The$acquisi+on$of$variable$word$ order$in$two6verb ... fileThe$acquisi+on$of$variable$word$...
52
The acquisi+on of variable word order in twoverb clusters in regional Dutch Varia+on in Language Acquisi+on WestFälische WilhelmsUniversität Münster Leonie Cornips Meertens Ins+tute (KNAW) /Maastricht University 10.02.2012
Transcript of The$acquisi+on$of$variable$word$ order$in$two6verb ... fileThe$acquisi+on$of$variable$word$...
The acquisi+on of variable word order in two-‐verb clusters in regional
Dutch Varia+on in Language Acquisi+on
West-‐Fälische Wilhelms-‐Universität Münster Leonie Cornips
Meertens Ins+tute (KNAW) /Maastricht University 10.02.2012
1. Introduc,on: two-‐verb clusters
Variable word order in two-‐verb clusters
AUXFIN-‐PART: 2-‐1/1-‐2 order (1) dat Jan gewerkt hee)/ hee) gewerkt that Jan worked has/ has worked
MODFIN-‐VINF: 2-‐1/1-‐2 order (2) dat Jan lezen kan/ kan lezen that Jan read can/ can read
2. Two-‐verb clusters: Heerlen Dutch spontaneous speech corpus (cf. Cornips 1994, 2009)
loca+on of Heerlen
Heerlen Dutch corpus: 33,5 hours of recorded spontaneous speech
low level of educa,on high level of educa,on
total
language background
young old young old
‘immigrant’ dialect ‘Heerlen Dutch’
3 5 8
6 6 0
5 8 8
5 10 3
19 29 19
total 16 12 21 18 67
word order in adult corpus
Individual varia,on: 1 = finite AUXFIN-‐PART: 15: Peter
a. dus waar (...) X hee) gewoo 1-‐2 order
thus where X has livedpart
b. waar X gewoond hee) 2-‐1 order
where X livedpart has
word order in adult corpus
Individual varia,on: 1 = finite MODFIN-‐VINF: 3: Jansen a. dus die een beetje lezen kunnen thus those a bit of readinf can
b. die dat (...) redelijk kunnen opbrengen those that reasonably can yieldinf
a= 2-‐1 order b= 1-‐2 order
number of tokens in adult corpus
33.5 hours of spontaneous speech: 1.230 tokens of two-‐verb clusters where
1 = finite:
• 558 tokens of AUX/PASSFIN-‐PART
• 554 tokens of MODFIN-‐VINF
word order in adult Heerlen Dutch corpus
2.1 Word order varia,on in AUXPERF/PASS – PART CLUSTER (1= finite or non-‐finite)
VARIATION AT THE LEVEL OF THE COMMUNITY
1 = FINITE
The use of the orders 1-2 or 2-1 in the AUXPERF/PASS-PART cluster at the level of group of speakers (community) in Heerlen; 1=finite
1=FINITE
AUXPERF/PASS-‐PART tokens %
2-‐1 376 67.4 1-‐2 182 32.6
558 100
Categorical and variable use of 1-‐2 and/or 2-‐1 order in the AUXPERF/PASS-‐PART cluster in Heerlen at the
level of the individual speaker
1=FINITE n speakers (n=67)
categorical use 1-‐2 1 categorical use 2-‐1 16 variable use 1-‐2/2-‐1 50
AUXPERF/PASS-PART adult Heerlen Dutch
AUX (1) = FINITE
External factors wrt distribuAon 2-‐1 versus 1-‐2 order
• no social stra+fica+on regarding age, level of educaAon, language background
=> STABLE VARIATION
AUXPERF/PASS-PART ADULT HEERLEN DUTCH
1=FINITE
Internal factors wrt 2-‐1 versus 1-‐2 order
• no effect of nega+on, extraposed PP, tense (past versus present tense ‘have’) regarding 1FIN-‐2 or 2-‐1FIN order in AUXFIN-‐PART cluster (Cornips & Ribbert 2006, Haegeman 1998, Wurmbrandt 2006)
word order in adult Heerlen Dutch corpus
2.2 Word order varia,on in MODFIN-‐VINF cluster (1=finite and non-‐finite)
VARIATION AT THE LEVEL OF THE COMMUNITY
MOD1 = FINITE
The use of the orders 1-‐2 or 2-‐1 in MODFIN-‐VINF cluster at the level of group of speakers (community level) in Heerlen (1=finite)
1=FINITE
MOD-‐INFINITIVE tokens %
2-‐1 14 .03 1-‐2 540 99.97
554 100
MODFIN-‐VINF cluster: kunnen 'can', moeten 'must', willen 'want', mogen 'may', zullen 'shall'
The use of the orders 1-‐2 or 2-‐1 in ASPFIN-‐VINF cluster at the level of group of speakers (community level) in Heerlen (1=finite)
1=FINITE
ASP-‐INFINITIVE tokens %
2-‐1 1 .0 1-‐2 117 100
118 100
ASPFIN-‐VINF cluster: doen 'do', laten, 'let', gaan 'go' and blijven 'remain'
Categorical and variable use of 1-‐2 and/or 2-‐1 order MOD/ASPFIN -‐VINF CLUSTER cluster at the
level of the individual speaker in Heerlen
1=FINITE n speakers (n=67)
categorical use 1-‐2 55 categorical use 2-‐1 0 variable use 1-‐2/2-‐1 12
Summary: two types of verb cluster in adult Heerlen Dutch
Type 1: AUXPERF/PASS-PART INDIVIDUAL LEVEL: 50/67 SPEAKERS SHOW
VARIABLE ORDER COMMUNITY LEVEL: 67%: 2-1 ORDER
33%: 1-2 ORDER TYPE 2: MOD/ASPFIN-‐VINF
INDIVIDUAL LEVEL: 12/67 SPEAKERS SHOW VARIABLE ORDER
COMMUNITY LEVEL: 0%: 2-1 ORDER 100%: 1-2 ORDER
Research ques+ons (cf. also Smith et al. 2007)
a) Are variable forms in evidence from the start of the acquisi+on process i.e. learnt at the same +me as categorical forms?
b) What effect does input have on the child’s acquisi+on of forms, not only in terms of frequency of use but also external and internal constraints on the variability?
c) are the linguis+c condi+onings (morphology of the deepest embedded verb and type of auxiliary) of the two verb clusters -‐ MODFIN-‐VINF and AUXFIN-‐PART -‐ acquired at the same +me and in the same way?
3. Acquisi,on of two types of verb clusters
• no informa+on about caregiver input, only adult to adult interac+on
• Smith et al (2007): we suggest that a guiding principle in the acquisi+on of (socio)linguis+c competence may lie in whether the variable is a marker or an indicator (Labov 1994:78) in the community in which the child is growing up.
• HYPOTHESIS: VARIABLE WORD ORDER CAN ONLY BE ACQUIRED FROM START WHEN VARIATION SIGNALS LANGUAGE CHANGE I.E. HAS SOCIAL MEANING IN COMMUNITY.
Two types of clusters have no social meaning. Modal/Aspectual-‐Infini+ve is categorical 1-‐2 and AUX-‐PART is variable 1-‐2/2-‐1 but no effect of external variables
Child: acquisi+on of input?
Methodology (taken from Zuckerman 2001)
Subjects:
AGE NUMBER
2;8 -‐ 3;10 n=19
5;0 -‐ 5;11 n=15
LocaAon: Heerlen
Methodology (taken from Zuckerman 2001)
Procedure: The method used was a ques+on answering task (elicited produc+on), regarding 30 pictures presented to the subjects. The task of answering a ques+on was employed in order to encourage the subjects to begin their answers with Omdat 'because', which is a complemen+zer in Dutch, and by that to encourage the use of an embedded structure. All the ques+ons where therefore Why-‐ques+ons.
Example: 15 test sentences elici+ng AUXFIN-‐PART
AUX+Par+ple cluster
Picture shows Kikker with a ball in the sky (Kikker 'frog' is a known child figure).
Experimenter: "Kikker throws a ball. The ball is in the sky" ("Kikker heew de bal gegooid. De bal vliegt in de lucht")
Expected Child’s answer
Puppet: Waarom vliegt de bal in de lucht? 'Why is the ball in the sky
Expected answer: Omdat Kikker de bal gegooid heeL because Frog the ball thrown has or:
Omdat Kikker de bal heeL gegooid because Frog the bal has thrown
MOD/ASP+Infini+ve cluster
• 15 test sentences elici+ng MODFIN-‐VINF
Picture shows Kikker with skates on the ice.
Experimenter: "Kikker wants to skate. He is on the ice" ("Kikker wil schaatsen. Hij is op het ijs")
MOD/ASP+Infini+ve cluster
Puppet: Waarom is Kikker op het ijs? 'Why is Kikker on the ice'
Expected answer: Omdat Kikker wil schaatsen 1-‐2 order because Frog want skate or: Omdat Kikker schaatsen wil 2-‐1 order because Frog skate want
Results MOD+Infini+ve cluster
age n=19 age: 2;8-‐3;10
n=15 age: 5;0-‐6;0
MOD-‐INF
tokens % tokens %
2-‐1 5/56 9 2/164 1
1-‐2 51/56 91 162/164 99
Results AUX+Part. cluster
age n=19 age: 2;8-‐3;10
n=15 age: 5;0-‐6;0
AUX-‐Part
tokens % tokens %
2-‐1 4/15 27 16/84 19
1-‐2 11/15 73 68/84 81
summary children’s results
Type 1: AUXPERF/PASS-PART 2;8-3;10 5;0-6;0
2-1 27% 19% 1-2 73% 81% TYPE 2: MOD/ASPFIN-‐VINF
2;8-3;10 5;0-6;0 2-1 9% 1% 1-2 91% 99%
Comparison children-‐adults Mod+Inf cluster
age adults n=67
children n=34
MOD-‐Inf
tokens % tokens %
2-‐1 1/118 0 7/320 2
1-‐2 117/118 100 312/320 98
Categorical and variable use of 1-‐2 and/or 2-‐1 order MOD/ASPFIN -‐VINF CLUSTER cluster by the
children
MOD-‐Infini,ve n children (n=34)
n adults (n=67)
categorical use 1-‐2
23 55
categorical use 2-‐1
0 0
variable use 1-‐2/2-‐1
3 12
Comparison children-‐adults Aux+Part cluster
age adults n=67
children n=34
Aux+Part
tokens % tokens %
2-‐1 376/558 67 20/99 20
1-‐2 182/558 33 79/99 80
Categorical and variable use of 1-‐2 and/or 2-‐1 order MOD/ASPFIN -‐VINF CLUSTER cluster by the
children
AUX-‐Part 2;8-‐3;10 n=19
5;0-‐6;0 n=15
n adults (n=67)
categorical use 1-‐2
5 6 1
categorical use 2-‐1
2 1 16
variable use 1-‐2/2-‐1
0 7 50
4. DISCUSSION
RESEARCH QUESTIONS:
a) Are variable forms in evidence from the start of the acquisi+on process i.e. learnt at the same +me as categorical forms?
NO: Children start out with categorical rules where there is variability in the adult community (see AUX-‐Part verb cluster)
b) What effect does input have on the child’s acquisi+on of forms, not only in terms of frequency of use but also internal constraints on the variability?
• Frequency of variable use by adults: no effect; • Wrt modal cluster: (i) similar word order 1-‐2, (i) similar frequency -‐ categorical Wrt perfecAve cluster: (i) opposite word order, (ii) different frequencies – categorical
HOW COME?
c) Are the linguis+c condi+onings (morphology of the deepest embedded verb and type of auxiliary) of the two verb clusters acquired at the same +me and in the same way?
NO
• 4.1 Developmental stages
acquisiAonal path
De Haan (1987), Jordens (1990, 2002), Blom (2003)
stages:
1 MOD
2 MOD + INF
3 AUX + PART
acquisiAonal path
• type of verbal cluster (children in Heerlen) MOD-‐INF much frequenter than AUX-‐Part
AUX-‐PART MOD-‐INF
age 3 15/285 5% 56/285 20%
age 5 84/225 37% 164/225 73%
acquisiAonal path • Mistakes by the Heerlen children
AUX + INF
but never MOD + PART
Infini+ve acquired before past par+ciple
acquisiAonal path • expected according test design AUX-‐PART replaced by MOD-‐INF but not the reverse:
expected: omdat hij voor hee) gelezen because he has read
BUT uzered: omdat hij voor wil lezen (Frederique 4;5) because he wants read
5. Analysis
Variable forms (cons+tu+ng a linguis+c variable) in evidence are acquired from the start if varia+on displays social stra+fica+on, i.e. varia+on signals language change.
In adult speech, variable perfec+ve cluster has no social meaning
This case i.e. perfec+ve cluster doesn’t cons+tute counterevidence.
5. Analysis
AUX-‐PART MOD-‐INF
age 3 categorical (n=0) categorical
age 5 variable (n=7) categorical
=> converging towards community pazern
Default behaviour: 1-‐2 order in two-‐verb cluster i.e. modal+infini+ve cluster
Overgeneraliza+on 1-‐2 order in AUX-‐PART cluster
5. Analysis
Claudia 3;5: ADJ AUX
beetje moe was adjec+ve lew of finite verb
a bit +red was
if the verb in the passive/perfec+ve cluster is an adjec+val element, it has to undergo verb raising, resul+ng in a 2-‐1 order
5. Analysis
Developmental stages: 1) child starts out with modal+infini+ve in 1-‐2
order which is the community order 2) child has to acquire past par+cple morphology
but default order (based on modal cluster) is 1-‐2 3) at the moment child acquires par+ciple
morphology, she is able to raise the verb -‐> 2-‐1 order
4) variable order emerges as in community
