text corpora)
DISCUSSION
What should be clear from the findings presented here is that rules of French
adjective placement that one typically finds in (at least American) textbooks
though not plainly incorrect are not fully descriptively accurate either at
least with respect to written French across several text genres In particular
such rules portray pre-N and post-N positions as mutually exclusive for
almost all adjectives when in fact they are not
Similar discrepancies are often found between rules for and by native
speakers and what those speakers actually do One need look no further than
the French for an example with colloquial speech often at odds with the
prescriptive bon usage and the admonishments of the Academie francaise One
might therefore also expect to find a discrepancy between textbook rules and
what teachers do in the classroom andor what students find to be
lsquoacceptablersquo in terms of word order As we have seen however this is not
the case These rules are faithfully reflected in classroom language use by
instructors to such an extent that when asked to evaluate the acceptability
of lsquoexpectedrsquo and lsquounexpected but nativelikersquo word orders (and one would
presume in production as well) both instructors and students paid much
less attention to problems in pragmatic implicature than did native speakers
a situation which does not appear to change even in the face of counter-
examples occasionally provided in the textbooks they use (see (6) above) and
most certainly in the written texts that they read at later course levels As
pointed out by an anonymous reviewer this situation can be considered a
testament to the power of classroom instruction (instructor input in
particular) if instruction is changed he or she reasons it is likely that
classroom learners will reflect this change
But should instruction be changed If so who along the continuum from
the theorist-researcher to the teacher-practitioner should inform and guide
this change At which level In what way Such questions are at the very
heart of two on-going debates that I re-examine before proposing some
BRUCE ANDERSON 301
answers the pedagogical versus descriptive norm debate (eg Owen 1993
Widdowson 2000 and the collection of articles in Gass et al 2002) and
the interrelated native-speaker-usage-as-goal debate in foreign language
pedagogy (eg the collection of articles in Blyth 2003)
On the whole the approach to instruction on French adjective
position outlined earlier appears advantageous in its simplicity providing
easily remembered rules of thumb that learners can apply in language
production That this approach to instruction does not as we have seen
straightforwardly follow the descriptive norm is a situation that linguists
utilizing a corpus-based methodology believe can and should be remedied
Conrad (2000) for instance sees in corpus-based research an immediate way
in which linguistics can inform language pedagogy freeing language teachers
from lsquohav[ing] to rely on judgments of grammatical accuracy or on native
speakersrsquo intuition about what sounds naturalrsquo (2000 555) Although the
subsequent study by Biber and Reppen (2002) uses a disclaimer to the effect
that frequency should not be the sole factor in material design they
nevertheless suggest that lsquoa selective revision of pedagogy to reflect actual
use as shown by frequency studies could result in radical changes that
facilitate the learning process for studentsrsquo (2002 199) Owen (1993)
however provides examples of how a total reliance on a corpus does not
necessarily yield better observation and more importantly questions
whether better observation automatically equates to better explanation
(1993 168) (but see Francis and Sinclair 1994 for rebuttal) For his part
Widdowson (2000) objects to the kind of equation just described as a
case of linguistics applied a mistaken belief that what is textually attested
lsquouniquely represents real language and that this reality should define the
foreign language subjectrsquo (2000 8) (but see Stubbs 2001 for rebuttal) What
is needed according to Widdowson is not simply linguistics applied
to pedagogy but a true role for applied linguistics an approach to usage that
also recognizes native speakersrsquo awareness of and opinions about such usage
The need for a truly applied linguistics perspective on native speaker
usage (including frequency patterns) is embodied in a pedagogical norm
originally conceptualized in Valdman (1989) and elaborated upon in
Bardovi-Harlig and Gass (2002) as
[A] combination of language systems and forms selected by linguistsand pedagogues to serve as the immediate language target ortargets that learners seek to acquire during their language studyIn other words pedagogical norms represent a mid-point or seriesof mid-points for learners as they progress toward acquiringnative language norms [thus constituting] a form of languagethat is acceptable to native speakers but easier to learn than the fullnative language system (p 3)
The textbook word-order rules for French adjective position reviewed
earlier would first appear to constitute an appropriate pedagogical norm
302 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
under Bardovi-Harlig and Gassrsquos (2002) definition because the learner is able
to produce a subset of the totality of possible utterances deemed acceptable to
native speakers which may turn out to be even more the case in spoken than
in written French (though the difference in mediums remains to be studied)
The learner data presented here from Anderson (2002 2007) however
provides one piece of evidence that textbook rules and classroom input do not
constitute a sufficient pedagogical norm as learners have stopped in their
progress toward acquiring native speaker (ie descriptively accurate) norms
As the articles in Blyth (2003) attest however it is highly debatable
whether the lsquoNative Speakerrsquo (as an idealized abstraction) should be the
appropriate model for the contemporary learner For Kramsch (2002 2003)
too much of a focus on an authentic native speaker norm reduces
foreign language (FL) study to a constant lsquoapproximation to and conformity
with stereotypical native speakers thus often silencing other FL-specific
forms of language potentialrsquo (2002 60) Language learning in instructional
settings on my understanding of Kramschrsquos (2003) argument should not be
viewed solely in a teleological sensemdashas language acquisition directed
toward a fixed nativelike end-pointmdashbut rather as language appropriation
by learners in the sense that they are constructing a linguistic and even
cultural identity lsquoin the interstices of national languages and on the
margins of monolingual speakersrsquo territoriesrsquo (2003 260) The political nature
of the debate as highlighted by Kramsch becomes all the more apparent
when one additionally considers the status of the foreign language on
the world stage (eg French as an international language in retreat)
and how deeply the normative attitudes of its speakers run (eg Francersquos
three-century-long history of acceptance and maintenance of the standard
(Lodge 1993))
The foregoing synopsis of the debates surrounding native speaker usage as
both the goal of foreign language instruction and the extent to which
pedagogical norms need apply has highlighted the real world complexities
involved in what would otherwise seem to be a simple equation Revise
pedagogy to reflect actual use as shown by frequency studies and this
will result in a more targetlike performance on the part of learners So we
return then to the questions posed earlier (1) Should instruction be
changed (2) If so who along the continuum from the theorist-researcher to
the teacher-practitioner should inform and guide this change (3) At which
level (4) In what ways
With respect to our first question because one can empirically
demonstrate that the pedagogical norm for French adjective position
prevents learners from eventually acquiring native speaker (ie descriptively
accurate) norms instruction should be changed so thatmdashirrespective of
whether the attainment of such a norm is necessarily desirablemdashlearners
have both a choice in selecting one grammatical or lexical form over the other
and an awareness of the meaning potential of each choice (following Kramsch
2002 71)
BRUCE ANDERSON 303
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
variation in position where 20 adjectives did so (See Figure 5 available to
online subscribers at httpapplijoxfordjournalsorg) Again most (but not
all) of these adjectives already figure into the textbook-determined list of
variable adjectives in (5) (See Appendix B available to online subscribers at
httpapplijoxfordjournalsorg for a list of adjectives varying in position by
course level)
To summarize the observational data just presented the ambient language
of the classroom clearly mimics the pedagogical treatment that American
university-level classroom learners receive from textbooks Although
adjectives are in general a very frequent part of the input the rate of
pre-N position (at roughly 25 percent) is essentially static across levels
pointing to the fact that classroom input across levels is in no way
contradictory to what learners are initially taught Post-N position is basic
(unmarked) and pre-N position is exceptional (marked) The adjectives that
do appear in pre-N position in classroom input are in most cases those that
learners already expect to appear in that position (cf the list in (4) and
Appendix A) The few adjectives that appear in both positions are again in
most cases those that learners expect to do so (cf the list in (5) and
Appendix B)
By the end of the first-year level then both instruction and classroom
input would lead learners to expect that adjectives will appear in post-N
position with the notable exception of the class of adjectives in (4) (those in
(5) having not yet been introduced) By the second-year level it could be
argued that instruction provides examples in the form of so-called change of
meaning adjectives that variation in position is to be associated with
variation in interpretation (and perhaps by implication always to
interpretation) However learnersrsquo explicit knowledge of interpretative
differences as exemplified by the adjectives in (5) typically consists of
arbitrary and unsystematic English glosses By arbitrary I mean that
instruction never provides an adequate explanation for why the pre-N
position of cher should correspond to beloved or for its post-N position to
correspond to expensive7 Though telling students following Bragger and Rice
(1996) that the meaning in post-N position is lsquoconcretersquo and the meaning in
pre-N position is lsquofigurativersquo might (somehow) help it does not account for
other adjectives in (5) like prochain lsquonextrsquo This latter adjective also
demonstrates the fact that English glosses are unsystematic in that prochain
translates as lsquonextrsquo in either position
Before considering our final research question I note that no research on
adjective position in native French speaker discourse is available to date
An anonymous Applied Linguistics reviewer questions the validity of compar-
ing primarily spoken classroom discoursemdashor any other kind of spoken
discoursemdashto written texts without first demonstrating that adjective position
is insensitive to this difference in medium While recognizing the potentially
important distinction in medium and the need for further research in this
area the validity of comparing classroom input to written texts in the present
BRUCE ANDERSON 297
study lies in the fact that such a comparison roughly corresponds to the
curriculum of typical university-level courses an emphasis on listening
comprehension and oral communication in the first- and second-year
language courses leading to an emphasis on reading comprehension
(typically literature-based) and proficiency in writing in the third- and
fourth-year courses Were it to turn out that the degree of variation in
adjective position found in written texts is restricted to just that medium
revisions to pedagogy would necessarily be restricted to writing-intensive
(composition creative composition stylistics) courses at the third- and
fourth-year levels (I return to this issue in the discussion section)
LEARNER INTUITIONS OF ACCEPTABILITY
Thus far we have seen an incongruity between textbook presentations of
adjective position and authentic text sources across a number of written text
genres (research question 1) and that ambient language use involving
adjective position in the classroom is clearly more congruent with and might
therefore be taken to reinforce the explicit rules found in textbooks
(research question 2) We are now in a position to address our third research
questionmdashthe extent to which intuitions of acceptability among classroom
learners are biased by the congruity between pedagogical treatment and
classroom input frequency patterns An empirical study previously reported
in Anderson (2002 2007) can shed some light on this question
Methodology
Anderson (2002 2007) investigated the acquisition of various morphosyn-
tactic properties of French noun phrases by administering an acceptability
judgment task in which participants had to evaluate the appropriateness of
sentences featuring among other things the pre- versus post-N position of
adjectives Each of the adjectives selected were lsquohighly variablersquo with respect
to position (cf Table 1) yet crucially do not figure in textbook lists of
adjectives said to change meaning based on position (cf the list in (10))
These include lourd lsquoheavyrsquo violent lsquoviolentrsquo charmant lsquocharmingrsquo precieux
lsquopreciousrsquo epais lsquothickrsquo etrange lsquostrangersquo and delicieux lsquodeliciousrsquo (See
Appendix C available to online subscribers at httpapplijoxfordjournals
org for test sentences) Study participants included English-speaking learners
of French in the second third and fourth years of undergraduate study
(nfrac14 80) who were attending courses at the same institution at which the
observational data were collected a group of advanced learners (nfrac14 20)
consisting of graduate student instructors at that same institution as well as
secondary education teachers and a group of native French speakers (nfrac14 27)
roughly matched in age and education level
Each adjective appeared in test sentences once prenominally and once
postnominally with each test sentence preceded once by a context
298 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
motivating acceptance and once by a context motivating rejection of one or
the other adjective order for a total of four positionndashinterpretation pairings
per adjective These were randomly distributed across two testing
instruments administered a week apart After reading each context and
paired test sentence study participants were directed to check boxes
labeled lsquofinersquo lsquooddrsquo or lsquocannot decidersquo to the prompt This particular sentence
sounds ___ in this context
As an example each of the sentences in (7) featuring the adjective lourd
lsquoheavyrsquo was paired with each of two contexts
(7) a Pierre doit laisser la lourde valise avec son amib Pierre doit laisser la valise lourde a lrsquoaeroport
lsquoPierre has to leave the heavy suitcase with his friendatthe airportrsquo
In the first context motivating a unique noun-referent interpretation
there was one and only one suitcase which happened to be heavy and
which lsquoPierrersquo had to leave behind with his friend In the second context
motivating a multiple noun-referent interpretation there were a number of
suitcases only one of which was heavy and it was this one that lsquoPierrersquo had
to leave behind at the airport (See Appendix D available to online
subscribers at httpapplijoxfordjournalsorg for the complete version of
each context)
When viewed solely in terms of truth-conditional semantics each of the
four positionndashinterpretation pairings is compatible save for one pre-N
position (7a) in a multiple noun-referent context This is because the singular
definite article la and pre-N adjective lourde together presuppose a unique
suitcase in context (which happens to be heavy) This presupposition
does not hold when the adjective is in post-N position where all that is
required is that there be at least one such suitcase in context The testing
instrument then allows us to examine how classroom learners react to
an lsquoexpectedrsquo word order (ie sentences like (7b) in either context) as well
as to an lsquounexpectedrsquo but semantically and pragmatically appropriate
word order (ie sentences like (7a) but only in the unique noun-referent
context)
Results
As the mean acceptance rates from native French speakers in Table 2 show
truth-conditional semantics was not the only factor determining acceptance
rates Two of the pairings received essentially chance-level rates of
acceptance (pre-N position in a multiple noun-referent context at 49
percent and post-N position in a unique context at 52 percent) whereas the
other two pairings showed much clearer above- or below-chance rates (pre-N
position in a unique noun-referent context at 74 percent and post-N
position in a multiple noun-referent context at 37 percent)
BRUCE ANDERSON 299
As was argued in Anderson (2002 2007) results for each of the four
positionndashinterpretation pairings among the native French speakers are
explainable in terms of a disproportionate interaction between semantics and
pragmatic implicature in particular the Gricean maxim of quantity An
utterance should be no more and no less informative than is required for the
current purposes of the exchange (Grice 1975) The positionndashinterpretation
pairing that was lsquotruersquo in terms of semantics and also the most pragmatically
appropriate was accepted at the highest rate (74 percent) the pairing that
was also lsquotruersquo in terms of semantics but was pragmatically the least
appropriate was accepted at the lowest rate (37 percent)8
Do learner group response patterns show the same bias toward pragmatic
implicature in their acceptance rates The comparison of mean acceptance
rates between native French speakers and learner groups in Table 3 shows
that learners performed in a particular (and particularly non-Frenchlike) way
in which the interaction goes in the opposite direction favoring semantics
over pragmatics
Recall that the only positionndashinterpretation pairing that was semantically
incompatible was pre-N position (7a) in a multiple noun-referent context
given the presupposition of uniqueness of the noun referent All learner
groups did indeed accept this pairing at the lowest rate of the four and with
the exception of the second-year learners (at thorn9 percent) did so more
strongly than the native French speakers at mean acceptance rates that were
11ndash15 percent lower)9 However they also accepted at much lower mean
rates (12 to 23 percent) the most felicitous positionndashinterpretation pairing
for the native French speakers pre-N position in a unique noun-referent
context Recall also that post-N position (7b) was semantically compatible
under either interpretation All learner groups accepted these two positionndash
interpretation pairings at much higher rates than the native French speakers
(thorn23 to thorn31 percent) including the least felicitous positionndashinterpretation
pairing for the native French speakersmdashpost-N position in a multiple noun-
referent context This is precisely what one would expect from the
pedagogical treatment these learners are provided and the congruity of
that treatment with classroom input which under-represents actual usage in
Table 2 Native French speaker mean acceptance rates for the uniquenonunique noun-referent distinction
Prenominal (AN)order ()
Postnominal (NA)order ()
Unique noun referent 74 52
Nonunique noun referent 49 37
Source Adapted from Anderson 2002 2007
300 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
text-based corpora and as we have now seen differs from native speaker
intuitions of overall (semantico-pragmatic) acceptability
Although a bias toward or even strict adherence to explicit pedagogical
rules in evaluating such sentences may not seem all that surprising what does
stand out in the data from Anderson (2002 2007) is that these differential
rates in acceptance between native and nonnative speakers are as true for the
advanced group in that studymdashconsisting of graduate student instructors and
second education teachersmdashas they are for the intermediate (second-year)
learner group This would seem to indicate no increased sensitivity to
contextual (pragmatic) factors in adjective position over the course of studying
French even though the advanced group had been exposed to much more
authentic input through texts written for and by native French speakers (ie
the descriptive norm uncovered in our examination of authentic text corpora)
DISCUSSION
What should be clear from the findings presented here is that rules of French
adjective placement that one typically finds in (at least American) textbooks
though not plainly incorrect are not fully descriptively accurate either at
least with respect to written French across several text genres In particular
such rules portray pre-N and post-N positions as mutually exclusive for
almost all adjectives when in fact they are not
Similar discrepancies are often found between rules for and by native
speakers and what those speakers actually do One need look no further than
the French for an example with colloquial speech often at odds with the
prescriptive bon usage and the admonishments of the Academie francaise One
might therefore also expect to find a discrepancy between textbook rules and
what teachers do in the classroom andor what students find to be
lsquoacceptablersquo in terms of word order As we have seen however this is not
the case These rules are faithfully reflected in classroom language use by
instructors to such an extent that when asked to evaluate the acceptability
of lsquoexpectedrsquo and lsquounexpected but nativelikersquo word orders (and one would
presume in production as well) both instructors and students paid much
less attention to problems in pragmatic implicature than did native speakers
a situation which does not appear to change even in the face of counter-
examples occasionally provided in the textbooks they use (see (6) above) and
most certainly in the written texts that they read at later course levels As
pointed out by an anonymous reviewer this situation can be considered a
testament to the power of classroom instruction (instructor input in
particular) if instruction is changed he or she reasons it is likely that
classroom learners will reflect this change
But should instruction be changed If so who along the continuum from
the theorist-researcher to the teacher-practitioner should inform and guide
this change At which level In what way Such questions are at the very
heart of two on-going debates that I re-examine before proposing some
BRUCE ANDERSON 301
answers the pedagogical versus descriptive norm debate (eg Owen 1993
Widdowson 2000 and the collection of articles in Gass et al 2002) and
the interrelated native-speaker-usage-as-goal debate in foreign language
pedagogy (eg the collection of articles in Blyth 2003)
On the whole the approach to instruction on French adjective
position outlined earlier appears advantageous in its simplicity providing
easily remembered rules of thumb that learners can apply in language
production That this approach to instruction does not as we have seen
straightforwardly follow the descriptive norm is a situation that linguists
utilizing a corpus-based methodology believe can and should be remedied
Conrad (2000) for instance sees in corpus-based research an immediate way
in which linguistics can inform language pedagogy freeing language teachers
from lsquohav[ing] to rely on judgments of grammatical accuracy or on native
speakersrsquo intuition about what sounds naturalrsquo (2000 555) Although the
subsequent study by Biber and Reppen (2002) uses a disclaimer to the effect
that frequency should not be the sole factor in material design they
nevertheless suggest that lsquoa selective revision of pedagogy to reflect actual
use as shown by frequency studies could result in radical changes that
facilitate the learning process for studentsrsquo (2002 199) Owen (1993)
however provides examples of how a total reliance on a corpus does not
necessarily yield better observation and more importantly questions
whether better observation automatically equates to better explanation
(1993 168) (but see Francis and Sinclair 1994 for rebuttal) For his part
Widdowson (2000) objects to the kind of equation just described as a
case of linguistics applied a mistaken belief that what is textually attested
lsquouniquely represents real language and that this reality should define the
foreign language subjectrsquo (2000 8) (but see Stubbs 2001 for rebuttal) What
is needed according to Widdowson is not simply linguistics applied
to pedagogy but a true role for applied linguistics an approach to usage that
also recognizes native speakersrsquo awareness of and opinions about such usage
The need for a truly applied linguistics perspective on native speaker
usage (including frequency patterns) is embodied in a pedagogical norm
originally conceptualized in Valdman (1989) and elaborated upon in
Bardovi-Harlig and Gass (2002) as
[A] combination of language systems and forms selected by linguistsand pedagogues to serve as the immediate language target ortargets that learners seek to acquire during their language studyIn other words pedagogical norms represent a mid-point or seriesof mid-points for learners as they progress toward acquiringnative language norms [thus constituting] a form of languagethat is acceptable to native speakers but easier to learn than the fullnative language system (p 3)
The textbook word-order rules for French adjective position reviewed
earlier would first appear to constitute an appropriate pedagogical norm
302 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
under Bardovi-Harlig and Gassrsquos (2002) definition because the learner is able
to produce a subset of the totality of possible utterances deemed acceptable to
native speakers which may turn out to be even more the case in spoken than
in written French (though the difference in mediums remains to be studied)
The learner data presented here from Anderson (2002 2007) however
provides one piece of evidence that textbook rules and classroom input do not
constitute a sufficient pedagogical norm as learners have stopped in their
progress toward acquiring native speaker (ie descriptively accurate) norms
As the articles in Blyth (2003) attest however it is highly debatable
whether the lsquoNative Speakerrsquo (as an idealized abstraction) should be the
appropriate model for the contemporary learner For Kramsch (2002 2003)
too much of a focus on an authentic native speaker norm reduces
foreign language (FL) study to a constant lsquoapproximation to and conformity
with stereotypical native speakers thus often silencing other FL-specific
forms of language potentialrsquo (2002 60) Language learning in instructional
settings on my understanding of Kramschrsquos (2003) argument should not be
viewed solely in a teleological sensemdashas language acquisition directed
toward a fixed nativelike end-pointmdashbut rather as language appropriation
by learners in the sense that they are constructing a linguistic and even
cultural identity lsquoin the interstices of national languages and on the
margins of monolingual speakersrsquo territoriesrsquo (2003 260) The political nature
of the debate as highlighted by Kramsch becomes all the more apparent
when one additionally considers the status of the foreign language on
the world stage (eg French as an international language in retreat)
and how deeply the normative attitudes of its speakers run (eg Francersquos
three-century-long history of acceptance and maintenance of the standard
(Lodge 1993))
The foregoing synopsis of the debates surrounding native speaker usage as
both the goal of foreign language instruction and the extent to which
pedagogical norms need apply has highlighted the real world complexities
involved in what would otherwise seem to be a simple equation Revise
pedagogy to reflect actual use as shown by frequency studies and this
will result in a more targetlike performance on the part of learners So we
return then to the questions posed earlier (1) Should instruction be
changed (2) If so who along the continuum from the theorist-researcher to
the teacher-practitioner should inform and guide this change (3) At which
level (4) In what ways
With respect to our first question because one can empirically
demonstrate that the pedagogical norm for French adjective position
prevents learners from eventually acquiring native speaker (ie descriptively
accurate) norms instruction should be changed so thatmdashirrespective of
whether the attainment of such a norm is necessarily desirablemdashlearners
have both a choice in selecting one grammatical or lexical form over the other
and an awareness of the meaning potential of each choice (following Kramsch
2002 71)
BRUCE ANDERSON 303
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
study lies in the fact that such a comparison roughly corresponds to the
curriculum of typical university-level courses an emphasis on listening
comprehension and oral communication in the first- and second-year
language courses leading to an emphasis on reading comprehension
(typically literature-based) and proficiency in writing in the third- and
fourth-year courses Were it to turn out that the degree of variation in
adjective position found in written texts is restricted to just that medium
revisions to pedagogy would necessarily be restricted to writing-intensive
(composition creative composition stylistics) courses at the third- and
fourth-year levels (I return to this issue in the discussion section)
LEARNER INTUITIONS OF ACCEPTABILITY
Thus far we have seen an incongruity between textbook presentations of
adjective position and authentic text sources across a number of written text
genres (research question 1) and that ambient language use involving
adjective position in the classroom is clearly more congruent with and might
therefore be taken to reinforce the explicit rules found in textbooks
(research question 2) We are now in a position to address our third research
questionmdashthe extent to which intuitions of acceptability among classroom
learners are biased by the congruity between pedagogical treatment and
classroom input frequency patterns An empirical study previously reported
in Anderson (2002 2007) can shed some light on this question
Methodology
Anderson (2002 2007) investigated the acquisition of various morphosyn-
tactic properties of French noun phrases by administering an acceptability
judgment task in which participants had to evaluate the appropriateness of
sentences featuring among other things the pre- versus post-N position of
adjectives Each of the adjectives selected were lsquohighly variablersquo with respect
to position (cf Table 1) yet crucially do not figure in textbook lists of
adjectives said to change meaning based on position (cf the list in (10))
These include lourd lsquoheavyrsquo violent lsquoviolentrsquo charmant lsquocharmingrsquo precieux
lsquopreciousrsquo epais lsquothickrsquo etrange lsquostrangersquo and delicieux lsquodeliciousrsquo (See
Appendix C available to online subscribers at httpapplijoxfordjournals
org for test sentences) Study participants included English-speaking learners
of French in the second third and fourth years of undergraduate study
(nfrac14 80) who were attending courses at the same institution at which the
observational data were collected a group of advanced learners (nfrac14 20)
consisting of graduate student instructors at that same institution as well as
secondary education teachers and a group of native French speakers (nfrac14 27)
roughly matched in age and education level
Each adjective appeared in test sentences once prenominally and once
postnominally with each test sentence preceded once by a context
298 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
motivating acceptance and once by a context motivating rejection of one or
the other adjective order for a total of four positionndashinterpretation pairings
per adjective These were randomly distributed across two testing
instruments administered a week apart After reading each context and
paired test sentence study participants were directed to check boxes
labeled lsquofinersquo lsquooddrsquo or lsquocannot decidersquo to the prompt This particular sentence
sounds ___ in this context
As an example each of the sentences in (7) featuring the adjective lourd
lsquoheavyrsquo was paired with each of two contexts
(7) a Pierre doit laisser la lourde valise avec son amib Pierre doit laisser la valise lourde a lrsquoaeroport
lsquoPierre has to leave the heavy suitcase with his friendatthe airportrsquo
In the first context motivating a unique noun-referent interpretation
there was one and only one suitcase which happened to be heavy and
which lsquoPierrersquo had to leave behind with his friend In the second context
motivating a multiple noun-referent interpretation there were a number of
suitcases only one of which was heavy and it was this one that lsquoPierrersquo had
to leave behind at the airport (See Appendix D available to online
subscribers at httpapplijoxfordjournalsorg for the complete version of
each context)
When viewed solely in terms of truth-conditional semantics each of the
four positionndashinterpretation pairings is compatible save for one pre-N
position (7a) in a multiple noun-referent context This is because the singular
definite article la and pre-N adjective lourde together presuppose a unique
suitcase in context (which happens to be heavy) This presupposition
does not hold when the adjective is in post-N position where all that is
required is that there be at least one such suitcase in context The testing
instrument then allows us to examine how classroom learners react to
an lsquoexpectedrsquo word order (ie sentences like (7b) in either context) as well
as to an lsquounexpectedrsquo but semantically and pragmatically appropriate
word order (ie sentences like (7a) but only in the unique noun-referent
context)
Results
As the mean acceptance rates from native French speakers in Table 2 show
truth-conditional semantics was not the only factor determining acceptance
rates Two of the pairings received essentially chance-level rates of
acceptance (pre-N position in a multiple noun-referent context at 49
percent and post-N position in a unique context at 52 percent) whereas the
other two pairings showed much clearer above- or below-chance rates (pre-N
position in a unique noun-referent context at 74 percent and post-N
position in a multiple noun-referent context at 37 percent)
BRUCE ANDERSON 299
As was argued in Anderson (2002 2007) results for each of the four
positionndashinterpretation pairings among the native French speakers are
explainable in terms of a disproportionate interaction between semantics and
pragmatic implicature in particular the Gricean maxim of quantity An
utterance should be no more and no less informative than is required for the
current purposes of the exchange (Grice 1975) The positionndashinterpretation
pairing that was lsquotruersquo in terms of semantics and also the most pragmatically
appropriate was accepted at the highest rate (74 percent) the pairing that
was also lsquotruersquo in terms of semantics but was pragmatically the least
appropriate was accepted at the lowest rate (37 percent)8
Do learner group response patterns show the same bias toward pragmatic
implicature in their acceptance rates The comparison of mean acceptance
rates between native French speakers and learner groups in Table 3 shows
that learners performed in a particular (and particularly non-Frenchlike) way
in which the interaction goes in the opposite direction favoring semantics
over pragmatics
Recall that the only positionndashinterpretation pairing that was semantically
incompatible was pre-N position (7a) in a multiple noun-referent context
given the presupposition of uniqueness of the noun referent All learner
groups did indeed accept this pairing at the lowest rate of the four and with
the exception of the second-year learners (at thorn9 percent) did so more
strongly than the native French speakers at mean acceptance rates that were
11ndash15 percent lower)9 However they also accepted at much lower mean
rates (12 to 23 percent) the most felicitous positionndashinterpretation pairing
for the native French speakers pre-N position in a unique noun-referent
context Recall also that post-N position (7b) was semantically compatible
under either interpretation All learner groups accepted these two positionndash
interpretation pairings at much higher rates than the native French speakers
(thorn23 to thorn31 percent) including the least felicitous positionndashinterpretation
pairing for the native French speakersmdashpost-N position in a multiple noun-
referent context This is precisely what one would expect from the
pedagogical treatment these learners are provided and the congruity of
that treatment with classroom input which under-represents actual usage in
Table 2 Native French speaker mean acceptance rates for the uniquenonunique noun-referent distinction
Prenominal (AN)order ()
Postnominal (NA)order ()
Unique noun referent 74 52
Nonunique noun referent 49 37
Source Adapted from Anderson 2002 2007
300 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
text-based corpora and as we have now seen differs from native speaker
intuitions of overall (semantico-pragmatic) acceptability
Although a bias toward or even strict adherence to explicit pedagogical
rules in evaluating such sentences may not seem all that surprising what does
stand out in the data from Anderson (2002 2007) is that these differential
rates in acceptance between native and nonnative speakers are as true for the
advanced group in that studymdashconsisting of graduate student instructors and
second education teachersmdashas they are for the intermediate (second-year)
learner group This would seem to indicate no increased sensitivity to
contextual (pragmatic) factors in adjective position over the course of studying
French even though the advanced group had been exposed to much more
authentic input through texts written for and by native French speakers (ie
the descriptive norm uncovered in our examination of authentic text corpora)
DISCUSSION
What should be clear from the findings presented here is that rules of French
adjective placement that one typically finds in (at least American) textbooks
though not plainly incorrect are not fully descriptively accurate either at
least with respect to written French across several text genres In particular
such rules portray pre-N and post-N positions as mutually exclusive for
almost all adjectives when in fact they are not
Similar discrepancies are often found between rules for and by native
speakers and what those speakers actually do One need look no further than
the French for an example with colloquial speech often at odds with the
prescriptive bon usage and the admonishments of the Academie francaise One
might therefore also expect to find a discrepancy between textbook rules and
what teachers do in the classroom andor what students find to be
lsquoacceptablersquo in terms of word order As we have seen however this is not
the case These rules are faithfully reflected in classroom language use by
instructors to such an extent that when asked to evaluate the acceptability
of lsquoexpectedrsquo and lsquounexpected but nativelikersquo word orders (and one would
presume in production as well) both instructors and students paid much
less attention to problems in pragmatic implicature than did native speakers
a situation which does not appear to change even in the face of counter-
examples occasionally provided in the textbooks they use (see (6) above) and
most certainly in the written texts that they read at later course levels As
pointed out by an anonymous reviewer this situation can be considered a
testament to the power of classroom instruction (instructor input in
particular) if instruction is changed he or she reasons it is likely that
classroom learners will reflect this change
But should instruction be changed If so who along the continuum from
the theorist-researcher to the teacher-practitioner should inform and guide
this change At which level In what way Such questions are at the very
heart of two on-going debates that I re-examine before proposing some
BRUCE ANDERSON 301
answers the pedagogical versus descriptive norm debate (eg Owen 1993
Widdowson 2000 and the collection of articles in Gass et al 2002) and
the interrelated native-speaker-usage-as-goal debate in foreign language
pedagogy (eg the collection of articles in Blyth 2003)
On the whole the approach to instruction on French adjective
position outlined earlier appears advantageous in its simplicity providing
easily remembered rules of thumb that learners can apply in language
production That this approach to instruction does not as we have seen
straightforwardly follow the descriptive norm is a situation that linguists
utilizing a corpus-based methodology believe can and should be remedied
Conrad (2000) for instance sees in corpus-based research an immediate way
in which linguistics can inform language pedagogy freeing language teachers
from lsquohav[ing] to rely on judgments of grammatical accuracy or on native
speakersrsquo intuition about what sounds naturalrsquo (2000 555) Although the
subsequent study by Biber and Reppen (2002) uses a disclaimer to the effect
that frequency should not be the sole factor in material design they
nevertheless suggest that lsquoa selective revision of pedagogy to reflect actual
use as shown by frequency studies could result in radical changes that
facilitate the learning process for studentsrsquo (2002 199) Owen (1993)
however provides examples of how a total reliance on a corpus does not
necessarily yield better observation and more importantly questions
whether better observation automatically equates to better explanation
(1993 168) (but see Francis and Sinclair 1994 for rebuttal) For his part
Widdowson (2000) objects to the kind of equation just described as a
case of linguistics applied a mistaken belief that what is textually attested
lsquouniquely represents real language and that this reality should define the
foreign language subjectrsquo (2000 8) (but see Stubbs 2001 for rebuttal) What
is needed according to Widdowson is not simply linguistics applied
to pedagogy but a true role for applied linguistics an approach to usage that
also recognizes native speakersrsquo awareness of and opinions about such usage
The need for a truly applied linguistics perspective on native speaker
usage (including frequency patterns) is embodied in a pedagogical norm
originally conceptualized in Valdman (1989) and elaborated upon in
Bardovi-Harlig and Gass (2002) as
[A] combination of language systems and forms selected by linguistsand pedagogues to serve as the immediate language target ortargets that learners seek to acquire during their language studyIn other words pedagogical norms represent a mid-point or seriesof mid-points for learners as they progress toward acquiringnative language norms [thus constituting] a form of languagethat is acceptable to native speakers but easier to learn than the fullnative language system (p 3)
The textbook word-order rules for French adjective position reviewed
earlier would first appear to constitute an appropriate pedagogical norm
302 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
under Bardovi-Harlig and Gassrsquos (2002) definition because the learner is able
to produce a subset of the totality of possible utterances deemed acceptable to
native speakers which may turn out to be even more the case in spoken than
in written French (though the difference in mediums remains to be studied)
The learner data presented here from Anderson (2002 2007) however
provides one piece of evidence that textbook rules and classroom input do not
constitute a sufficient pedagogical norm as learners have stopped in their
progress toward acquiring native speaker (ie descriptively accurate) norms
As the articles in Blyth (2003) attest however it is highly debatable
whether the lsquoNative Speakerrsquo (as an idealized abstraction) should be the
appropriate model for the contemporary learner For Kramsch (2002 2003)
too much of a focus on an authentic native speaker norm reduces
foreign language (FL) study to a constant lsquoapproximation to and conformity
with stereotypical native speakers thus often silencing other FL-specific
forms of language potentialrsquo (2002 60) Language learning in instructional
settings on my understanding of Kramschrsquos (2003) argument should not be
viewed solely in a teleological sensemdashas language acquisition directed
toward a fixed nativelike end-pointmdashbut rather as language appropriation
by learners in the sense that they are constructing a linguistic and even
cultural identity lsquoin the interstices of national languages and on the
margins of monolingual speakersrsquo territoriesrsquo (2003 260) The political nature
of the debate as highlighted by Kramsch becomes all the more apparent
when one additionally considers the status of the foreign language on
the world stage (eg French as an international language in retreat)
and how deeply the normative attitudes of its speakers run (eg Francersquos
three-century-long history of acceptance and maintenance of the standard
(Lodge 1993))
The foregoing synopsis of the debates surrounding native speaker usage as
both the goal of foreign language instruction and the extent to which
pedagogical norms need apply has highlighted the real world complexities
involved in what would otherwise seem to be a simple equation Revise
pedagogy to reflect actual use as shown by frequency studies and this
will result in a more targetlike performance on the part of learners So we
return then to the questions posed earlier (1) Should instruction be
changed (2) If so who along the continuum from the theorist-researcher to
the teacher-practitioner should inform and guide this change (3) At which
level (4) In what ways
With respect to our first question because one can empirically
demonstrate that the pedagogical norm for French adjective position
prevents learners from eventually acquiring native speaker (ie descriptively
accurate) norms instruction should be changed so thatmdashirrespective of
whether the attainment of such a norm is necessarily desirablemdashlearners
have both a choice in selecting one grammatical or lexical form over the other
and an awareness of the meaning potential of each choice (following Kramsch
2002 71)
BRUCE ANDERSON 303
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
motivating acceptance and once by a context motivating rejection of one or
the other adjective order for a total of four positionndashinterpretation pairings
per adjective These were randomly distributed across two testing
instruments administered a week apart After reading each context and
paired test sentence study participants were directed to check boxes
labeled lsquofinersquo lsquooddrsquo or lsquocannot decidersquo to the prompt This particular sentence
sounds ___ in this context
As an example each of the sentences in (7) featuring the adjective lourd
lsquoheavyrsquo was paired with each of two contexts
(7) a Pierre doit laisser la lourde valise avec son amib Pierre doit laisser la valise lourde a lrsquoaeroport
lsquoPierre has to leave the heavy suitcase with his friendatthe airportrsquo
In the first context motivating a unique noun-referent interpretation
there was one and only one suitcase which happened to be heavy and
which lsquoPierrersquo had to leave behind with his friend In the second context
motivating a multiple noun-referent interpretation there were a number of
suitcases only one of which was heavy and it was this one that lsquoPierrersquo had
to leave behind at the airport (See Appendix D available to online
subscribers at httpapplijoxfordjournalsorg for the complete version of
each context)
When viewed solely in terms of truth-conditional semantics each of the
four positionndashinterpretation pairings is compatible save for one pre-N
position (7a) in a multiple noun-referent context This is because the singular
definite article la and pre-N adjective lourde together presuppose a unique
suitcase in context (which happens to be heavy) This presupposition
does not hold when the adjective is in post-N position where all that is
required is that there be at least one such suitcase in context The testing
instrument then allows us to examine how classroom learners react to
an lsquoexpectedrsquo word order (ie sentences like (7b) in either context) as well
as to an lsquounexpectedrsquo but semantically and pragmatically appropriate
word order (ie sentences like (7a) but only in the unique noun-referent
context)
Results
As the mean acceptance rates from native French speakers in Table 2 show
truth-conditional semantics was not the only factor determining acceptance
rates Two of the pairings received essentially chance-level rates of
acceptance (pre-N position in a multiple noun-referent context at 49
percent and post-N position in a unique context at 52 percent) whereas the
other two pairings showed much clearer above- or below-chance rates (pre-N
position in a unique noun-referent context at 74 percent and post-N
position in a multiple noun-referent context at 37 percent)
BRUCE ANDERSON 299
As was argued in Anderson (2002 2007) results for each of the four
positionndashinterpretation pairings among the native French speakers are
explainable in terms of a disproportionate interaction between semantics and
pragmatic implicature in particular the Gricean maxim of quantity An
utterance should be no more and no less informative than is required for the
current purposes of the exchange (Grice 1975) The positionndashinterpretation
pairing that was lsquotruersquo in terms of semantics and also the most pragmatically
appropriate was accepted at the highest rate (74 percent) the pairing that
was also lsquotruersquo in terms of semantics but was pragmatically the least
appropriate was accepted at the lowest rate (37 percent)8
Do learner group response patterns show the same bias toward pragmatic
implicature in their acceptance rates The comparison of mean acceptance
rates between native French speakers and learner groups in Table 3 shows
that learners performed in a particular (and particularly non-Frenchlike) way
in which the interaction goes in the opposite direction favoring semantics
over pragmatics
Recall that the only positionndashinterpretation pairing that was semantically
incompatible was pre-N position (7a) in a multiple noun-referent context
given the presupposition of uniqueness of the noun referent All learner
groups did indeed accept this pairing at the lowest rate of the four and with
the exception of the second-year learners (at thorn9 percent) did so more
strongly than the native French speakers at mean acceptance rates that were
11ndash15 percent lower)9 However they also accepted at much lower mean
rates (12 to 23 percent) the most felicitous positionndashinterpretation pairing
for the native French speakers pre-N position in a unique noun-referent
context Recall also that post-N position (7b) was semantically compatible
under either interpretation All learner groups accepted these two positionndash
interpretation pairings at much higher rates than the native French speakers
(thorn23 to thorn31 percent) including the least felicitous positionndashinterpretation
pairing for the native French speakersmdashpost-N position in a multiple noun-
referent context This is precisely what one would expect from the
pedagogical treatment these learners are provided and the congruity of
that treatment with classroom input which under-represents actual usage in
Table 2 Native French speaker mean acceptance rates for the uniquenonunique noun-referent distinction
Prenominal (AN)order ()
Postnominal (NA)order ()
Unique noun referent 74 52
Nonunique noun referent 49 37
Source Adapted from Anderson 2002 2007
300 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
text-based corpora and as we have now seen differs from native speaker
intuitions of overall (semantico-pragmatic) acceptability
Although a bias toward or even strict adherence to explicit pedagogical
rules in evaluating such sentences may not seem all that surprising what does
stand out in the data from Anderson (2002 2007) is that these differential
rates in acceptance between native and nonnative speakers are as true for the
advanced group in that studymdashconsisting of graduate student instructors and
second education teachersmdashas they are for the intermediate (second-year)
learner group This would seem to indicate no increased sensitivity to
contextual (pragmatic) factors in adjective position over the course of studying
French even though the advanced group had been exposed to much more
authentic input through texts written for and by native French speakers (ie
the descriptive norm uncovered in our examination of authentic text corpora)
DISCUSSION
What should be clear from the findings presented here is that rules of French
adjective placement that one typically finds in (at least American) textbooks
though not plainly incorrect are not fully descriptively accurate either at
least with respect to written French across several text genres In particular
such rules portray pre-N and post-N positions as mutually exclusive for
almost all adjectives when in fact they are not
Similar discrepancies are often found between rules for and by native
speakers and what those speakers actually do One need look no further than
the French for an example with colloquial speech often at odds with the
prescriptive bon usage and the admonishments of the Academie francaise One
might therefore also expect to find a discrepancy between textbook rules and
what teachers do in the classroom andor what students find to be
lsquoacceptablersquo in terms of word order As we have seen however this is not
the case These rules are faithfully reflected in classroom language use by
instructors to such an extent that when asked to evaluate the acceptability
of lsquoexpectedrsquo and lsquounexpected but nativelikersquo word orders (and one would
presume in production as well) both instructors and students paid much
less attention to problems in pragmatic implicature than did native speakers
a situation which does not appear to change even in the face of counter-
examples occasionally provided in the textbooks they use (see (6) above) and
most certainly in the written texts that they read at later course levels As
pointed out by an anonymous reviewer this situation can be considered a
testament to the power of classroom instruction (instructor input in
particular) if instruction is changed he or she reasons it is likely that
classroom learners will reflect this change
But should instruction be changed If so who along the continuum from
the theorist-researcher to the teacher-practitioner should inform and guide
this change At which level In what way Such questions are at the very
heart of two on-going debates that I re-examine before proposing some
BRUCE ANDERSON 301
answers the pedagogical versus descriptive norm debate (eg Owen 1993
Widdowson 2000 and the collection of articles in Gass et al 2002) and
the interrelated native-speaker-usage-as-goal debate in foreign language
pedagogy (eg the collection of articles in Blyth 2003)
On the whole the approach to instruction on French adjective
position outlined earlier appears advantageous in its simplicity providing
easily remembered rules of thumb that learners can apply in language
production That this approach to instruction does not as we have seen
straightforwardly follow the descriptive norm is a situation that linguists
utilizing a corpus-based methodology believe can and should be remedied
Conrad (2000) for instance sees in corpus-based research an immediate way
in which linguistics can inform language pedagogy freeing language teachers
from lsquohav[ing] to rely on judgments of grammatical accuracy or on native
speakersrsquo intuition about what sounds naturalrsquo (2000 555) Although the
subsequent study by Biber and Reppen (2002) uses a disclaimer to the effect
that frequency should not be the sole factor in material design they
nevertheless suggest that lsquoa selective revision of pedagogy to reflect actual
use as shown by frequency studies could result in radical changes that
facilitate the learning process for studentsrsquo (2002 199) Owen (1993)
however provides examples of how a total reliance on a corpus does not
necessarily yield better observation and more importantly questions
whether better observation automatically equates to better explanation
(1993 168) (but see Francis and Sinclair 1994 for rebuttal) For his part
Widdowson (2000) objects to the kind of equation just described as a
case of linguistics applied a mistaken belief that what is textually attested
lsquouniquely represents real language and that this reality should define the
foreign language subjectrsquo (2000 8) (but see Stubbs 2001 for rebuttal) What
is needed according to Widdowson is not simply linguistics applied
to pedagogy but a true role for applied linguistics an approach to usage that
also recognizes native speakersrsquo awareness of and opinions about such usage
The need for a truly applied linguistics perspective on native speaker
usage (including frequency patterns) is embodied in a pedagogical norm
originally conceptualized in Valdman (1989) and elaborated upon in
Bardovi-Harlig and Gass (2002) as
[A] combination of language systems and forms selected by linguistsand pedagogues to serve as the immediate language target ortargets that learners seek to acquire during their language studyIn other words pedagogical norms represent a mid-point or seriesof mid-points for learners as they progress toward acquiringnative language norms [thus constituting] a form of languagethat is acceptable to native speakers but easier to learn than the fullnative language system (p 3)
The textbook word-order rules for French adjective position reviewed
earlier would first appear to constitute an appropriate pedagogical norm
302 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
under Bardovi-Harlig and Gassrsquos (2002) definition because the learner is able
to produce a subset of the totality of possible utterances deemed acceptable to
native speakers which may turn out to be even more the case in spoken than
in written French (though the difference in mediums remains to be studied)
The learner data presented here from Anderson (2002 2007) however
provides one piece of evidence that textbook rules and classroom input do not
constitute a sufficient pedagogical norm as learners have stopped in their
progress toward acquiring native speaker (ie descriptively accurate) norms
As the articles in Blyth (2003) attest however it is highly debatable
whether the lsquoNative Speakerrsquo (as an idealized abstraction) should be the
appropriate model for the contemporary learner For Kramsch (2002 2003)
too much of a focus on an authentic native speaker norm reduces
foreign language (FL) study to a constant lsquoapproximation to and conformity
with stereotypical native speakers thus often silencing other FL-specific
forms of language potentialrsquo (2002 60) Language learning in instructional
settings on my understanding of Kramschrsquos (2003) argument should not be
viewed solely in a teleological sensemdashas language acquisition directed
toward a fixed nativelike end-pointmdashbut rather as language appropriation
by learners in the sense that they are constructing a linguistic and even
cultural identity lsquoin the interstices of national languages and on the
margins of monolingual speakersrsquo territoriesrsquo (2003 260) The political nature
of the debate as highlighted by Kramsch becomes all the more apparent
when one additionally considers the status of the foreign language on
the world stage (eg French as an international language in retreat)
and how deeply the normative attitudes of its speakers run (eg Francersquos
three-century-long history of acceptance and maintenance of the standard
(Lodge 1993))
The foregoing synopsis of the debates surrounding native speaker usage as
both the goal of foreign language instruction and the extent to which
pedagogical norms need apply has highlighted the real world complexities
involved in what would otherwise seem to be a simple equation Revise
pedagogy to reflect actual use as shown by frequency studies and this
will result in a more targetlike performance on the part of learners So we
return then to the questions posed earlier (1) Should instruction be
changed (2) If so who along the continuum from the theorist-researcher to
the teacher-practitioner should inform and guide this change (3) At which
level (4) In what ways
With respect to our first question because one can empirically
demonstrate that the pedagogical norm for French adjective position
prevents learners from eventually acquiring native speaker (ie descriptively
accurate) norms instruction should be changed so thatmdashirrespective of
whether the attainment of such a norm is necessarily desirablemdashlearners
have both a choice in selecting one grammatical or lexical form over the other
and an awareness of the meaning potential of each choice (following Kramsch
2002 71)
BRUCE ANDERSON 303
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
As was argued in Anderson (2002 2007) results for each of the four
positionndashinterpretation pairings among the native French speakers are
explainable in terms of a disproportionate interaction between semantics and
pragmatic implicature in particular the Gricean maxim of quantity An
utterance should be no more and no less informative than is required for the
current purposes of the exchange (Grice 1975) The positionndashinterpretation
pairing that was lsquotruersquo in terms of semantics and also the most pragmatically
appropriate was accepted at the highest rate (74 percent) the pairing that
was also lsquotruersquo in terms of semantics but was pragmatically the least
appropriate was accepted at the lowest rate (37 percent)8
Do learner group response patterns show the same bias toward pragmatic
implicature in their acceptance rates The comparison of mean acceptance
rates between native French speakers and learner groups in Table 3 shows
that learners performed in a particular (and particularly non-Frenchlike) way
in which the interaction goes in the opposite direction favoring semantics
over pragmatics
Recall that the only positionndashinterpretation pairing that was semantically
incompatible was pre-N position (7a) in a multiple noun-referent context
given the presupposition of uniqueness of the noun referent All learner
groups did indeed accept this pairing at the lowest rate of the four and with
the exception of the second-year learners (at thorn9 percent) did so more
strongly than the native French speakers at mean acceptance rates that were
11ndash15 percent lower)9 However they also accepted at much lower mean
rates (12 to 23 percent) the most felicitous positionndashinterpretation pairing
for the native French speakers pre-N position in a unique noun-referent
context Recall also that post-N position (7b) was semantically compatible
under either interpretation All learner groups accepted these two positionndash
interpretation pairings at much higher rates than the native French speakers
(thorn23 to thorn31 percent) including the least felicitous positionndashinterpretation
pairing for the native French speakersmdashpost-N position in a multiple noun-
referent context This is precisely what one would expect from the
pedagogical treatment these learners are provided and the congruity of
that treatment with classroom input which under-represents actual usage in
Table 2 Native French speaker mean acceptance rates for the uniquenonunique noun-referent distinction
Prenominal (AN)order ()
Postnominal (NA)order ()
Unique noun referent 74 52
Nonunique noun referent 49 37
Source Adapted from Anderson 2002 2007
300 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
text-based corpora and as we have now seen differs from native speaker
intuitions of overall (semantico-pragmatic) acceptability
Although a bias toward or even strict adherence to explicit pedagogical
rules in evaluating such sentences may not seem all that surprising what does
stand out in the data from Anderson (2002 2007) is that these differential
rates in acceptance between native and nonnative speakers are as true for the
advanced group in that studymdashconsisting of graduate student instructors and
second education teachersmdashas they are for the intermediate (second-year)
learner group This would seem to indicate no increased sensitivity to
contextual (pragmatic) factors in adjective position over the course of studying
French even though the advanced group had been exposed to much more
authentic input through texts written for and by native French speakers (ie
the descriptive norm uncovered in our examination of authentic text corpora)
DISCUSSION
What should be clear from the findings presented here is that rules of French
adjective placement that one typically finds in (at least American) textbooks
though not plainly incorrect are not fully descriptively accurate either at
least with respect to written French across several text genres In particular
such rules portray pre-N and post-N positions as mutually exclusive for
almost all adjectives when in fact they are not
Similar discrepancies are often found between rules for and by native
speakers and what those speakers actually do One need look no further than
the French for an example with colloquial speech often at odds with the
prescriptive bon usage and the admonishments of the Academie francaise One
might therefore also expect to find a discrepancy between textbook rules and
what teachers do in the classroom andor what students find to be
lsquoacceptablersquo in terms of word order As we have seen however this is not
the case These rules are faithfully reflected in classroom language use by
instructors to such an extent that when asked to evaluate the acceptability
of lsquoexpectedrsquo and lsquounexpected but nativelikersquo word orders (and one would
presume in production as well) both instructors and students paid much
less attention to problems in pragmatic implicature than did native speakers
a situation which does not appear to change even in the face of counter-
examples occasionally provided in the textbooks they use (see (6) above) and
most certainly in the written texts that they read at later course levels As
pointed out by an anonymous reviewer this situation can be considered a
testament to the power of classroom instruction (instructor input in
particular) if instruction is changed he or she reasons it is likely that
classroom learners will reflect this change
But should instruction be changed If so who along the continuum from
the theorist-researcher to the teacher-practitioner should inform and guide
this change At which level In what way Such questions are at the very
heart of two on-going debates that I re-examine before proposing some
BRUCE ANDERSON 301
answers the pedagogical versus descriptive norm debate (eg Owen 1993
Widdowson 2000 and the collection of articles in Gass et al 2002) and
the interrelated native-speaker-usage-as-goal debate in foreign language
pedagogy (eg the collection of articles in Blyth 2003)
On the whole the approach to instruction on French adjective
position outlined earlier appears advantageous in its simplicity providing
easily remembered rules of thumb that learners can apply in language
production That this approach to instruction does not as we have seen
straightforwardly follow the descriptive norm is a situation that linguists
utilizing a corpus-based methodology believe can and should be remedied
Conrad (2000) for instance sees in corpus-based research an immediate way
in which linguistics can inform language pedagogy freeing language teachers
from lsquohav[ing] to rely on judgments of grammatical accuracy or on native
speakersrsquo intuition about what sounds naturalrsquo (2000 555) Although the
subsequent study by Biber and Reppen (2002) uses a disclaimer to the effect
that frequency should not be the sole factor in material design they
nevertheless suggest that lsquoa selective revision of pedagogy to reflect actual
use as shown by frequency studies could result in radical changes that
facilitate the learning process for studentsrsquo (2002 199) Owen (1993)
however provides examples of how a total reliance on a corpus does not
necessarily yield better observation and more importantly questions
whether better observation automatically equates to better explanation
(1993 168) (but see Francis and Sinclair 1994 for rebuttal) For his part
Widdowson (2000) objects to the kind of equation just described as a
case of linguistics applied a mistaken belief that what is textually attested
lsquouniquely represents real language and that this reality should define the
foreign language subjectrsquo (2000 8) (but see Stubbs 2001 for rebuttal) What
is needed according to Widdowson is not simply linguistics applied
to pedagogy but a true role for applied linguistics an approach to usage that
also recognizes native speakersrsquo awareness of and opinions about such usage
The need for a truly applied linguistics perspective on native speaker
usage (including frequency patterns) is embodied in a pedagogical norm
originally conceptualized in Valdman (1989) and elaborated upon in
Bardovi-Harlig and Gass (2002) as
[A] combination of language systems and forms selected by linguistsand pedagogues to serve as the immediate language target ortargets that learners seek to acquire during their language studyIn other words pedagogical norms represent a mid-point or seriesof mid-points for learners as they progress toward acquiringnative language norms [thus constituting] a form of languagethat is acceptable to native speakers but easier to learn than the fullnative language system (p 3)
The textbook word-order rules for French adjective position reviewed
earlier would first appear to constitute an appropriate pedagogical norm
302 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
under Bardovi-Harlig and Gassrsquos (2002) definition because the learner is able
to produce a subset of the totality of possible utterances deemed acceptable to
native speakers which may turn out to be even more the case in spoken than
in written French (though the difference in mediums remains to be studied)
The learner data presented here from Anderson (2002 2007) however
provides one piece of evidence that textbook rules and classroom input do not
constitute a sufficient pedagogical norm as learners have stopped in their
progress toward acquiring native speaker (ie descriptively accurate) norms
As the articles in Blyth (2003) attest however it is highly debatable
whether the lsquoNative Speakerrsquo (as an idealized abstraction) should be the
appropriate model for the contemporary learner For Kramsch (2002 2003)
too much of a focus on an authentic native speaker norm reduces
foreign language (FL) study to a constant lsquoapproximation to and conformity
with stereotypical native speakers thus often silencing other FL-specific
forms of language potentialrsquo (2002 60) Language learning in instructional
settings on my understanding of Kramschrsquos (2003) argument should not be
viewed solely in a teleological sensemdashas language acquisition directed
toward a fixed nativelike end-pointmdashbut rather as language appropriation
by learners in the sense that they are constructing a linguistic and even
cultural identity lsquoin the interstices of national languages and on the
margins of monolingual speakersrsquo territoriesrsquo (2003 260) The political nature
of the debate as highlighted by Kramsch becomes all the more apparent
when one additionally considers the status of the foreign language on
the world stage (eg French as an international language in retreat)
and how deeply the normative attitudes of its speakers run (eg Francersquos
three-century-long history of acceptance and maintenance of the standard
(Lodge 1993))
The foregoing synopsis of the debates surrounding native speaker usage as
both the goal of foreign language instruction and the extent to which
pedagogical norms need apply has highlighted the real world complexities
involved in what would otherwise seem to be a simple equation Revise
pedagogy to reflect actual use as shown by frequency studies and this
will result in a more targetlike performance on the part of learners So we
return then to the questions posed earlier (1) Should instruction be
changed (2) If so who along the continuum from the theorist-researcher to
the teacher-practitioner should inform and guide this change (3) At which
level (4) In what ways
With respect to our first question because one can empirically
demonstrate that the pedagogical norm for French adjective position
prevents learners from eventually acquiring native speaker (ie descriptively
accurate) norms instruction should be changed so thatmdashirrespective of
whether the attainment of such a norm is necessarily desirablemdashlearners
have both a choice in selecting one grammatical or lexical form over the other
and an awareness of the meaning potential of each choice (following Kramsch
2002 71)
BRUCE ANDERSON 303
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
text-based corpora and as we have now seen differs from native speaker
intuitions of overall (semantico-pragmatic) acceptability
Although a bias toward or even strict adherence to explicit pedagogical
rules in evaluating such sentences may not seem all that surprising what does
stand out in the data from Anderson (2002 2007) is that these differential
rates in acceptance between native and nonnative speakers are as true for the
advanced group in that studymdashconsisting of graduate student instructors and
second education teachersmdashas they are for the intermediate (second-year)
learner group This would seem to indicate no increased sensitivity to
contextual (pragmatic) factors in adjective position over the course of studying
French even though the advanced group had been exposed to much more
authentic input through texts written for and by native French speakers (ie
the descriptive norm uncovered in our examination of authentic text corpora)
DISCUSSION
What should be clear from the findings presented here is that rules of French
adjective placement that one typically finds in (at least American) textbooks
though not plainly incorrect are not fully descriptively accurate either at
least with respect to written French across several text genres In particular
such rules portray pre-N and post-N positions as mutually exclusive for
almost all adjectives when in fact they are not
Similar discrepancies are often found between rules for and by native
speakers and what those speakers actually do One need look no further than
the French for an example with colloquial speech often at odds with the
prescriptive bon usage and the admonishments of the Academie francaise One
might therefore also expect to find a discrepancy between textbook rules and
what teachers do in the classroom andor what students find to be
lsquoacceptablersquo in terms of word order As we have seen however this is not
the case These rules are faithfully reflected in classroom language use by
instructors to such an extent that when asked to evaluate the acceptability
of lsquoexpectedrsquo and lsquounexpected but nativelikersquo word orders (and one would
presume in production as well) both instructors and students paid much
less attention to problems in pragmatic implicature than did native speakers
a situation which does not appear to change even in the face of counter-
examples occasionally provided in the textbooks they use (see (6) above) and
most certainly in the written texts that they read at later course levels As
pointed out by an anonymous reviewer this situation can be considered a
testament to the power of classroom instruction (instructor input in
particular) if instruction is changed he or she reasons it is likely that
classroom learners will reflect this change
But should instruction be changed If so who along the continuum from
the theorist-researcher to the teacher-practitioner should inform and guide
this change At which level In what way Such questions are at the very
heart of two on-going debates that I re-examine before proposing some
BRUCE ANDERSON 301
answers the pedagogical versus descriptive norm debate (eg Owen 1993
Widdowson 2000 and the collection of articles in Gass et al 2002) and
the interrelated native-speaker-usage-as-goal debate in foreign language
pedagogy (eg the collection of articles in Blyth 2003)
On the whole the approach to instruction on French adjective
position outlined earlier appears advantageous in its simplicity providing
easily remembered rules of thumb that learners can apply in language
production That this approach to instruction does not as we have seen
straightforwardly follow the descriptive norm is a situation that linguists
utilizing a corpus-based methodology believe can and should be remedied
Conrad (2000) for instance sees in corpus-based research an immediate way
in which linguistics can inform language pedagogy freeing language teachers
from lsquohav[ing] to rely on judgments of grammatical accuracy or on native
speakersrsquo intuition about what sounds naturalrsquo (2000 555) Although the
subsequent study by Biber and Reppen (2002) uses a disclaimer to the effect
that frequency should not be the sole factor in material design they
nevertheless suggest that lsquoa selective revision of pedagogy to reflect actual
use as shown by frequency studies could result in radical changes that
facilitate the learning process for studentsrsquo (2002 199) Owen (1993)
however provides examples of how a total reliance on a corpus does not
necessarily yield better observation and more importantly questions
whether better observation automatically equates to better explanation
(1993 168) (but see Francis and Sinclair 1994 for rebuttal) For his part
Widdowson (2000) objects to the kind of equation just described as a
case of linguistics applied a mistaken belief that what is textually attested
lsquouniquely represents real language and that this reality should define the
foreign language subjectrsquo (2000 8) (but see Stubbs 2001 for rebuttal) What
is needed according to Widdowson is not simply linguistics applied
to pedagogy but a true role for applied linguistics an approach to usage that
also recognizes native speakersrsquo awareness of and opinions about such usage
The need for a truly applied linguistics perspective on native speaker
usage (including frequency patterns) is embodied in a pedagogical norm
originally conceptualized in Valdman (1989) and elaborated upon in
Bardovi-Harlig and Gass (2002) as
[A] combination of language systems and forms selected by linguistsand pedagogues to serve as the immediate language target ortargets that learners seek to acquire during their language studyIn other words pedagogical norms represent a mid-point or seriesof mid-points for learners as they progress toward acquiringnative language norms [thus constituting] a form of languagethat is acceptable to native speakers but easier to learn than the fullnative language system (p 3)
The textbook word-order rules for French adjective position reviewed
earlier would first appear to constitute an appropriate pedagogical norm
302 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
under Bardovi-Harlig and Gassrsquos (2002) definition because the learner is able
to produce a subset of the totality of possible utterances deemed acceptable to
native speakers which may turn out to be even more the case in spoken than
in written French (though the difference in mediums remains to be studied)
The learner data presented here from Anderson (2002 2007) however
provides one piece of evidence that textbook rules and classroom input do not
constitute a sufficient pedagogical norm as learners have stopped in their
progress toward acquiring native speaker (ie descriptively accurate) norms
As the articles in Blyth (2003) attest however it is highly debatable
whether the lsquoNative Speakerrsquo (as an idealized abstraction) should be the
appropriate model for the contemporary learner For Kramsch (2002 2003)
too much of a focus on an authentic native speaker norm reduces
foreign language (FL) study to a constant lsquoapproximation to and conformity
with stereotypical native speakers thus often silencing other FL-specific
forms of language potentialrsquo (2002 60) Language learning in instructional
settings on my understanding of Kramschrsquos (2003) argument should not be
viewed solely in a teleological sensemdashas language acquisition directed
toward a fixed nativelike end-pointmdashbut rather as language appropriation
by learners in the sense that they are constructing a linguistic and even
cultural identity lsquoin the interstices of national languages and on the
margins of monolingual speakersrsquo territoriesrsquo (2003 260) The political nature
of the debate as highlighted by Kramsch becomes all the more apparent
when one additionally considers the status of the foreign language on
the world stage (eg French as an international language in retreat)
and how deeply the normative attitudes of its speakers run (eg Francersquos
three-century-long history of acceptance and maintenance of the standard
(Lodge 1993))
The foregoing synopsis of the debates surrounding native speaker usage as
both the goal of foreign language instruction and the extent to which
pedagogical norms need apply has highlighted the real world complexities
involved in what would otherwise seem to be a simple equation Revise
pedagogy to reflect actual use as shown by frequency studies and this
will result in a more targetlike performance on the part of learners So we
return then to the questions posed earlier (1) Should instruction be
changed (2) If so who along the continuum from the theorist-researcher to
the teacher-practitioner should inform and guide this change (3) At which
level (4) In what ways
With respect to our first question because one can empirically
demonstrate that the pedagogical norm for French adjective position
prevents learners from eventually acquiring native speaker (ie descriptively
accurate) norms instruction should be changed so thatmdashirrespective of
whether the attainment of such a norm is necessarily desirablemdashlearners
have both a choice in selecting one grammatical or lexical form over the other
and an awareness of the meaning potential of each choice (following Kramsch
2002 71)
BRUCE ANDERSON 303
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
answers the pedagogical versus descriptive norm debate (eg Owen 1993
Widdowson 2000 and the collection of articles in Gass et al 2002) and
the interrelated native-speaker-usage-as-goal debate in foreign language
pedagogy (eg the collection of articles in Blyth 2003)
On the whole the approach to instruction on French adjective
position outlined earlier appears advantageous in its simplicity providing
easily remembered rules of thumb that learners can apply in language
production That this approach to instruction does not as we have seen
straightforwardly follow the descriptive norm is a situation that linguists
utilizing a corpus-based methodology believe can and should be remedied
Conrad (2000) for instance sees in corpus-based research an immediate way
in which linguistics can inform language pedagogy freeing language teachers
from lsquohav[ing] to rely on judgments of grammatical accuracy or on native
speakersrsquo intuition about what sounds naturalrsquo (2000 555) Although the
subsequent study by Biber and Reppen (2002) uses a disclaimer to the effect
that frequency should not be the sole factor in material design they
nevertheless suggest that lsquoa selective revision of pedagogy to reflect actual
use as shown by frequency studies could result in radical changes that
facilitate the learning process for studentsrsquo (2002 199) Owen (1993)
however provides examples of how a total reliance on a corpus does not
necessarily yield better observation and more importantly questions
whether better observation automatically equates to better explanation
(1993 168) (but see Francis and Sinclair 1994 for rebuttal) For his part
Widdowson (2000) objects to the kind of equation just described as a
case of linguistics applied a mistaken belief that what is textually attested
lsquouniquely represents real language and that this reality should define the
foreign language subjectrsquo (2000 8) (but see Stubbs 2001 for rebuttal) What
is needed according to Widdowson is not simply linguistics applied
to pedagogy but a true role for applied linguistics an approach to usage that
also recognizes native speakersrsquo awareness of and opinions about such usage
The need for a truly applied linguistics perspective on native speaker
usage (including frequency patterns) is embodied in a pedagogical norm
originally conceptualized in Valdman (1989) and elaborated upon in
Bardovi-Harlig and Gass (2002) as
[A] combination of language systems and forms selected by linguistsand pedagogues to serve as the immediate language target ortargets that learners seek to acquire during their language studyIn other words pedagogical norms represent a mid-point or seriesof mid-points for learners as they progress toward acquiringnative language norms [thus constituting] a form of languagethat is acceptable to native speakers but easier to learn than the fullnative language system (p 3)
The textbook word-order rules for French adjective position reviewed
earlier would first appear to constitute an appropriate pedagogical norm
302 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
under Bardovi-Harlig and Gassrsquos (2002) definition because the learner is able
to produce a subset of the totality of possible utterances deemed acceptable to
native speakers which may turn out to be even more the case in spoken than
in written French (though the difference in mediums remains to be studied)
The learner data presented here from Anderson (2002 2007) however
provides one piece of evidence that textbook rules and classroom input do not
constitute a sufficient pedagogical norm as learners have stopped in their
progress toward acquiring native speaker (ie descriptively accurate) norms
As the articles in Blyth (2003) attest however it is highly debatable
whether the lsquoNative Speakerrsquo (as an idealized abstraction) should be the
appropriate model for the contemporary learner For Kramsch (2002 2003)
too much of a focus on an authentic native speaker norm reduces
foreign language (FL) study to a constant lsquoapproximation to and conformity
with stereotypical native speakers thus often silencing other FL-specific
forms of language potentialrsquo (2002 60) Language learning in instructional
settings on my understanding of Kramschrsquos (2003) argument should not be
viewed solely in a teleological sensemdashas language acquisition directed
toward a fixed nativelike end-pointmdashbut rather as language appropriation
by learners in the sense that they are constructing a linguistic and even
cultural identity lsquoin the interstices of national languages and on the
margins of monolingual speakersrsquo territoriesrsquo (2003 260) The political nature
of the debate as highlighted by Kramsch becomes all the more apparent
when one additionally considers the status of the foreign language on
the world stage (eg French as an international language in retreat)
and how deeply the normative attitudes of its speakers run (eg Francersquos
three-century-long history of acceptance and maintenance of the standard
(Lodge 1993))
The foregoing synopsis of the debates surrounding native speaker usage as
both the goal of foreign language instruction and the extent to which
pedagogical norms need apply has highlighted the real world complexities
involved in what would otherwise seem to be a simple equation Revise
pedagogy to reflect actual use as shown by frequency studies and this
will result in a more targetlike performance on the part of learners So we
return then to the questions posed earlier (1) Should instruction be
changed (2) If so who along the continuum from the theorist-researcher to
the teacher-practitioner should inform and guide this change (3) At which
level (4) In what ways
With respect to our first question because one can empirically
demonstrate that the pedagogical norm for French adjective position
prevents learners from eventually acquiring native speaker (ie descriptively
accurate) norms instruction should be changed so thatmdashirrespective of
whether the attainment of such a norm is necessarily desirablemdashlearners
have both a choice in selecting one grammatical or lexical form over the other
and an awareness of the meaning potential of each choice (following Kramsch
2002 71)
BRUCE ANDERSON 303
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
under Bardovi-Harlig and Gassrsquos (2002) definition because the learner is able
to produce a subset of the totality of possible utterances deemed acceptable to
native speakers which may turn out to be even more the case in spoken than
in written French (though the difference in mediums remains to be studied)
The learner data presented here from Anderson (2002 2007) however
provides one piece of evidence that textbook rules and classroom input do not
constitute a sufficient pedagogical norm as learners have stopped in their
progress toward acquiring native speaker (ie descriptively accurate) norms
As the articles in Blyth (2003) attest however it is highly debatable
whether the lsquoNative Speakerrsquo (as an idealized abstraction) should be the
appropriate model for the contemporary learner For Kramsch (2002 2003)
too much of a focus on an authentic native speaker norm reduces
foreign language (FL) study to a constant lsquoapproximation to and conformity
with stereotypical native speakers thus often silencing other FL-specific
forms of language potentialrsquo (2002 60) Language learning in instructional
settings on my understanding of Kramschrsquos (2003) argument should not be
viewed solely in a teleological sensemdashas language acquisition directed
toward a fixed nativelike end-pointmdashbut rather as language appropriation
by learners in the sense that they are constructing a linguistic and even
cultural identity lsquoin the interstices of national languages and on the
margins of monolingual speakersrsquo territoriesrsquo (2003 260) The political nature
of the debate as highlighted by Kramsch becomes all the more apparent
when one additionally considers the status of the foreign language on
the world stage (eg French as an international language in retreat)
and how deeply the normative attitudes of its speakers run (eg Francersquos
three-century-long history of acceptance and maintenance of the standard
(Lodge 1993))
The foregoing synopsis of the debates surrounding native speaker usage as
both the goal of foreign language instruction and the extent to which
pedagogical norms need apply has highlighted the real world complexities
involved in what would otherwise seem to be a simple equation Revise
pedagogy to reflect actual use as shown by frequency studies and this
will result in a more targetlike performance on the part of learners So we
return then to the questions posed earlier (1) Should instruction be
changed (2) If so who along the continuum from the theorist-researcher to
the teacher-practitioner should inform and guide this change (3) At which
level (4) In what ways
With respect to our first question because one can empirically
demonstrate that the pedagogical norm for French adjective position
prevents learners from eventually acquiring native speaker (ie descriptively
accurate) norms instruction should be changed so thatmdashirrespective of
whether the attainment of such a norm is necessarily desirablemdashlearners
have both a choice in selecting one grammatical or lexical form over the other
and an awareness of the meaning potential of each choice (following Kramsch
2002 71)
BRUCE ANDERSON 303
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
Because researchers utilizing authentic corpora are the only ones who can
empirically demonstrate incongruities between pedagogical rules and native
speaker usage they are the ones who should play the most significant role in
guiding such changes Yet they must do so following Widdowson (2000)
within a sociolinguistic framework an approach to usage that also recognizes
native speakersrsquo awareness of and opinions about such usage Stated in other
terms and following Kerr (2002) the application of corpus-based findings to
pedagogy entails asking to what extent the frequency patterns of native
speaker usage culled from native-speaker corpora in both written and
spoken mediums can help us establish a more refined pedagogical norm
rather than a pedagogical norm that is necessarily congruous with a
descriptive norm at each stage
In response to our third question Kerr (2002) has already suggested that
instruction on variant word-order constructions be delayed until the third-
year level of undergraduate instruction and beyond particularly for those
constructions in which variant word orders are correlated with preceding
contextual information (as is the present case) since learners cannot be
expected to use such constructions appropriately until their focus of attention
can go beyond the sentence frame (2002 195)
Finally with respect to our fourth question the foregoing discussion
would seem to indicate that within American universities at least the
traditional third-year undergraduate Advanced Grammar course represents
the most appropriate venue for increased attention to variant word-
order constructions which would in turn serve to transform such courses
into ones that truly are advanced and not simply remedial In order for this
transformation to take place at least with respect to French adjective
position one must first recognize that the acceptability of pre-N
versus post-N order is not a matter of grammaticality in French as it is in
English but rather a matter of permissibility given lexical and contextual
(pragmatic) factors (Waugh 1977) This entails that we should treat
differences in adjective position and any other instances of flexible yet
grammatical word order as cases of pragmatically marked and unmarked
structures
Following Dryer (1995) a particular word order can be described as
pragmatically unmarked (even if it is not the most frequently attested order)
if one can show that it is the default ordermdashthat is the order that is used
elsewhere once the contexts(s) of use of the pragmatically marked order have
been characterized (1995 103) The task of concisely characterizing contexts
in which the other word order or orders are used falls most naturally upon
the corpus-based applied linguist Anderson (2002) demonstrated that
uniqueness of the noun referent is one such situation in which pragmatically
marked pre-N adjective order is used though there are likely more contexts
to be discovered As suggested by research on data-driven language learning
(in eg Aston 2001) this same discovery task might then be mimicked
304 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
as the primary pedagogical activity for instructors and learners to engage in
within a truly advanced grammar course
Final version received May 2006
SUPPLEMENTARY MATERIAL
The supplementary material mentioned in the text is available to online
subscribers at httpapplijoxfordjournalsorg
NOTES
1 Bardovi-Harligrsquos (1987) use of the
phrase lsquodistribution of data available as
inputrsquo (1987 400) is equated here with
frequency though she herself equates it
with saliency Use of either term carries
with it the assumption that what is
more frequent is more easily apperceived
(to use Gassrsquos (1997) terminology) and
thus more likely to be incorporated
into the interlanguage grammar
This assumption is clearly evident in
Bardovi-Harligrsquos claim that lsquohigh sal-
ience encourages the acquisition of a
construction before it would otherwise
be expected [on the basis of cross-
linguistic markedness]rsquo (1987 402)
2 Though a third-year remedial gram-
mar course was not observed for the
present study an examination of one
popular textbook for such courses
La grammaire a lrsquoœuvre (Barson 1996
225ndash8) offers only slightly more
information in its overview of adjec-
tive position Post-N position is again
presented as the general rule and a
list of adjectives that exceptionally
appear in pre-N position is provided
this list now contains a total of fifteen
adjectives again including some not
found in the lists provided in the first-
and second-year textbooks but exclud-
ing others The treatment of variable
adjective position is a bit more
nuanced than that found in the
other textbooks Variation in position
is again correctly equated with varia-
tion in interpretation but capturing
this change in interpretation through
the use of English glossesmdashthe sole
strategy in second-year textbooksmdashis
restricted to just six adjectives (cher
ancien pauvre propre already included
in (5) but also meme lsquosamersquo vs lsquoitselfrsquo
and seul lsquoonlyrsquo vs lsquoalonersquo) The
adjectives dernier and prochain also
already included in (5) are explained
as post-N in expressions of time
(eg le week-end dernier lsquolast weekendrsquo)
but pre-N when construed as part of a
series similar to ordinal adjectives
(eg la derniere fois lsquothe last timersquo cf
le dernier acte de cette piece lsquothe last act
of this playrsquo) Three more adjectivesmdash
certain different and diversmdashare said to
have a literal meaning in post-N
position (as lsquocertain [sure]rsquo lsquodif-
ferent [not similar]rsquo and lsquodiverse
[varied]rsquo respectively) whereas in
pre-N position all three signify a vague
number (eg certaines erreurs de calcul
lsquo[a] certain [number of] calculus
errorsrsquo)
3 Larsson (1993) limits his study to 113
adjectives having a valorisation positive
lsquopositive valuersquo because his source
material is principally travel catalo-
gues and tourist guides both of which
tend to play up the positive attributes
BRUCE ANDERSON 305
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
of the locales they are describing
The lists of adjectives from Forsgren
(1978) and Wilmet (1980) are not
limited by such lexical semantic
considerations
4 As pointed out to me by A Valdman
(personal communication 10 Decem-
ber 1999) the frequency counts pro-
vided in the three corpora synthesized
in Table 1 are potentially inaccurate
representations of the extent to which
adjectives vary in position because no
indication of the range of nouns with
which the adjectives appear is provided
Thus although an adjective like plein
lsquofullrsquo may appear roughly half the time
in pre-N position it may do so only in a
number of fixed expressions such as la
pleine lune lsquothe full moonrsquo Though this
may be so the cutoff point of 25 or
greater total attestations goes some way
in ensuring that collocations like la
pleine lune were not the only source of
variable position data Moreover in the
results I later present from Anderson
(2002 2007) the adjectives used in the
test sentences in Appendix C varied
in position with a range of nouns as
shown by a concordance run on texts
found in the ARTFL database
5 One might additionally question as
did an anonymous Applied Linguistics
reviewer why variation in adjective
position in textbooks was not also
examined during the observational
study As discussed in the following
section no classroom input corpus
could exhaustively tally the totality of
input directed at students (especially
sources of input the learner has a
choice in seeking out outside of class)
Because there was no assurance that
cultural readings such as the ones
serving as the source of the
examples in (6) actually constituted
input in or outside of the class they
were excluded from the tally If
however a cultural reading was read
silently or aloud in class the tokens of
adjective position in that reading were
included
6 Unfortunately the contexts within
which these specific instances of non-
nativelike use in student output
occurred could not be fully recorded
The researcher (himself an advanced
nonnative speaker of French) simply
determined that the use of a particular
adjective in a particular position
sounded odd given the contextual
information in the studentrsquos utterance
by placing an asterisk next to the
adjective in the appropriate adjective
position sub-quadrant on the tally
sheet
7 Such an explanation involves a base
meaning for the adjective cher as
lsquogreater than average in valuersquo with
its two polysemes determined syntac-
tically In post-N position an objective
interpretation obtains (ie greater
than average in value according to
some objective criterion such as
monetary value [frac14 expensive]) In
pre-N position a subjective interpreta-
tion obtains (ie greater than average
in value according to some subjective
criterion such as emotional value
[frac14 beloved])
8 Use of the noun phrase la valise lourde
lsquothe heavy suitcasersquo in a sentence
following a context in which the
reader has been told that there were
several suitcases would be lsquofinersquo in
semantic terms (as all that is required
is at least one such suitcase to be left)
but particularly lsquooddrsquo in pragmatic
terms (as it creates for a sentence
that is less informative than it could
be for the current purposes of the
exchange)
9 I argue elsewhere (Anderson 2002
2007) that this finding contradicts the
claim that learner grammars in con-
trast to native speaker grammars are
wholly based on lsquowhat they have
heard and how oftenrsquo (a possibility
306 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
that Bley-Vroman 2004 263 enter-
tains) This is because there is nothing
in classroom input frequency to pre-
vent generalizing the acceptability of
pre-N position of an adjective such as
lourde in (7a) from one interpretation
to the other Knowledge of syntaxndash
semantic composition in a poverty-
of-the-stimulus situation such as this
(ie knowledge to the effect that the
pairing of (7a) with a multiple noun-
referent context is the least acceptable
in truth-conditional semantic terms
despite impoverished input on the
part of L2 French learners) implies
an interlanguage grammar that is
epistemologically equivalent to that
of native speakers
REFERENCES
AndersonB2002 The fundamental equivalence
of native and interlanguage grammars Evi-
dence from argument licensing and adjective
position in L2 French Unpublished doctoral
dissertation Indiana University
Anderson B 2007 lsquoLearnability and parametric
change in the nominal system of L2 Frenchrsquo
Language Acquisition 14 forthcoming
Aston G (ed) 2001 Learning with Corpora
Houston TX Athelstan
Bardovi-Harlig K 1987 lsquoMarkedness and
salience in second-language acquisitionrsquo
Language Learning 37 385ndash407
Bardovi-Harlig K and S Gass 2002 lsquoIntroduc-
tionrsquo in S Gass K Bardovi-Harlig S S Magnan
and J Walz (eds) Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins pp 1ndash12
Barson J 1996 La grammaire a lrsquoœuvre 5th edn
Boston Heinle amp Heinle
Biber D 1988 Variation across Speech and Writing
Cambridge Cambridge University Press
Biber D 1995 Dimensions of Register Variation
A Cross-linguistic Perspective Cambridge Cam-
bridge University Press
Biber D and R Reppen 2002 lsquoWhat does
frequency have to do with grammar teachingrsquo
Studies in Second Language Acquisition 24
199ndash208
Biber D S Conrad and R Reppen 1998
Corpus Linguistics Investigating Language Structure
and Use Cambridge Cambridge University
Press
Biber D S Johansson G Leech S Conrad
and E Finegan 1999 Longman Grammar of
Spoken and Written English London Longman
Bley-Vroman R 2004 lsquoCorpus linguistics and
second language acquisition Rules and fre-
quency in the acquisition of English multiple
wh-questionsrsquo in P Leistyna and C F Meyer
(eds) Corpus Analysis Language Structure and
Language Use Amsterdam Rodopi pp 255ndash72
Bley-Vroman R and N Yoskinaga 2000 lsquoThe
acquisition of multiple wh-questions by high-
proficiency non-native speakers of Englishrsquo
Second Language Research 16 3ndash26
Blyth C S (ed) 2003 The Sociolinguistics of
Foreign-language Classrooms Contributions of the
Native the Near-native and the Non-native Speaker
Boston MA Heinle
Bragger J D and D B Rice 1996 Quant a moi
Boston Heinle amp Heinle
Cinque G 1994 lsquoOn the evidence for partial
N-movement in the Romance DPrsquo in
G Cinque J Koster J-Y Pollock L Rizzi and
R Zanuttini (eds) Paths Towards Universal
Grammar Washington DC Georgetown
University Press pp 85ndash110
Conrad S 2000 lsquoWill corpus linguistics revolu-
tionize grammar teaching in the 21st centuryrsquo
TESOL Quarterly 34 548ndash60
Delbecque N 1990 lsquoWord order as a reflection
of alternate conceptual construals in French
and Spanish Similarities and divergences
in adjective positionrsquo Cognitive Linguistics 1
349ndash416
Delmonier D 1980 lsquoLa place de lrsquoadjectif en
francais Bilan des points de vue et theories du
XXe sieclersquo Cahiers de Lexicologie 37 5ndash24
Dryer M S 1995 lsquoFrequency and pragmatically
unmarked word orderrsquo in P Downing and
M Noonan (eds) Word Order in Discourse
Amsterdam Benjamins pp 105ndash35
Ellis N C 2002 lsquoFrequency effects in language
processing A review with implications for the-
ories of implicit and explicit language acquisi-
tionrsquo Studies in Second Language Acquisition 24
143ndash88
BRUCE ANDERSON 307
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT
Eubank L and K R Gregg 2002 lsquoNews flashmdash
Hume still deadrsquo Studies in Second Language
Acquisition 24 237ndash47
Forsgren M 1978 La place de lrsquoadjectif epithete en
francais contemporain [Studia Romanica Upsa-
liensia 20] Uppsala Sweden Almqvist and
Wiksell
Francis G and J Sinclair 1994 lsquolsquolsquoI bet he drinks
Carling Black Labelrsquorsquo A riposte to Owen on
corpus grammarrsquo Applied Linguistics 15
190ndash200
Gass S 1997 Input Interaction and the Second
Language Learner Mahwah NJ Erlbaum
Gass S K Bardovi-Harlig S S Magnan and
J Walz (eds) 2002 Pedagogical Norms for Second
and Foreign Language Learning and Teaching
Amsterdam Benjamins
Greenberg J 1966 Language Universals The
Hague Mouton
Grice H P 1975 lsquoLogic and conversationrsquo in
P Cole and J L Morgan (eds) Syntax and
Semantics Vol 3 Speech Acts New York Academic
Press pp 41ndash58
Holmes J 1988 lsquoDoubt and uncertainty in ESL
textbooksrsquo Applied Linguistics 9 21ndash44
Kasper G 1979 lsquoCommunication strategies
Modality reductionrsquo Interlanguage Studies
Bulletin 42 266ndash83
Kerr B J 2002 lsquoVariant word-order construc-
tionsrsquo in S Gass K Bardovi-Harlig S S
Magnan and J Walz (eds) Pedagogical Norms
for Second and Foreign Language Learning and
Teaching Amsterdam Benjamins pp 183ndash98
Koike D A and J E Liskin-Gasparro 2003
lsquoPrivilege of the nonnative speaker meets
practical needs of the language teacherrsquo in C
S Blyth (ed) The Sociolinguistics of Foreign Lan-
guage Classrooms Boston MA Heinle pp 263ndash6
Kramsch C 2002 lsquoStandard norm and varia-
bility in language learningrsquo in S Gass K
Bardovi-Harlig S S Magnan and J Walz
(eds) Pedagogical Norms for Second and Foreign
Language Learning and Teaching Amsterdam
Benjamins pp 59ndash79
Kramsch C 2003 lsquoThe privilege of the non-
native speakerrsquo in C S Blyth (ed) The Socio-
linguistics of Foreign-language Classrooms Boston
MA Heinle pp 251ndash62
Larsson B 1993 La place et le sens des adjectifs
epithetes de valorisation positive [Etudes romanes
de Lund 50] Lund Sweden Lund University
Press
Lodge RA 1993 French From Dialect to Standard
London Routledge
Magnan S S L Martin-Berg W J Berg and
Y Rochette Ozzello 2002 Paroles 2nd edn
Fort Worth TX Harcourt College
Mazurkewich I 1984 lsquoAcquisition of dative
alternation by second language learners
and linguistic theoryrsquo Language Learning 34
91ndash109
Oates M D and J F Dubois 2003 Personnages
3rd edn Boston Houghton Mifflin
OrsquoConnorDiVitoN 1991 lsquoIncorporating native
speaker norms in second language materialsrsquo
Applied Linguistics 12 383ndash95
Owen C 1993 lsquoCorpus-based grammar and the
Heineken effect Lexico-grammatical descrip-
tion for language learnersrsquo Applied Linguistics
14 167ndash87
Ross J 1977 lsquoGood-bye to whom hello who torsquo
in the CLS Book of Squibs Cumulative Index
1968ndash1977 Chicago Chicago Linguistics
Society pp 88ndash90
St Onge S R St Onge and K Kulick 2003
Interaction Boston Heinle amp Heinle
Stubbs M 2001 lsquoTexts corpora and problems of
interpretation A response to Widdowsonrsquo
Applied Linguistics 22 149ndash72
Valdman A 1989 lsquoClassroom foreign language
learning and language variation The notion of
pedagogical normsrsquo in R Eisenstein-Miriam
(ed) The Dynamic Interlanguage Empirical Studies
in Second Language Variation New York Plenum
pp 261ndash78
Valdman A and C Pons 1997 Chez nous Upper
Saddle River NJ Prentice Hall
van Riemsdijk H 1978 A Case Study in Syntactic
Markedness The Binding Nature of Prepositional
Phrases Lisse The Netherlands Peter de Riddler
Waugh L 1977 A Semantic Analysis of Word Order
Leiden Brill
Widdowson H G 2000 lsquoOn the limitations of
linguistics appliedrsquo Applied Linguistics 21 3ndash25
WilmetM 1980 lsquoAnteposition et postposition de
lrsquoepithete qualificative en francais contempor-
ainrsquo Travaux de Linguistique 7 179ndash204
308 PEDAGOGICAL RULES AND THEIR FREQUENCY IN THE INPUT