Transcript Syntactic priming in bilinguals

Syntactic Priming in Bilinguals:
Effects of verb repetition
in an L2-monolingual and cross-lingual setting
Sofie Schoonbaert1, Robert Hartsuiker1, & Martin Pickering2
1Ghent
University, Belgium
2University of Edinburgh, Scotland
ISB2005, March 20-24th, Barcelona
Some background
Speaking in a second language
 Bilingualism
 Speech production
Some background
The Architecture of BILINGUAL MEMORY
 At least ‘some’ information is shared between languages
SEMANTIC/LEXICAL
SYNTACTIC
(perception studies of Dijkstra et al., Brysbaert et al.;
production studies of Costa et al., Colome et al.)
 mainly studying semantic & lexical representations)
 ‘SHARED SYNTAX’ vs. ‘SEPARATE SYNTAX’ hypothesis
syntactic rules like:
passive-active / prepositional dative [PO]-double object
dative [DO]
 Syntactic priming in bilinguals??
Syntactic priming studies
 Speech production
 SYNTACTIC PRIMING:
tendency to repeat a recently encountered structure
In monolinguals:
(Bock, 1986, 1989; Bock & Loebell, 1990; Branigan et al., 2000; Hartsuiker & Kolk,
1998; Hartsuiker et al., 1999; Hartsuiker & Westenberg, 2000; Pickering & Branigan,
1998, 1999; Pickering et al., 2002; Potter & Lombardi, 1998)
 BOCK (1986): repeating auditory prime sentence – describing
visually presented picture with passive-active / DO-PO structures
 VERB REPETITION BETWEEN PRIME-TARGET (Pickering &
Branigan, 1998)
 lexical boost to syntactic priming
Model
(adopted from Pickering & Branigan, 1998)
combinatorial nodes
unrelated verbs
[DO]
[PO]
PRIME:
‘The chef gives a gun to the
boxer’ (PO)
TARGET verb: to send
category node
lemma stratum
lemma nodes
give
word-form stratum
send
Model
(adopted from Pickering & Branigan, 1998)
combinatorial nodes
identical verbs condition
[DO]
[PO]
PRIME:
‘The chef gives a gun to the
boxer’ (PO)
TARGET verb: to give
category node
lemma stratum
lemma nodes
give
word-form stratum
send
Syntactic priming studies
Between monolinguals in dialogue:
Confederate technique (► dialogue game; Branigan et al.,
2000)
PRIME
Match: press 1
Mismatch: press 2
‘The chef gives a
gun to the boxer’
[PO]
The chef
gives the
boxer a gun
Participant
Confederate
(L2)
Dialogue game
(introduced by Branigan, et al., 2000)
TARGET
‘- - - - ? - - - -’
Match: press 1
Mismatch: press 2
[PO or DO]
nun
swimmer
Confederate
Participant
(L2)
Dialogue game
(introduced by Branigan, et al., 2000)
Syntactic priming studies
Between monolinguals in dialogue:
Confederate technique (► dialogue game; Branigan et al., 2000)
In bilinguals:
- (Loebell & Bock, 2003;Meijer & Fox Tree, 2003)
- In a dialogue game: Hartsuiker, et al., 2004
Model
(adopted from Hartsuiker et al., 2004)
L1 prime:
Passive
(verb perseguir)
conceptual nodes
L2 response on
target picture
(verb ‘to hit’):
L2
language nodes
L1
 More Passives
than Actives in L2
lemma nodes
combinatorial nodes
category node
Present study
 syntactic priming in L2 ?
 cross-linguistic priming (L1->L2)?
• with dative structures
• with Dutch-English bilinguals in dialogue
 verb manipulation
Experiment 1 :

L2 -> L2
Looking for syntactic priming of dative structures in Dutch-English bilinguals
PRIME
‘The chef gives a
gun to the boxer’
TARGET
‘- - - - ? - - - -’
[PO]
[PO or DO]
The chef
gives the
boxer a gun
Confederate
(L2)
Participant
(L2)
Experiment 1 :
L2 -> L2
Design:
2 (DATIVE PRIME: DO - PO) x 2 (VERB TYPE: identical - unrelated)
MAIN
Syntactic priming effect !
INTERACTION
verb
repetition
enhances the
syntactic
priming effect
Prime Type
Verb Type
Identical
Unrelated
give-give
give-show
DO
PO
.55
.91
.72
.81
PRIMING
36%
9%
LEXICAL BOOST
Prop. PO
responses
Experiment 2 :

L1 -> L2
Looking for syntactic priming of dative structures in Dutch-English bilinguals
PRIME
TARGET
‘De kok geeft de
bokser een geweer’
‘- - - - ? - - - -’
[PO]
[PO or DO]
De kok geeft
de bokser
een geweer
Confederate
Participant
(L1)
(L2)
Bilingual version of a dialogue game
(introduced by Branigan, et al., 2000)
Experiment 2 :
L1 -> L2
Design:
2 (DATIVE PRIME: DO - PO) x 2 (VERB TYPE: translation - unrelated)
MAIN
Syntactic priming effect !
INTERACTION
translation
equivalence
enhances the
syntactic
priming effect
Prime Type
Verb Type
Translation
Unrelated
geven-give
geven-show
DO
PO
.74
.91
.77
.85
PRIMING
17%
8%
TRANSLATION
EQUIVALENCE BOOST
Prop. PO
responses
EXP 1
Prime Type
Verb Type
Identical
Unrelated
DO
PO
.55
.91
.72
.81
PRIMING
36%
9%
L2-L2
EXP 2
L1-L2
LEXICAL BOOST
Prime Type
Verb Type
Translation
Unrelated
DO
PO
.74
.91
.77
.85
PRIMING
17%
8%
TRANSLATION EQUIVALENCE BOOST
Discussion
Basic results :
 Syntactic priming within L2
 Cross-linguistic syntactic priming (from L1 to L2)
 Boost to syntactic priming by:
1/ repetition of the verb (36%)
2/ translation equivalent verbs (17%)
(in prime and target)
HOW COULD THIS HAPPEN ?
Model EXP1
(adopted from Hartsuiker et al., 2004)
L2
L2prime:
prime:
PO
PO
(verb
give’)
(verb‘to
‘give’)
conceptual nodes
L2response
responseon
on
L2
targetpicture
picture
target
(verb ‘to give’)
 morePO
priming
More
than
with
identical
DO
verbs
lemma nodes
combinatorial nodes
An integrated account of bilingual language representation
Model EXP2
(adopted from Hartsuiker et al., 2004)
L1
L1prime:
prime:
PO
PO
(verb
(verb‘geven’)
‘geven’)
conceptual nodes
L2response
responseon
on
L2
targetpicture
picture
target
(verb ‘to give’)
 morePO
priming
More
than
with
translation
DO
equivalents
lemma nodes
combinatorial nodes
An integrated account of bilingual language representation
Discussion
We believe that:
- the syntactic priming effect within L2 (EXP1) develops in a similar
way as in L1 (see Pickering & Branigan, 1998)
- cross-linguistic syntactic priming occurs AND was enhanced with
translation equivalent verbs
due to simultaneous activation of
--- a combinatorial node, specifying the dative structure (Pickering &
Branigan, 1998) and
--- the translation equivalent’s lemma (due to the connection between
the semantic representation and the lemma)
This activation increases the probability of selecting the same
structure with the translation equivalent (cfr. Cleland & Pickering,
2003)
General Conclusions
 Bilingualism
---> a single lexical-syntactic memory system
With…
 ‘SHARED SYNTAX’ vs. ‘SEPARATE SYNTAX’ hypothesis
 Shared concepts activating words from both languages
 Words from the non-target language influence syntactic
choice (via cascading of activation)
(with moderately proficient Dutch-English bilinguals)
 Speech production
 Reliable SYNTACTIC PRIMING across languages
- a translation equivalence boost
Thanks to you
to Rob Hartsuiker & Martin Pickering (supervising)
to the FSR-Flanders (funding)
Exp 1 : L2 -> L2
Plot of Means
significant2-way
2-way
interaction
interaction
F(1,31)=40,08; p<,0000
0,95
Prop. of Correct PO Answers
0,90
0,85
0,80
diffIdentical
VERB
show-give
same
VERB
Unrelated
give-GIVE
0,75
0,70
0,65
0,60
0,55
0,50
PO responses
DO
DO
PO
PO
PRIME
Exp 2 : L1 -> L2
significant 2-way interaction
Proportion of Correct PO answers
F(1,24)=5,70; p<,0252
PO responses
0,95
0,93
0,90
0,88
0,85
0,82
0,80
0,78
0,75
0,73
0,70
DIFFERENT verb
Translation
ex. tonen-GIVE
SAME
verb
Unrelated
ex. geven-GIVE
DO
PO
DATIVE PRIME
100%
PO
Responses
80%
100%
PO
Responses
60%
60%
40%
40%
20%
20%
0%
80%
Identical
Unrelated
DO PRIME
55%
72%
PO PRIME
91%
81%
0%
Translation
Unrelated
DO PRIME
74%
77%
PO PRIME
91%
85%
Discussion
A closer look at the results: some models
 Levelt & colleagues
Lemma = specifies syntactic proporties of a word
= contactpoint between meaning and form
(the base form of words)
 Pickering & Branigan, 1998 (extension)
(include combinatorial nodes, that are linked to/shared by the
lemma nodes)
To account for repeated verb effect in SP
(repeated verb: residual activation in both verb&combinat node;
different verb: only residual activation in the combinat node)
 Hartsuiker et al., 2004 (extension for bilinguals)
Pickering & Branigan 1998 JML
Experiment 1 :

L2 -> L2
Looking for syntactic priming of dative structures in Dutch-English bilinguals
PRIME
Match: press 1
Mismatch: press 2
‘The chef gives
the boxer a gun’
[DO]
The chef
gives the
boxer a gun
Confederate
Subject
(L2)
Monolingual dialogue game in L2
(introduced by Branigan, et al., 2000)
Experiment 1 :

L2 -> L2
Looking for syntactic priming of dative structures in Dutch-English bilinguals
TARGET
‘- - - - ? - - - -’
Match: press 1
Mismatch: press 2
[PO or DO]
nun
swimmer
Confederate
Subject
(L2)