Week 4 - Matthews & Yip (2009) Contact Induced Grammaticalization

7/27/2019 Week 4 - Matthews & Yip (2009) Contact Induced Grammaticalization

1/30

Studies in Language: (), . ./sl...mat / - John Benjamins Publishing Company

Contact-induced grammaticalization

Evidence rom bilingual acquisition*

Stephen Matthews and Virginia YipUniversity o Hong Kong / Chinese University o Hong Kong

It is widely acknowledged that developments in bilingual individuals parallel,and ultimately underlie, those taking place in the course o contact-inducedchange. In this paper we address the poorly understood relationship between theindividual and community-level processes, ocusing on the process o grammati-calization in circumstances o language contact and the corresponding develop-mental processes in bilingual acquisition. Te phenomena chosen or discussionare drawn rom Singapore Colloquial English (SCE) and rom the Hong KongBilingual Corpus (Yip & Matthews 2000, 2007). Parallel developments in SCEand bilingual acquisition are analysed as cases o contact-induced grammatical-ization as defined by Heine and Kuteva (2003; 2005), with some modifications.Te emergence o alreadyas a marker o aspect presents a case o ordinarycontact-induced grammaticalization, while the development o grammaticalunctions ogiverepresents a case o replica grammaticalization. One implica-tion o these findings is that bilingual first language acquisition is a possibleroute or substrate influence, both in general and specifically in the developmento contact languages such as pidgins and creoles.

. Introduction

It is a truth widely acknowledged that developments in bilingual individuals par-allel, and ultimately underlie, those taking place in the course o contact-inducedlanguage change: the bilingual individual is the ultimate locus o language con-tact (Romaine 1996: 573). In bilingual individuals we observe processes such ascode-switching, transer and other orms o grammatical interaction; in languagesin contact, we observe processes such as lexical borrowing, calquing and contact-

induced grammaticalization, while the outcomes include language shif, pidgini-zation and creolization. Yet the relationship between the individual and language-level processes remains poorly understood, and this is especially so in the domaino grammaticalization.


2/30


In this paper we ocus on the relationship between grammaticalization incircumstances o language contact, and the corresponding processes in bilingual

development at the individual level. Particular reerence is made to Singapore Col-loquial English (SCE) as a contact language and the development o bilingual chil-dren acquiring Cantonese and English as described in Yip and Matthews (2000,2007). We ocus on two grammatical phenomena attested in SCE which are paral-leled in bilingual development, and which we see as representative o two kindso contact-induced grammaticalization: the emergence o alreadyas a marker oaspect is a case o ordinary contact-induced grammaticalization, and the devel-opment o grammatical unctions ogivea case o replica grammaticalization asdefined by Heine and Kuteva (2003, 2005). A third case, the urther grammatical-

ization o onein relative clauses, is discussed in Yip and Matthews (2007).In Section 2, we outline a ramework or analysis in which the relevant notions

are defined and a modified model or contact-induced grammaticalization is sug-gested. Tis is ollowed by some background inormation on the bilingual childrenand data collection in Section 3. Section 4 discusses the emergence o alreadyas amarker o aspect as a case o ordinary contact-induced grammaticalization, whileSection 5 treats the development o grammatical unctions ogiveas a case o rep-lica grammaticalization. Some conclusions are drawn in Section 6.

. Contact-induced grammaticalization

Grammaticalization has traditionally been viewed as a process internal to a lan-guage, as opposed to external or contact-induced changes such as structural bor-rowing, calquing and substrate influence. Indeed, linguists have ofen debatedwhether a particular development was to be attributed to contact or to internaldevelopment through grammaticalization. Tis dichotomy is still evident in cur-rent textbooks: or example, the treatment o contact-induced change in Toma-son (2001) does not mention grammaticalization, while the treatment o gram-maticalization in rask (1996) does not mention language contact, and Winord(2003: 350) retains the traditional assumption that grammaticalization involvesgrammatical change that is internally motivated. In recent work this dichotomyhas begun to break down, due to findings in two areas in particular:

a. areal typology, where it has become clear that similar instances o grammati-

calization show strong geographical clustering, implicating mutual influenceas a actor in the process (Ansaldo 1999; Bisang 1998; Dahl 2001; Enfield 2003;Heine and Kuteva 2005);


3/30

Stephen Matthews and Virginia Yip

b. creoles and other contact languages, where it has been observed that apparentcases o grammaticalization could result rom substrate influence: what at

first sight looks like internal grammaticalization may well be due to influencerom other languages as well (Arends and Bruyn 1994: 120).

Both these points can be illustrated by the grammaticalization o the verb say.In certain linguistic areas such as West Arica and Southeast Asia, a verb origi-nally meaning say ofen serves as a complementizer meaning that. ypically thischange occurs though reanalysis o a serial verb construction, as in Cantonesewhere the verb waa6say ollows another verb such asgong2talk in (1):1

(1) Keoi5 tung4 ngo5 gong2 waa6 lei5 mou5 cin2

he with me talk say you not.have money He told me you had no money.

While the second verb waa6say appears redundant here, its structural unctionbecomes clear where it introduces a complement clause ollowing a verb such as

ji5wai4believe whose meaning does not involve speech at all:

(2) Keoi5 ji5wai4 waa6 ting1jat6 heoi3 taam3 lei5 she believe say tomorrow go visit you

She thought (that) she was visiting you tomorrow.Here waa6no longer means say but serves as a complementizer, comparable tothat. Tis grammatical use o say recurs in many Chinese dialects with differentsay verbs, which are grammaticalized as complementizers to different degrees:or example, in the case o kngsay in aiwanese Southern Min, the grammati-calization process is more advanced than in that o Cantonese waa6 (Chappell2008). Moreover, the same process applies to the verb shuosay in the variety oMandarin spoken in aiwan (Cheng 1997a). Since this is not a general property o

Mandarin Chinese, the grammatical use o shuosay as complementizer appearsto be the result o substrate influence rom aiwanese Southern Min a primaacie instance o contact-induced grammaticalization.

In the extreme case o language contact represented by creole languages, somehave questioned whether grammaticalization takes place at all in such cases. InSranan the word taki, derived rom English talk, is used as a complementizermeaning that in a similar pattern to waa6say in (2). Te model is provided byWest Arican languages such as Gbe and wi in which the complementizer thatderives rom the verb say (Plag 1995). Bruyn (1996) suggested that such cases in-

volve apparent grammaticalization, where the appearance o grammaticalizationresults rom three steps:

i. grammaticalization o item X has already occurred in language A


4/30


ii. item Y in language B is identified with item Xiii. a range o unctions is transerred rom item X to item Y

According to this model, no process o grammaticalization actually takes place inthe contact language. Rather, a term such as say in the lexifier language is identi-fied with an equivalent in the substrate language, based on the lexical meaningso these words. Stage (ii) o Bruyns model ormalizes the traditional notion ointerlingual identification, as used by Weinreich (1953) and widely adopted in thefield o second language acquisition. In step (iii), the whole range o unctions othe substrate item X, including grammatical as well as lexical usages, is transerredto the developing contact language.

A similar view is implied by the Relexification model o creole ormation, inwhich words o the lexifier language are assigned lexical entries rom the corre-sponding lexical item in the substrate language, thus taking on the grammaticalas well as lexical unctions o the substrate item. In such a model, there is no needto assume that grammaticalization takes place (Leebvre 1998: 40); instead, whatis being transerred into the creole is a lexical item with all o its unctions, thusa multiunctional lexical entry (Leebvre 2004: 180). We shall question these as-sumptions on two grounds in particular:

i. general properties o grammaticalization, such as persistence, can be seen toapply in the contact language (see especially Section 2.3);

ii. what is transerred may not be the whole multiunctional lexical entry, but asegment o the pathway or continuum o unctions associated with a singleorm. A similar point is raised by Siegel (2006: 31) who notes that morphemeswith orms rom the L2 or lexifier ofen have only some o the properties o thecorresponding morphemes in the L1 or substrate. In other words, transer isofen partial.

. Ordinary contact-induced grammaticalization

Heine and Kuteva (2003; 2005) have taken on the challenge o re-thinking gram-maticalization to take account o the role o language contact. Tey propose arelatively explicit model that makes a useul point o departure, though severalaspects o this model invite debate. Teir ordinary contact-induced grammatical-ization defines a process in which the initial steps (a) and (b) are similar to thoseo Bruyns (1996) apparent grammaticalization, but in step (c) Heine and Kuteva

invoke universal strategies o grammaticalization, implying that the usual prin-ciples o grammaticalization apply. Te process assumes a model language M anda replica language R in which a grammatical phenomenon is reproduced.


5/30


Ordinary contact-induced grammaticalization (Heine and Kuteva 2005: 81)a. Speakers o language R notice that in language M there is a grammatical cat-

egory Mx.

b. Tey develop an equivalent category Rx in language R on the basis o patternsavailable in R.

c. this end, they draw on universal strategies o grammaticalization, using con-struction Ry in order to develop Rx.

d. Tey grammaticalize category Ry to Rx.

Te act that this definition explicitly reers to what individual speakers do, as op-posed to what happens in languages (the ocus o much research on language con-tact), makes the model potentially applicable to bilingual acquisition. As it stands,

however, the definition incurs a certain monolingual bias: Speakers o language Rin step (a) seems to imply that R is the first or privileged language or the speakersconcerned.2o apply the model to bilingual acquisition we will need to replaceSpeakers o language R with Speakers with knowledge o at least two languages,M and R. Tere is also a question o metalinguistic awareness raised by notice instep (a), which we shall discuss later in connection with replica grammaticaliza-tion.

An example given by Heine and Kuteva (2005) involves aspect marking inBislama, the English-based pidgin o Vanuatu. Te Austronesian languages serv-ing as models or Bislama have a durative aspect marker, a prefix u-in the case oVetmbao (Keesing 1991: 328)

(3) Naji ng-u-xoel dram he he--dig yam Hes in the process o digging yams

In Bislama, an equivalent category is ormed by using the verb stapstay (derivedrom English stop) beore the main verb:

(4) Em i stap pik-im yam he dig- yam Hes in the process o digging yams

Tis is a case o ordinary contact-induced grammaticalization because step (c)involves a universal pathway o grammaticalization (SAY > Imperective aspect)to create a new grammatical category in language R. Te case o already as anaspect marker discussed in Section 2 below represents ordinary contact-inducedgrammaticalization o this kind.


6/30


. Replica grammaticalization

Replica grammaticalization is proposed as a special case o contact-induced gram-

maticalization in which not only is an equivalent grammatical category created inthe replica language R, but it is derived through the same pathway o grammatical-ization rom lexical to grammatical (or rom grammatical to more grammatical)as in the model language M:

Replica grammaticalization (Heine and Kuteva 2003: 539)a. Speakers o language R notice that in language M there is a grammatical cat-

egory Mx.b. Tey develop an equivalent category Rx, using material available in their own

language (R).c. o this end, they replicate a grammaticalization process they assume to have

taken place in language M, using an analogical ormula o the kind [My > Mx]:[Ry > Rx].

d. Tey grammaticalize category Ry to Rx.

We have already noted the monolingual bias o Speakers o language R in step (a),or which we substitute Speakers with knowledge o at least two languages, M andR. Similarly, or available in their own language (R) in step (b) we would substi-tute the simpler available in language R. Deeper problems with this model involvethe assumptions that speakers notice a grammatical category (step a) and repli-cate a grammaticalization process they assume to have taken place in language M(step c). Tese ormulations imply a metalinguistic awareness (in steps (a) and (c))and a historical perspective (in step c), which are available to the linguist, but not(at least not directly) available to a bilingual speaker, let alone a bilingual child. Inorder or the model to be viable, these notions will have to reormulated or glossedin such a way that it is plausible to attribute steps (a-d) to individual speakers. Forstep (a), it will suffice to say there is a pattern in language M which speakers try to

adapt in order to express similar content in language R. A more proound problemis posed by step (c): the processual nature o the ormula [My > Mx], meaning thata change has taken place in the model language (possibly centuries ago) cannotpossibly be accessible to speakers without explicit study. o identiy such a pro-cess (such as a shif rom say to that) requires evidence o a kind that or mostlanguages is not available even to linguists, who in the absence o historical re-cords can only hypothesize such changes. It is thereore not easible to assume thatspeakers replicate a grammaticalization process they assume to have taken place

in language M (step (c) in the above model; our emphasis). It is as i the speakersresponsible or the changes were budding historical linguists.3

What is available to the speaker is the implicit knowledge that a single pho-nological orm is associated with more than one unction: or example, the orm


7/30


waa6in Cantonese serves both as a verb meaning say and as a complementizerthat. We would thereore replace the analogical ormula [My > Mx]:[Ry > Rx]

in step (c) by a ormula such as [My ~ Mx]:[Ry ~ Rx], where ~ represents a con-tinuum rom a lexical sense y to a grammatical unction x, both associated withthe same phonological orm (or at least with phonologically relatable orms, to theextent that that grammaticalization is accompanied by phonological reduction).Such a continuum arises with the development rom lexical give to passive, wherethe pathway o grammaticalization is set out by Lord, Yap and Iwasaki (2002):

(5) Lexical verb: Permissive: Passive: give > allow > by

As the arrows indicate, this schema represents a historical course o development,reconstructed on the basis o comparative evidence. It is, however, possible to statethis situation in synchronic terms, thus avoiding the budding historical linguistparadox.4

A synchronic maniestation o the pathway can be ound in languages inwhich all three unctions are perormed by the same word as in the Min dialect oChaozhou (Matthews, Xu and Yip 2005).

(6) Ua tsau-ze ke i pu ts

I yesterday give 3 book I gave him a book yesterday. (lexical give)

(7) I bo ke ua ti tsi pu ts 3 not give 1 read this book He didnt let me read this book. (permissive let)

(8) Pu ts ke na boi klau book give person buy Te book has been bought already. (passive)

We have already suggested that this should be seen as a continuum rather than,say, three distinct unctions. Evidence or such a continuum comes rom interme-diate cases compatible with more than one interpretation o the give verb:

(9) Ke i ti tsee give/let 3 see one while Give (it) to him to read or a while. (lexical give) or Let him read (it) or a while. (permissive)

(10) Mai ke na lia tio l not-want let/ people catch 2 Dont let anyone catch you. (permissive) or Dont get (yoursel ) caught. (passive)


8/30


Such cases are termed bridging contextsby Evans and Wilkins (2000: 156) andincorporated as a crucial step in the grammaticalization process by Heine (2002).

We shall see similar examples o bridging contexts in the bilingual acquisition datain Section 5.3.

. Contact as catalyst and principles o grammaticalization

We have argued that in order to be viable, the model outlined by Heine and Kuteva(2003; 2005) must be reormulated without the assumption that diachronic pro-cesses are accessible to speakers. By reormulating their analogical ormula interms o synchronic polyunctionality, we come closer to the alternative models o

Bruyn (1996) and Leebvre (1998) and to the polysemy copying model consideredby Heine and Kuteva (2005: 100). We differ rom Bruyns notion o apparent gram-maticalization and Leebvres Relexification model in assuming that general prin-ciples o grammaticalization are applicable, not only to the substrate language(s)in which grammaticalization originally took place, but also to the contact languageaffected by it. Tese principles include the ollowing (Hopper 1991):

Directionality: change is overwhelmingly rom lexical to grammatical, andgrammatical to more grammatical, while changes in the converse directionare relatively rare;

Persistence: a grammaticalized orm may retain characteristics o the lexicalsource rom which it derives;

Divergence: grammaticalized orms gradually diverge rom their lexical sourc-es in orm (or example, by undergoing phonological reduction) and otherproperties;

Layering: grammaticalization introduces a new layer which coexists with old-er layers within the same unctional domain, ofen resulting in specialization.

Because these principles are relevant to the language contact phenomena at issue,we retain the notion o contact-induced grammaticalization. Recall that differentChinese dialects have grammaticalized the say complementizer to different ex-tents (Section 1). Tis is equally true o the verbs with the lexical meaning acquirewhich have developed modal and other grammatical unctions in Southeast Asianlanguages (Enfield 2003). It is thereore not the case that the set o structures andunctions associated with a grammaticalized item has been transerred wholesalerom one language to the next, as Bruyns and Leebvres models imply (see Sec-

tion 2 above). Rather, the partial equivalences established on the basis o interlin-gual identification have led to parallel developments along a single pathway, oralternatively (as we argue in Section 4.4) along a number o related pathways o


9/30


grammaticalization. We thus see contact as a catalyst driving change along path-ways o grammaticalization.

While the same general principles remain applicable, it is also possible thatcontact-induced grammaticalization in circumstances o intensive contact differsrom typical internal grammaticalization with respect to certain o these prin-ciples. One such difference involves whether grammaticalization involves filling aunctional gap. In connection with layering, Hopper and raugott observe:

ypically, grammaticalization does not result in the filling o any obvious unc-tional gap. On the contrary, the orms that have been grammaticalized competewith existing constructions so amiliar in unction that any explanation involvingfilling a gap seems out o the question. (Hopper and raugott 1993: 125)

By contrast, in contact-induced grammaticalization gap-filling is recognized asa motivating actor:

With the replication o a category on the model o another language, the replicalanguage may acquire a category or which previously there was no or no appro-priate equivalent. (Heine and Kuteva 2005: 124)

Contact-induced changes o this kind are system-altering changes (Aikhenvald2006: 19). Heine and Kuteva (2005) give many examples rom contact languages

including pidgins and creoles; such changes are to be expected in an expandingpidgin such as ok Pisin, where various grammatical categories may previouslyhave been missing. In bilingual acquisition, especially where there is a dominantlanguage, contact-induced grammaticalization may serve to fill temporary gapswhere the target language structure has yet to be acquired: we shall see evidenceor this in the case o verbal aspect (Section 4) and the passive (Section 5).

Another possible difference concerns the role o requency. In classical caseso grammaticalization as internal change, items become grammaticalized in con-

texts in which they are used with high requency. In the model proposed by Bybee(2003), such changes are conditioned by requency: as a construction becomesrequently used, it becomes a distinct, autonomous new construction, and lexicalitems associated with the construction may become semantically bleached and/or phonologically reduced. Tus the construction going to[verb] loses the senseo motion and reduces togonnain the course o grammaticalization as a markero uturity. In contact-induced grammaticalization, however, the grammaticalizedmodel already exists or bilingual speakers and can be transerred into the replicalanguage, regardless o its requency. We shall see that even relatively low-requen-

cy grammatical usages such as the passive use o give verbs in Chinese dialectscan be transerred to the replica language (Section 5).


10/30


. Data for this study

Te data or this study come rom a series o projects investigating grammaticaldevelopment in Hong Kong Cantonese-English bilingual children. Longitudinaldata or a total o six children between ages 1;03 and 4;06 are available in the HongKong Bilingual Child Language Corpus within the CHILDES archive (MacWhin-ney, 2000).5All children were exposed to both languages rom birth, and theirparents adopted the one-parent-one-language strategy in addressing the children.Te corpus is based on weekly or bi-weekly recordings o hal an hour or eachlanguage during which the child is interacting with two research assistants (eachspeaking a different language) in their daily activities, sometimes with the parents,

siblings and other caretakers present. Further details o the participants and datacollection are given in Yip and Matthews (2007).

Although corpus data are currently available or six bilingual children, thedata in this article come primarily rom the authors own three bilingual children(immy, Sophie and Alicia) or whom the longitudinal corpus data are supple-mented by diary data. Te availability o diary data or these three siblings enablesus to address structures such as the give passive (Section 5.3) which appear rarely,i at all, in the longitudinal corpus data. Te diary was kept rom 1;036;00 or

immy, 1;065;06 or Sophie, and 1;005;04 or Alicia. Te diaries include severalentries per week and were intended to complement the video and audio record-ings. Both parents were involved in recording the diary data in the two languages,although the coverage o English data was more extensive than or Cantonese.Te contexts mostly involved interaction between the child and parents at homeor occasionally away rom home. Relevant contextual inormation was given asar as possible in the diary entries. We believe that the diary data are reliable tothe extent that they are systematic: the patterns described here are all instantiatedat least three times. Such recurrent patterns imply developing competence ratherthan perormance alone. Tere is, however, inevitably a selection bias such thatunusual and non-native-like utterances are more likely to be recorded than unre-markable and well-ormed ones. For this reason, we use the diary data essentiallyor qualitative analysis, and do not base any quantitative claims on them.

. Alreadyas marker of perfective aspect

ense and aspect constitute an area in which substrate effects are commonlyidentified in contact languages, as in the case o Bislama discussed in Section 2.1above. An extreme case is offered by Singapore Colloquial English (SCE), whereBao (2005) argues or systemic transer o the whole aspectual system rom the


11/30


Chinese substrate. Te development o alreadyas a marker o perective aspectin SCE is paralleled in the bilingual data. In terms o Heine and Kutevas model,

this constitutes a case o ordinary contact-induced grammaticalization, wherebya category in the model language is replicated using equivalent material in thereplica language, but without recapitulating the same pathway o grammaticaliza-tion.

. Already in Singapore Colloquial English (SCE)

Several studies have described the use o alreadyas an aspect marker in SCE. Hoand Platt (1993) and Bao (1995; 2005) relate this usage to perective aspect mark-

ing in Hokkien and other Chinese dialects. A typical example cited in Bao (2005)is (11) where the verb washis uninflected and alreadyexpresses perective aspect:

(11) I wash my hand already I have washed/washed my hands.

Bao (2005) notes that whereas in English alreadycan precede or ollow the verb,in SCE alreadyconsistently ollows the verb, as does the perective marker le inMandarin Chinese as in (12):

(12) Wo xi-le shou I wash- hand I (have) washed my hands.

Bao concludes that the substrate source o already is unmistakable (Bao 2005:243). His account o the SCE aspectual system relies heavily on comparisons withMandarin, which itsel played at most a minor role in the ormation o SCE. In thecase o the perective aspect, however, the general argument remains valid sincesimilar models exist in the relevant substrate dialects: in particular, the aspect

marker liau, cognate with Mandarin le, is ound in the Min dialects (Hokkien andeochew) which were dominant in Singapore when SCE was developing.

In terms o Heine and Kutevas (2005) model, this would be a case o ordi-nary contact-induced grammaticalization (see Section 2.1): the Mandarin aspectmarker le(and its equivalents in other dialects such as liauin Hokkien and zo2in Cantonese) lack a lexical sense which can be extended through replica gram-maticalization as in the case ogivediscussed below (see Section 4).6Instead, thenearest equivalent in English to the perective aspectual category grammaticalized

in the substrate Chinese dialects is sought in the adverb already.


12/30


. Already in bilingual development

In her Singapore case study, (Kwan-erry 1989) describes the use o alreadyin her

bilingual child Elvoos development, as in the ollowing examples:

(13) You eat your cream already? (Elvoo 3;06)

(14) Alice ell down in the hole already. (Elvoo 3;06)

(15) Now my school is close already. (Elvoo 3;10)

(16) Te car is stop already. (Elvoo 4;00)

Kwan-erry (1989: 38) related the use o the adverb alreadyto the early acquisition

o the Cantonese perective marker zo2. In the ollowing example Elvoo producedCantonese zo2in a code-mixed sentence (17) which is syntactically parallel to anEnglish sentence (18) with already(Kwan-erry 1989: 39):

(17) Ngo5 sik6 go3 cake zo2 (Elvoo 3;09) I eat cake Ive eaten the cake already.

(18) I eat the cake already. (Elvoo 3;09)

Although it is ungrammatical to put the suffix zo2in sentence-final position as in(17), Cantonese zo2and alreadyare apparently equivalent to Elvoo, who was ob-served to use both alongside each other, providing ample evidence or interlingualidentification. In a dialogue with his mother, or example, Elvoo used the verb dietogether with already(Kwan-erry 1989: 40):

(19) Mother: O! Ngo5 sei2 zo2 laa3. Oh I die Oh! I have died! (i.e. Oh! I am dead!)

CHI: Die, die already. (Elvoo 3;08)

Elvoos English utterance is a translation o the mothers prior Cantonese utter-ance, with both zo2and alreadyin the postverbal position, suggesting that the twoorms are equivalent or him. Elvoo also used alreadywith stative verbs to denotea new situation or state reached (Kwan-erry 1989: 40):

(20) Te tongue red already, you see? (Elvoo 3;06) [Te tongue has turned red, you see?]

(21) Ze Ze is not here already. (Elvoo 3;08) [Sister isnt here anymore.]


13/30


Similarly, in our data rom Hong Kong bilingual children, all three Cantonese-dominant siblings use alreadywith uninflected verbs. Tey differ somewhat in the

position o already, and in their use o combinations o alreadywith adverbs suchas nowand all. immy uses alreadyboth beore and afer the verb, as in the ollow-ing examples rom diary data:

(22) I already eat. [pointing to plate o ruit] (immy 2;04;09)

(23) I find already the glasses. (immy 2;07;10)

Much as in SCE, the use o alreadyexpresses the perective aspect: Ive ound theglasses. Both the contexts o use and the time course o acquisition suggest that

he is ollowing the Cantonese model. Te perective marker zo2is consistently thefirst aspect marker to be acquired in Cantonese, and immy uses it productivelyby age 2:7

(24) Faan1 lai4 zo2, hei2 san1 zo2 (immy 2;01;16) come back raise body Im back, Ive got up.

Evidence that immys use o alreadyas an aspect marker is modelled on Canton-ese zo2comes rom an idiomatic usage o alreadyollowing a modal verb:

(25) I give you to eat apple. Have to cut already first. (immy 2;11;16)

Here alreadyis unexpected to the extent that the apple has yet to be cut, but cor-responds exactly to the adult usage o the perective marker zo2with a modal verb(Matthews and Yip 1994: 199) as in (26):

(26) Jiu3 cit3-zo2 sin1 need cut- first You have to cut it first.

immy also commonly uses the combination already now:

(27) I drink this already now. I drink already this. I drink this already now. [holding beer can] (immy 2;06;17)

(28) I open already now. [holding up opened present] (immy 2;07;04)

(29) I want balloon, the green one broke already now. (immy 2;07;30)

A possible model or the combination already nowis the combination [V-zo2laa3],

where zo2 is the perective aspect and laa3 a sentence particle expressing cur-rent relevance: nowarguably represents the closest approximation to this meaningavailable in English. A pair o parallel utterances by immy, recorded on the same


14/30


day as (27), suggests that immy makes the interlingual identification that is acrucial step in all models o the process:

(30) [o ather] Lei5 sik6-zo2 laa3 you eat- Youve eaten (already). [o helper] He has eat already now. (immy 2;06;17)

Here, although the verb eatitsel remains uninflected, the appearance o hasin theposition o perect auxiliary shows immy on the way to developing a target-liketense/aspect system. Similarly, between ages 2 and 3, Sophie requently uses anuninflected verb ollowed by already:

(31) She wake already. (Sophie 2;06;09)

(32) INV: Where are they? CHI: Eat already. (Sophie 2;07;22) INV: Ah, eat already. [laughs]

(33) Daddy, I ask already. (Sophie 3;00;21)

At this stage Sophie has yet to acquire the English present perect orm as in Hesgoneand the use o alreadyfills the gap, as shown by the ollowing exchange wherethe adult researcher models the target orm has gonebut Sophie persists in ollow-ing her own grammar:8

(34) INV: Where is monster now? CHI: He go already. He go already . INV: Te monster < has gone already >. Brother: < Hes dead again > [ [/] he go already the monster. (Sophie 2;10;21)

In addition to laa3which is comparable to sentence-final lein Mandarin and liauin Hokkien, Cantonese has the particle saai3indicating completion or exhaustiveeffect. Te combination [V saai3 laa3] as seen in (35) appears to underlie the co-occurrence o alland alreadyin Alicias and Sophies English:

(35) Jam2 saai3 laa3! drink all Ive drunk it all! [holding up glass] (Alicia 2;05;18)

(36) Daddy drink all already? (Alicia 2;08;29)

(37) CHI: He eat the he eat all already. (Sophie 2;09;24) INV: He has < eaten > [/] eaten up all o the ood. Okay


15/30


A precocious example o impromptu translation by Alicia shows how she treatsthe construction [V saai3 laa3] as equivalent to already, thus making the interlin-

gual identification necessary or contact-induced grammaticalization (step (b) inSection 2.1):

(38) Father: Its dark already CHI: hak1 saai3 laa3! dark all Its all dark already. (Alicia 1;10;16)

Other examples confirm Kwan-errys observation that alreadyollowing an ad-jective denotes a change o state:

(39) [coming in wearing pink dress] I today wear pink. I today wear pink. [later, re-appearing in red dress] I all red already. (Alicia 2;09;05)

Here the child is pointing out a change o state: instead o pink she is now wearingred. Tis is again consistent with the semantics o the Cantonese perective aspectmarker zo2, which gives an inchoative sense when used with a stative verb (Francisand Matthews 2005).

o summarize, the bilingual childrens development recapitulates the develop-

ment o alreadyas a marker o perective aspect in SCE: alreadyis used in postver-bal or clause-final position to express perective notions such as completion andchange o state. Both in SCE and in the bilingual data, the question o the modelis rather more complex than in cases o replica grammaticalization such as that o

give(Section 5). In SCE, the aspect markers le(in Mandarin) and/or liau(in Hok-kien) serve as the models; in Cantonese, the models include [V zo2 laa3] and [Vsaai3 laa3], all o which are grammatically optional: that is, there are variants suchas [V saai3] and [V laa3]. Nevertheless there is strong evidence that the childrenare ollowing the Cantonese models, including:

i. parallel utterances and translations as in (19) and (30);ii. language-specific details such as the use o perective marking with a modal

verb as in (25) ollowing the Cantonese model (26);iii. the inchoative use o alreadywith adjectives as in (20) and (39).

. Give-passives and replica grammaticalization

Replica grammaticalization is a special case o contact-induced grammaticaliza-tion where the development o grammatical unctions in a lexical item ollowsthe same pathway o grammaticalization as in the model language (Section 2.2).


16/30


Tis process may be illustrated by the development o grammatical unctions ogive verbs in various Chinese dialects. Cantonese instantiates the three stages il-

lustrated or Chaozhou in Section 2.2 above:Cantonese bei2give > permissive > passiveIn aiwanese southern Min (sao 1988), in addition to the passive, the permissivegives rise to a causative usage as in (40):

aiwanese hogive > permissive > passivepermissive > causative

(40) I h go chin siong-sim (Cheng 1997b: 203)

he give me truly sad He made me sad.

Replica grammaticalization (as defined by Heine and Kuteva but with modifica-tions as outlined in Section 2.2 above) can apply to any segment o these variouspathways. For example, the progression [give > permissive > causative] can spreadin the orm o a polysemy [give ~ permissive ~ causative] or (more perspicuously,using lexical concepts) [give ~ allow ~ make]. A bidialectal speaker could, or ex-ample, apply the set o meanings and unctions associated with hogive in aiwan-

ese to bei2give in Cantonese. Te result would be that in this speakers Cantoneseusage, bei2 is grammaticalized, having causative unctions alongside its lexicalsense o give and other grammaticalized unctions.

A striking eature o these cases is the continued existence o the give verbitsel alongside its grammaticalized counterparts. It is characteristic o grammati-calization in Southeast Asian languages that the original lexical item remains inuse alongside its grammaticalized counterpart. Tis relates to the typology o theselanguages (their typological poise in the sense o Enfield 2003) in which (a) a-fixes are ew and dispreerred, which does not avour the grammaticalized itembecoming an affix, and (b) lexical tone must be preserved on all syllables, prevent-ing phonological reduction o the grammaticalizing orms (Ansaldo 1999). Tecontinued existence o the lexical verb thus inhibits divergence o the grammati-calized morpheme rom the source lexical item. At the same time, it may acilitateinterlingual identification on the basis o the lexical meaning: this in turn wouldhelp to explain why the Southeast Asian linguistic area is characterized to such anextent by recurrent patterns o grammaticalization (Heine and Kuteva 2005: 203;Matthews 2006b: 225).

Finally, the continued existence o the lexical verb makes it difficult to main-tain that grammaticalization merely involves reanalysis, as suggested by Leeb-

vre (1998: 42). Grammaticalized verbs have acquired new unctions, but havenot necessarily been reanalyzed: or example, Cantonese bei2and Chaozhou ke


17/30


give remain verbs in their permissive and arguably even in their passive unctions(ang 2001; Matthews, Xu and Yip 2005).

. Give-passives in Malay contact varieties

Give-passives recur in several varieties o Malay, again with a number o differentgive verbs. In at least some o these varieties it is clear that Chinese, specificallySouthern Min and/or Cantonese influence is involved. Baba Malay, or example,has a clear Hokkien substrate and uses kasigive in passives (Ansaldo and Mat-thews 1999):

(41) Lu punya avourite girl nanti kasi lain orang book out you avourite girl later give other man book out Your avourite girl will be booked out by another man.

In Kedah Malay the verb bagiis used similarly (Yap and Iwasaki 2003):

(42) Rumah kita habis bagi api jilat! house we finish give fire lick Our house was completely licked by the fire!

A particularly Sinitic property o these give passives is that the agent phrase (apifire in (42)) is obligatory. Tis can be attributed to retention o the subcategoriza-tion o the lexical source verb give as a three-place predicate (Matthews, Xu andYip 2005). It is thereore a case o persistence, a puzzling but pervasive eature ogrammaticalization (see Section 2.3). Note also that in Malay, as in Chinese, di-erent verbs are used in each dialect; actual borrowing o orms is the exception tothe rule that what is spreading is the pattern o grammaticalization.

. Tegive-passive in Singapore Colloquial EnglishBao and Wee (1999: 5) describe two passive constructions in SCE,give-passive asin (43) and the kenapassive (where kenais borrowed rom Malay) as in (44):

(43) John give his boss scold. John was scolded by his boss

(44) Te durian kena eat by him already Te durian has been eaten by him

As in the southern Chinese dialects discussed above, the agent his bossin (43) isrequired in thegivepassive, whereas in the kenapassive it is optional. Tere is thuslayering and specialization in the domain o the passive. Te agent requirement


18/30


may be attributed to the give passives in the Chinese substrate dialects as in Hok-kien hor(as in (45), Bao and Wee 1999: 7) and Cantonese bei(46):9

(45) Ah Hock tapai hor lang me Ah Hock always give people scold Ah Hock always gets scolded by people.

(46) Keoi5 seng4jat6 bei2 jan4 laau6 s/he always give people scold He is always being scolded.

While it is clear that the substrate Chinese dialects somehow underlie the givepassive, it is not obvious whether this is a case o simple calquing or o contact-induced grammaticalization. Some observations on the SCEgivepassive made byBao and Wee suggest that the unctions o the substrate items are not simply trans-erred en masse:

We suggest that hor/beiandgivemay have undergone different degrees o gram-maticalization. Te lexical meaning o give or hor/beirequires an animate subject.In Hokkien or Cantonese, hor/beiis ully grammaticalized, losing the meaning oanimacy. In SgE,giveis only partially grammaticalized. As such, it is less produc-tive, and retains the animacy requirement. (Bao and Wee 1999: 8)

Te requirement or an animate subject is one that naturally applies to permis-sive unctions: the sense allow onesel to be scolded implies responsibility or theaction. In SCE, however, the requirement or an animate subject extends to the

give-passive. Tis is evidence o persistence, the phenomenon whereby the lexi-cal source o a grammaticalized item constrains its grammatical unctions (Hop-per and raugott 1993: 3) and thus consistent with Heine and Kutevas accountin which general principles o grammaticalization (such as persistence) apply tocontact-induced grammaticalization as they do to grammaticalization in general

(see Section 2.1). Tis in turn argues or retaining the term grammaticalizationin such cases, and is consistent with the contact-as-catalyst model outlined in Sec-tion 2.3.

. Ontogenetic grammaticalization o givein bilingual children

Parts o the pathway o grammaticalization discussed above are paralleled in thelanguage development o individual children. We may reer to this phenomenon

as ontogenetic grammaticalization as discussed by Ziegeler (1997), or idiogram-maticization as defined by Muwene (2008: 1734): the innovation whereby an in-dividual speaker uses an item in a new, incipiently grammatical unction.


19/30


One question arising here is whether a bilingual child could use passivegivewithout also using permissive give, thereby missing a step in the process. I we

explain the phenomena in terms o the traditional notion o calquing, there wouldseem to be no reason why this should not occur. I contact-induced grammati-calization or idiogrammaticization is taking place in ontogeny, however, then theollowing hypotheses can be derived:

i. earlier stages o grammaticalization must exist or the child, subject to an im-plicational hierarchy: passive permissive lexical give;

ii. bridging contexts must exist at the individual level in order or the child toextend lexical give to permissive, or permissive to passive unction.

Tese two hypotheses can be tested with corpus and diary data respectively. a-ble 1 shows the various uses o the verb givein the corpus data or six bilingualchildren (or details o the children and corpus, see Section 3 above and Yip andMatthews 2007).

Table 1. Frequency ogivein lexical and permissive unctions in the Hong Kong Bilin-gual Child Language Corpus

child Lexical permissive

immy 19 1Sophie 35 3

Alicia 7

Llywelyn 12

Charlotte 37 2

Kathryn 15 1

otal 125 7

It can be seen that all six children usegivein the lexical (transer) sense, and our

out o six children are ound to use it in a permissive sense at least once. A total o125 tokens o lexical giveand 7 tokens o permissivegiveare attested in the cor-pus data. Te requency o lexicalgiveis much higher that o permissivegive.Tecorpus based on regular recordings is apparently not dense enough to capture therelatively inrequent passive usage ogive, which is attested only in the diary dataor Sophie. Diary data or Sophie show that she usesgivein all the main unctionsassociated with bei2give in Cantonese, with the order o emergence being: lexical> permissive > passive. Lexical give is attested rom age 2;01:

(47) Give you one. (Sophie 2;01;17)

(48) I give you. I want to watch this one. [holding video] (Sophie 2;05)

(49) You give me that one, one only. [pointing to afer-shave] (Sophie 2;06)


20/30


Te contextual indications show that the lexical meaning o transer o possessionis intended. Te permissive use ogiveis attested rom age 3;03:

(50) You open give me see. [giving Daddy Fathers Day present] (Sophie 3;03;20)

(51) Daddy I give you see. [appearing in swimsuit] (Sophie 3;04;06)

(52) I immy dont give me to play this one, then I not be her brother [i.e. I immy doesnt let me play with this, I wont be his sister]. (Sophie 3;07;06)

Here the contexts indicate that transer o possession is not involved, but givemeans let. Passive uses ogiveare attested occasionally rom 3;03 to 4;11:

(53) [ather holds up broken pen] Tis one give immy give immy break it. (Sophie 3;03)

(54) Here is give immy scratch [points to scratched leg] (Sophie 3;06)

(55) Daddy, I already give the mosquito to bite [shows bite] (Sophie 4;09)

(56) Father: How about your coat? Child: Give Popo taken. (Sophie 4;11)

Tese examples are much less requent than instances o the permissive usage ogive, but this is equally true o child Cantonese, where only 12% o uses o bei2can be identified as passive (Wong 2003: 336). Te diary data suffice to show thatSophie extends her use ogiveto passive unctions, much as in SCE. She does soto fill a gap, in that at this stage (between ages 3 and 5) she does not command theEnglish passive. Tis can be seen rom her attempts to express passive meanings atage 5, which lack morphological or other indicators o passive voice:

(57) Last night I bite here Te mosquito bite, last night. [showing a mosquito bite] (Sophie 5;00;26)

(58) Daddy, I cut! [showing a cut in her finger] (Sophie 5;00;28)

Just as we argued with regard to Chinese dialects (Section 2.2), Sophies Englishshows evidence o bridging contexts usages consistent with more than oneinterpretation along the pathway o grammaticalization. Tus the ollowing ex-amples can be read as lexical give as in give it to me to see, or permissive let asin let me see:

(59) Give me see, give me see (Sophie 2;02)


21/30


(60) Daddy, I give you to drink caa4-caa4(i.e. tea). I give you to drink my caa4. Okay Daddy (Sophie 3;08)

Some later examples allow either a permissive or a passive reading:10

(61) Daddy, wake up. Otherwise you got nothing to eat. Give all body everybody to eat already. (Sophie 4;11)

Here Sophie is warning her ather that there will be nothing lef to eat i he doesnot get up quickly. A passive reading [Te ood] will have been eaten is possible,but this can be seen as an extension o the permissive reading Youll let everyoneeat it all (and it will be your ault), where the subject retains responsibility or the

action, as in the case o SCE discussed in Section 5.2 above.o summarize, the developmental sequence seen in Sophies English is consis-tent with contact-induced grammaticalization in:

a. Order o acquisition: the lexical stage precedes the permissive and the passiveusages ogive;

b. Mode o progression: bridging contexts mediate between the various senses inwhichgiveis used.

Under the view o contact as catalyst (Section 2.3), it is not predicted that a bilin-

gual child should transer the whole range o uses o Cantonese bei2to Englishgive, as Sophie did. Rather, we expect to see segments o the pathway undergoingtranser. Consistent with this prediction, immy and Alicia used permissivegive,but there are no clear examples o passivegive:

(62) Ghost have dinner, and give ghost eat dinner and (immy 2;09;02)

(63) How about a apple? I give you to eat apple. Have to cut already first. Apple is this one or me, skin or you. (immy 2;11;16)

(64) Give me go there [pulls Sophies hand] (Alicia 2;09;19)

immy has no need o thegivepassive as a gap-filling strategy, since unlike Sophie,he produces target-like passives as early as 3;02:

(65) My space rocket is crashed up. Te house was crashed. Tis is broken. Tiswas crashed. Tis was crashed by the car. [reerring to Lego models]

(immy 3;02;17)

Some relevant questions here concern monolingual Cantonese development: to

what extent does the ontogenetic development o bei2recapitulate historical de-velopments (see Ziegeler 1997)? Wong (2003; 2004) shows that the developmentalsequence or bei2in monolingual Cantonese is as ollows: transer (lexical give) >


22/30


permissive > dative > passive. While this sequence is consistent with ontogeneticdevelopment recapitulating grammaticalization, Wong (2004: 337) notes that it

also matches the requency o the respective constructions, so that other explana-tions are possible, including those based on input requency and syntactic com-plexity. For Wongs younger children (1;112;08), the lexical sense transer makesup 69%, permissive 12%, and passive only 1% (the remaining 11% is accountedor by dative giveas discussed in 5.4 below). For older children (2;043;02) thepercentage o permissive rises to 39% while the passive usage remains inrequentat around 2%.

In the bilingual data, the childrens Cantonese exhibits all the relevant unc-tions o bei2give as well as bridging contexts such as the ollowing, which appears

intermediate between a permissive and passive interpretation:

(66) Lei5 siu2sam1 m4 bei2 jan4 zuk1 lei5 aa3! (immy 2;09;01) you careul not give people catch you Be careul not to let people catch you/Be careul not to get caught.

Although various translations are possible, there would seem to be no way todistinguish permissive and passive readings here (semantically, the utteranceis vague rather than ambiguous). Indeed, such examples suggest that the unc-

tional continuum posited in Section 2 also exists as a continuum in the childrensgrammar.

. Dative constructions with bei2give

Apart rom its permissive and passive unctions, Cantonese bei2give is grammat-icalized as a dative marker (arguably becoming a preposition). Tis occurs whenbei2appears as the second verb o a serial verb construction (Matthews 2006a: 77)as in (67):

(67) Lei5 waan4 aan1 bei2 ngo5 you return back give me Give [it] back to me.

Tis usage represents a separate pathway o grammaticalization rom that whichleads to the permissive and passive unctions already discussed, in which it is bei2as thefirstverb in a serial verb construction that undergoes grammaticalization.Lord, Yap and Iwasaki (2002) recognize a second pathway o grammaticalization

here: lexical give > dative > beneactive. Te Cantonese-dominant bilingual chil-dren transer this usage too. Examples where they repeat the same message in bothlanguages clearly demonstrate the equivalence between Englishgiveand Canton-ese bei2in this unction:


23/30


(68) I cut will give you [cutting water melon]. Lei5 tsiah, ngo5 zoi6 cit3 bei2 lei5 sik6. you eat I again cut give you eat Eat it, and Ill cut some more or you to eat. (immy 2;11;21)11

(69) Po4po2 buy give me. Po4po2 maai5 bei2 ngo5 ge3. grandma buy give me Grandma bought these or me. (Alicia 3;02;26)

Another example suggests that Alicia treats giveas equivalent to the prepositionfor:

(70) Daddy, this one give you. [presents dish o toy ood] Daddy, this one or you. (Alicia 3;04;04)

A similar case is recorded in Sophies diary data, where the mothers comment onher use ogiveprompts Sophie to reormulate her point usingfor:

(71) Child: Tis is give Mummys. Mother: Keoi5 cyun4bou6 waa6 give ge3. she everything say give

She says give or everything. Child: Tis one or Mummy, Daddy. (Sophie 3;07;09)

Alicia nevertheless continues to use dativegive, especially in serial verb construc-tions parallel to the Cantonese as seen in (67):

(72) You know, I got a Swan Lake book give Lulu. (Alicia 4;05;15)

Later examples include the copular construction [Its giveNP] which is no longera serial verb construction but suggests a more advanced stage o grammaticaliza-

tion: (73) Father: What have you got? [A carries bag containing stamps] Child: Its give Lulu. Later. On Saturday. (Alicia 4;00;30)

Te equivalent construction in Cantonese uses bei2 give without the copularverb:

(74) Ni1 di1 bei2 lei5, ni1 di1 bei2 lei5 laa1 (Alicia 3;02;01)

this give you this give you Tis is or you, this is or you.


24/30


Te case o dative give,alongside the permissive and passive giveas already dis-cussed, illustrates an important general point: different pathways o grammatical-

ization may be based on the same verb, occurring in different syntactic environ-ments. Tis is a common pattern especially in isolating languages o Southeast Asia,as noted in Section 3.

. Conclusions

We have shown that a range o developmental phenomena in bilingual childrenare compatible with Heine and Kutevas model o contact-induced grammatical-

ization, subject to the modifications introduced in Section 2.1 and Section 2.2:in particular, the model cannot plausibly reer to speakers knowledge o gram-maticalization processes. Instead, what is transerred must involve synchronicallyidentifiable patterns o polyunctionality.

Tese parallel developments may also be compatible with other models suchas the traditional notion o calquing, polysemy copying as discussed by Heineand Kuteva (2005: 100) and Relexification (Leebvre 1998). We have pursued thegrammaticalization account because there is evidence that general principles o

grammaticalization apply (such as persistence: see Section 2.3 and Section 4.2),and because the model makes the necessary steps more explicit than the tradition-al notion o calquing. Whether we term these phenomena apparent grammati-calization or contact-induced grammaticalization may ultimately be a matter oterminology. What we have shown is that bilingual children can and do replicatethe process by which certain grammatical patterns spread across languages: theprocess by which substrate influence o Chinese dialects affects Singapore Col-loquial English (SCE), and by which patterns o polyunctionality such as the epi-demic o acquire modals spread throughout Southeast Asia (Enfield 2003). Morespecifically, we have verified several parts o the process as described by Heine andKuteva (2005), including:

Interlingual identification: the childrens parallel usage as in the case o already(Section 4.2) and dativegive(Section 5.4) demonstrates the perceived equiva-lence between the model and replica languages;

Intermediate steps: childrens development shows that give-passives developvia permissive usages, mediated by bridging contexts (see Section 5.3, c. Hei-

ne and Kuteva 2005: 102); Gap-filling: the children create perective (Section 4.2) and passive orms(Section 5.3) to plug gaps where they have yet to acquire the target language


25/30


strategies. Gap-filling is argued to be a motivation or contact-induced gram-maticalization, especially in contact languages (Heine and Kuteva 2005: 124).

One implication o these findings is that bilingual first language acquisition is apossible route or substrate influence, both in general and specifically in the devel-opment o contact languages such as pidgins and creoles. Parallel phenomena inbilingual development and in SCE illustrate this possibility. In making these pointswe are not arguing that SCE is a creole, or that the bilingual children are develop-ing a home creole. While SCE has been in act considered almost a creole (Hoand Platt 1993; Gupta 1994), this is no longer an issue to the extent that creoles areno longer seen as a structurally distinct class o languages (Corne 1999; Muwene

2001). Te mechanisms o interaction between English and the substrate gram-mars remain the same, whether we consider the resulting language to be a varietyo English, an English-lexifier creole, or some other orm o mixed language.

A major difference between the bilingual acquisition data and the Singapor-ean counterparts involves resolution and differential outcomes. From age 5, thebilingual children attended international schools where the teachers and many othe children are more or less monolingual native speakers o English. In this envi-ronment, grammatical peculiarities such as Sophies give-passive and uninflected

verbs with alreadygive way to more standard English voice and tense/aspect orms.Te contact-induced eatures that occur as developmental stages in these childrennever develop into a contact language. But what would happen given a whole com-munity o similar children? Just such a community is thought to have given rise toSCE, which has its origins in English-medium schools in ethnically mixed districtsincluding Eurasians, Jews, Armenians and Straits Chinese (Gupta 1994: 33). Teteachers included Chinese and Eurasians whose English may have been Chinese-influenced. Te Straits Chinese spoke Baba Malay which was itsel heavily influ-enced by Hokkien (Ansaldo and Matthews 1999). In such a social environment,

eatures which develop through interactive development in bilingual and multilin-gual children can eed into the eature pool o a developing contact language.

Notes

* Tis research has been ully supported by the Research Grants Council o the HongKong Special Administrative Region, China (project nos. HKU336/94H, CUHK4002/97H,CUHK4014/02H and CUHK 4692/05H) and direct grants rom the Chinese University o Hong

Kong (01/02, 03/04, 06/07). Tis paper was first developed at the Research Centre or Linguisticypology at La robe University, where Alexandra Aikhenvald and Peter Bakker provided help-ul comments. An earlier version o this paper was presented at the panel on grammaticalization,reanalysis and relexification on contact-induced change at the Annual Meeting o the Society


26/30


or Pidgin and Creole Linguistics in 2006: we thank Clancy Clements and Armin Schwegleror organizing this event, Joan Bybee or her thoughtul commentary, and several participantsor their eedback, notably Dany Adone, Adrienne Bruyn, Claire Leebvre and Bao Zhiming.

Feedback rom Salikoko Muwene is greatly appreciated and hereby acknowledged. We thankour children or their contributions to this paper and all the members o our research team whohave contributed to this work, in particular Uta Lam or her technical assistance.

. Cantonese examples are given in the JyutPingRomanization system, developed by the Lin-guistic Society o Hong Kong (ang et al 2002) to meet both linguistic criteria and the con-straints imposed by computer applications. IPA correspondences are given in Matthews and Yip(1994: 400). Te numbers at the end o each syllable represent the tones: 1 (high level), 2 (highrising), 3 (mid level), 4 (low alling), 5 (high rising) and 6 (low level).

. Heine and Kuteva (2005: 237ff) distinguish between L1>L2 replication, where the speakersfirst language serves as the model and a second language as the replica language, and L2 >L1replication, where the second language itsel serves as model and the first as replica. However,this distinction still assumes first and second languages rather than simultaneous bilingualism.Te distinction also proves difficult to draw, especially since in some situations the same lan-guage serves as both model and replica language (ibid: 239).

. Te problem here is analogous to that posed by the property o persistence, which reers tothe way in which grammaticalized items apparently retain characteristics o the lexical itemrom which they developed (Hopper 1991). Te paradox is that speakers cannot know that sucha development has occurred. Te challenge is to explain these phenomena without attributing

such knowledge to speakers o the language. Either the properties o the grammaticalized itemare somehow ossilized, or there must be some ontogenetic mechanism which recapitulates ormimics aspects o the historical development (Ziegeler 1997).

. Tose recognizing this problem sometimes appeal to a notion o panchrony intended tosubsume diachronic and synchronic perspectives. We avoid this concept since it circumventsthe problem rather than tackling it head-on, and introduces intractable new problems (New-meyer 1998: 284). We believe that the Jakobsonian insight o diachrony-in-synchrony can bemore effectively pursued by distinguishing diachronic processes rom (a) their synchronic re-flexes, and (b) their ontogenetic counterparts (see also note 3).

. Te longitudinal corpus data are available at the CHILDES database http://childes.psy.cmu.edu/under the Bilingual older.

. Te Malay sudahalready may be regarded as an adverb tending to grammaticalize into amarker o perective aspect. o the extent that Malay sudahserves as an additional model oralreadyin SCE (c. Malay kenawhich provides one passive orm, see 5.2), replica grammatical-ization could be involved.

. immy places the aspect marker zo2afer the verbal complex (faan1 lai4 zo2,hei2 san1 zo2),whereas in adult usage it would attach to the first verb o each complex (faan1-zo2 lai4,hei2-zo2

san1). Tis developmental stage is also attested in monolingual children.

. Te use o right-dislocation as in he go already the monster(34) is highly productive in bothchild and adult Cantonese. Right-dislocation is also possible in English, as in Hes already gone,the monster; the requency o the phenomenon suggests Cantonese influence, although the
http://childes.psy.cmu.edu/http://childes.psy.cmu.edu/http://childes.psy.cmu.edu/http://childes.psy.cmu.edu/


27/30


congruence between the English and Cantonese structures no doubt avours its use by the bi-lingual children.

. Te passive marker is commonly spelt horin Singapore ollowing English orthography, re-flecting the Hokkien pronunciation [h] with an open vowel.

. Note that the use o the infinitive in giveeverybody to eatis consistent with the passivereading, since some o Sophiesgive-passives use the infinitive, as in I already give the mosquitoto bitein (55).

. immy uses the word tsiaheat rom the Chiu Chow dialect as spoken by his grandmother.Tis is one o the ew Chiu Chow words that he uses, ofen or jocular effect.

References

Aikhenvald, Alexandra Y. 2006. Grammars in contact: A cross-linguistic perspective. In Gram-mars in contact, Alexandra Y. Aikhenvald & R.M.W.Dixon (eds.), 166. Oxord: OUP.

Ansaldo, Umberto. 1999. A typology o comparatives in Sinitic: Areal typology and patterns ogrammaticalization. Ph.D. dissertation, Stockholm University.

Ansaldo, Umberto & Stephen Matthews. 1999. Te Min substrate and creolization in Baba Ma-lay.Journal of Chinese Linguistics27: 3868.

Arends, Jacques & Adrienne Bruyn. 1994. Gradualist and developmental hypotheses. In Pid-

gins and Creoles: An introduction, Jacques Arends, Pieter Muysken & Norval Smith (eds.),111120. Amsterdam: John Benjamins.

Bao, Zhiming. 1995. Already in Singapore English. World Englishes14: 181188.Bao, Zhiming. 2005. Te aspectual system o Singapore English and the systemic substratist

explanation.Journal of Linguistics41: 237267.Bao, Zhiming & Lionel Wee. 1999. Te passive in Singapore English. World Englishes18: 111.Bisang, Walter. 1998. Grammaticalization and language contact, constructions and positions.

Te limits ofgrammaticalization, Anna Giacalone & Paolo Ramat (eds.), 1358. Amster-dam: John Benjamins.

Bruyn, Adrienne. 1996. On identiying instances o grammaticalization in creole languages.

Changing meanings, changing functions: Papers relating to grammaticalization in contactlanguages, Philip Baker & Anand Syea (eds.), 2946. London: Westminster University.

Bybee, Joan. 2003. Mechanisms o change in grammaticization: Te role o requency. Te hand-book of historical linguistics, Brian D. Joseph & Richard D. Janda, 60223. Oxord: Black-well.

Chappell, Hilary. 2008. Variation in the grammaticalization o verba dicendi rom complemen-tizers in Sinitic languages. Linguistic ypology12: 4598.

Cheng, Robert L. 1997a. aiyu yu aiwan Guoyu li de ziju jiegou biaozhi jiang yu kan (Tecomplementation markers say and see as complementizers in aiwanese and aiwanMandarin). In aiwanese andMandarin structures and their developmental trends in ai-

wan, Vol. II., Robert L.Cheng (ed.), 10532. aipei: Yuan-Liou Publishing.Cheng, Robert L. 1997b. Causative constructions in aiwanese. In aiwanese and Mandarin

structures and their developmental trends inaiwan, Vol. II., Robert L. Cheng (ed.), 201251. aipei: Yuan-Liou Publishing.


28/30


Corne, Chris. 1999. From French to Creole: Te development o new vernaculars in the Frenchcolonial world. London: University o Westminster Press.

Dahl, sten. 2001. Principles o areal typology. In Language typology and language universals:

An international handbook, Martin Haspelmath; Ekkehard Knig; Wul sterreicher &Wolgang Raible (eds.),14561470. Berlin: Mouton de Gruyter.

Enfield, Nicholas. 2003. Linguistic epidemiology: Semantics and grammar oflanguage contact inmainland Southeast Asia. London: Routledge Curzon.

Evans, Nicholas & David Wilkins. 2000. In the minds ear: the semantic extensions o perceptionverbs in Australian languages. Language 76: 54692.

Francis, Elaine & Stephen Matthews. 2005. A multi-dimensional approach to the category verbin Cantonese.Journal of Linguistics 41: 267305.

Gupta, Anthea Fraser. 1994. Te step-tongue: Childrens English in Singapore. Clevedon: Mul-tilingual Matters.

Heine, Bernd. 2002. On the role o context in grammaticalization. In New Reflections on Gram-maticalization, Ilse Wischer & Gabriele Diewald (eds.), 83101. Amsterdam: John Benja-mins.

Heine, Bernd & ania Kuteva. 2003. On contact-induced grammaticalization. Studies in Lan-guage27: 52972.

Heine, Bernd & ania Kuteva. 2005. Language contact and grammatical change. Cambridge:CUP.

Ho, Mian-Lian & David Platt. 1993. Dynamics of a contact continuum: Singapore English. Ox-ord: Clarendon Press.

Hopper, Paul. 1991. On some principles o grammaticalization. In Approaches to grammatical-ization, Vol. I., Elizabeth raugott & Bernd Heine (eds.), 1735. Amsterdam: John Benja-mins.

Hopper, Paul & Elizabeth raugott. 1993. Grammaticalization. Cambridge: CUP.Keesing, Roger M. 1991. Substrates, calquing and grammaticalization in Melanesian pidgin.

Approaches togrammaticalization, Vol. I., Elizabeth C. raugott & Bernd Heine (eds.), 315342. Amsterdam: John Benjamins.

Kwan-erry, Anna. 1989. Te specification o stage by a child learning English and Cantonesesimultaneously: A study o acquisition processes. In Interlingual Processes, Hans W. Dechert& Manred Raupach (eds.), 3348.bingen: Gunter Narr.

Leebvre, Claire. 1998. Creole genesis and the acquisition of grammar. Cambridge: CUP.Leebvre, Claire. 2004. Issues in the study of Pidgin and Creole languages.Amsterdam: John Ben-jamins.

Lord, Carol; Foong-Ha Yap & Shoichi Iwasaki. 2002. Grammaticalization o give: Arican andAsian perspectives. New reflections on grammaticalization, Ilse Wischer & Gabriele Diewald(eds.), 217235. Amsterdam: John Benjamins.

MacWhinney, Brian. 2000. Te CHILDES Project: ools for Analyzing alk, 3rd edn. Manwah NJ: Lawrence Erlbaum.

Matthews, Stephen. 2006a. On serial verbs in Cantonese. Serial verbs: A cross-linguistic typology,Alexandra Y. Aikhenvald & Robert M.W.Dixon (eds.), 6987. Oxord: OUP.

Matthews, Stephen. 2006b. Cantonese grammar in areal perspective. Grammars in contact, Al-exandra Y. Aikhenvald & Robert M.W.Dixon (eds.), 220236. Oxord: OUP.Matthews, Stephen; Huiling Xu & Virginia Yip. 2005. Passive and unaccusative in the Jieyang

dialect o Chaozhou.Journal of East Asian Linguistics14: 267298.


29/30


Matthews Stephen & VirginiaYip. 1994. Cantonese: A comprehensive grammar. London: Rout-ledge.

Muwene, Salikoko S. 2001. Te ecology of language evolution. Cambridge: CUP.

Muwene, Salikoko S. 2008. Language evolution: Contact, competition and change.London: Con-tinuum International.

Newmeyer, Frederick J. 1998. Language form and language function. Cambridge MA: Te MIPress.

Plag, Ingo. 1995. Te emergence o takias a complementizer in Sranan: On substrate influence,universals and gradual creolization. In Te early stages of creolization, Jacques Arends (ed.),113148.Amsterdam: John Benjamins.

Romaine, Suzanne. 1996. Bilingualism. In Te handbook of second language acquisition, WilliamC. Ritchie & ej K. Bhatia (eds.), 571605. New York NY: Academic Press.

Siegel, Jeff. 2006. Links between SLA and creole studies: Past and present. L2 acquisition and

Creole genesis: Dialogues, Claire Leebvre; Lydia White & Christine Jourdan (eds.), 1546.Amsterdam: John Benjamins.

ang, Sze-Wing. 2001. A complementation approach to Chinese passives and its consequences.Linguistics 39: 257295.

ang, Sze-Wing; Fan Kwok; Tomas Hun-ak Lee; Caesar Lun; Kang Kwong Luke; Peter ung& Kwan Hin Cheung. 2002. Guide to LSHK Cantoneseromanization of Chinese characters.2nd edn. Hong Kong: Linguistic Society o Hong Kong.

Tomason, Sarah G. 2001. Language contact: An introduction. Edinburgh: EUP.rask, R. L. 1996. Historical linguistics. London: Arnold.sao, Feng-Fu. 1988. Te unctions o Mandaringei and aiwanese hou in the double object and

passive constructions. Te structure of aiwanese: A modern synthesis, Robert L. Cheng &Shuanan Huang (eds.), 165202. aipei: Crane Publishing.

Weinreich, Uriel. 1953. Languages in contact. Te Hague: Mouton.Winord, Donald. 2003.An introduction to contact linguistics. Oxord: Blackwell.Wong, Kwok-Shing. 2003. Te emergence o bei2 give constructions in Cantonese-speaking

children. PhD dissertation, University o Hong Kong.Wong, Kwok-Shing. 2004. Te acquisition o polysemous orms: Te case o bei2give in Can-

tonese. Up and down the cline: Te nature of grammaticalization, Olga Fischer; Muriel &Harry Perridon (eds.), 325343. Amsterdam: John Benjamins.

Yap, Foong Ha & Shoichi Iwasaki. 2003. From causatives to passives: A passage in some East andSoutheast Asian languages. Cognitive linguistics and non-IndoEuropean languages, EugeneH. Casad & Gary B. Palmer (eds.), 419445.Berlin: Mouton de Gruyter.

Yip, Virginia & Stephen Matthews. 2000. Syntactic transer in a Cantonese-English bilingualchild. Bilingualism: Language and Cognition3: 193208.

Yip, Virginia & Stephen Matthews. 2007. Te bilingual child: Early development and languagecontact.Cambridge: CUP.

Ziegeler, Debra. 1997. Retention in diachronic and ontogenetic grammaticalization. CognitiveLinguistics8: 207241.


30/30


Authors address

Stephen Matthews

School o HumanitiesUniversity o Hong KongPokulamHong Kong

[email protected]

Virginia Yip

Department o Linguistics andModern LanguagesChinese University o Hong KongShatin, New erritoriesHong Kong

[email protected]
mailto:[email protected]:[email protected]:[email protected]:[email protected]

Week 4 - Matthews & Yip (2009) Contact Induced Grammaticalization

Documents

Transcript of Week 4 - Matthews & Yip (2009) Contact Induced Grammaticalization