Download presentation
Presentation is loading. Please wait.
1
Does sentence constraint influence word recognition in bilinguals? Evidence from Event-Related Potentials and RTs Pascal E. A. Brenders, Janet G. van Hell, & Ton Dijkstra NICI / BSI
2
Does sentence context influence word recognition in bilinguals? Consider these sentences: “The father brought his sick daughter to the DOCTOR.” “De vader bracht zijn zieke dochter naar de DOCTOR.” Three factors could influence the recognition of the word DOCTOR: semantic constraint (high vs. low)semantic constraint (high vs. low) lexical aspects of DOCTOR (cognates vs. noncognates)lexical aspects of DOCTOR (cognates vs. noncognates) language of sentence context (Dutch vs. English)language of sentence context (Dutch vs. English) We will investigate these factors in an English lexical decision task
3
Van Hell (1998) 1 Does sentence context influence word recognition in bilinguals? Task: English lexical decision Participants: Dutch-English bilinguals Manipulations: Semantic constraint:high vs. low Semantic constraint:high vs. low Cognate status:cognates vs. noncognates Cognate status:cognates vs. noncognates Language of sentence and target was always English (L2)
4
Van Hell (1998) 2 Stimuli: Cognatesdoctor(L1:dokter) Cognatesdoctor(L1:dokter) Non-cognatesflower(L1:bloem) Non-cognatesflower(L1:bloem) Cognates: similar meaning, orthography, and phonology between L1: Dutch, and L2: English
5
Van Hell (1998) Results
6
Van Hell (1998) Conclusions Conclusions Cognate effect in L2 remains in low constraint sentences Cognate effect in L2 remains in low constraint sentences Cognate effect in L2 remains in isolation Cognate effect in L2 remains in isolation In high constraint sentences cognate effect in the right direction, In high constraint sentences cognate effect in the right direction, but not significant but not significant
7
Current study Extending van Hell (1998) by introducing the following factors: semantic constraint:isolation semantic constraint:isolationhighlow cognate status:cognates cognate status:cognatesnoncognates language of sentence context:L1 L2 language of sentence context:L1 L2 L2 L2 measuring RTs and Event-Related Potentials measuring RTs and Event-Related Potentials
8
Behavioral experiment experimentTask English lexical decision (L2) Participants N = 22 Dutch-English bilinguals MeasurementsRTs
9
Stimuli Word stimuli: 56 English words and 56 matched pseudowords. All words were matched on length and log frequency Cognates Noncognates doctorapplefinger (n=28) (n=28)flowerbikemirror(n=28)
10
Sentence contexts English language Dutch language High constraint The father took the ill child to the doctor De vader ging met het zieke kind naar de doctor Low constraint The mother made an appointment with the doctor De moeder maakte een afspraak met de doctor Four sentence contexts were constructed for all 112 word stimuli:
11
Rapid Serial Visual Presentation mode (RSVP mode) Each word was presented for 345 ms, followed by a 300 ms blank (SOA = 645 ms). English lexical decision on sentence final words that were followed by a dot.
12
Behavioral results Isolated targets Cognate status effect in isolation (third presentation)
13
Behavioral results Sentence context 1 No main effects of language of sentence context No main effects of language of sentence context Semantic constraint effect Semantic constraint effect Cognate status effect Cognate status effect
14
Behavioral results Sentence context 2 Semantic constraint effect Semantic constraint effect Cognate status effect Cognate status effect
15
Conclusions behavioral experiment For isolated words, we found cognate facilitation, For isolated words, we found cognate facilitation, replicating the well known cognate effect from the literature replicating the well known cognate effect from the literature Cognate effects also in sentence contexts Cognate effects also in sentence contexts No main effects of language of sentence context No main effects of language of sentence context Faster RTs for targets at the end of high constraint sentences than Faster RTs for targets at the end of high constraint sentences than for low constraint sentences for low constraint sentences Semantic constraint and language of context have no impact on Semantic constraint and language of context have no impact on cognate effects cognate effects
16
ERP experiment experimentTask Silent reading Participants N = 26 Dutch-English bilinguals MeasurementsERPs
17
Method ERP experiment: Electrode placement on the scalp Recording parameters of the ERP study: EEG was recorded from 27 electrodes, Referenced to the left mastoid and digitized on-line with a sampling frequency of 200 Hz using a 12 bit A/D converter.
18
Component N400Index of semantic integration N400Index of semantic integration Between 300 and 500 ms after word onset Peaking at around 400 ms (negative amplitude)
19
Results ERP experiment: Main effect: semantic constraint Semantic constraint effect
20
Results ERP experiment: Cognates and noncognates in high and low constraint sentence contexts Semantic constraint effectSemantic constraint effect Independent of lexical characteristics of the targets:Independent of lexical characteristics of the targets: cognates or noncognates
21
Results ERP experiment: Main effect: cognate status collapsed No cognate status effect
22
Results ERP experiment: Cognates and noncognates in high or low constraint sentence contexts No cognate status effectNo cognate status effect Independent of semantic constraint:Independent of semantic constraint: high constraint vs. low constraint
23
Results ERP experiment: Cognates and noncognates in English or Dutch sentences No cognate effect in English sentencesNo cognate effect in English sentences Cognate status effect in Dutch sentencesCognate status effect in Dutch sentences
24
Results ERP experiment: Cognates and noncognates in isolation Cognate status effect in isolation
26
Results ERP experiment: Cognates and noncognates in English, high and low constraint sentence contexts Cognate status effect in English, Cognate status effect in English, high constraint sentence contexts high constraint sentence contexts Cognates are integrated more easily Cognates are integrated more easily than noncognates than noncognates (facilitation by Dutch reading) (facilitation by Dutch reading) No cognate status effect in English, No cognate status effect in English, low constraint sentence contexts low constraint sentence contexts
27
Results ERP experiment: Cognates and noncognates in Dutch, high and low constraint sentence contexts Cognate status effect in Dutch, Cognate status effect in Dutch, high constraint sentence contexts high constraint sentence contexts Cognates are integrated more difficult Cognates are integrated more difficult than noncognates than noncognates (inhibition by English reading) (inhibition by English reading) Cognate status effect in Dutch, Cognate status effect in Dutch, low constraint sentence contexts low constraint sentence contexts Cognates are integrated more difficult Cognates are integrated more difficult than noncognates than noncognates (inhibition by English reading) (inhibition by English reading)
28
Conclusions ERP experiment Although not all analysis have been done we may draw the following conclusions: Semantic constraint effect (N400) Semantic constraint effect (N400) independent of lexical characteristics of target items: cognates vs. noncognates There are cognate status effects, but they depend on both the There are cognate status effects, but they depend on both the language and the constraint of the sentence language and the constraint of the sentence Cognate status effect in isolated contexts (third presentation) Cognate status effect in isolated contexts (third presentation)
29
Overall conclusions 1 RTs: Semantic constraint and language of context have no impact on cognate status effects. ERPs: The three-way interaction of language of the sentence context, semantic constraint, and cognate status indicates that cognate status effects are affected by semantic constraint and language of the sentence context after all.
30
Overall conclusions 2 Behavioral experiment experimentERPexperiment Differenttask English lexical decision (L2) Silent reading Different measurements End-process of lexical decision task On-line recordings Dissociation of RT and ERP results:
32
Results ERP experiment * p < 0.05 ** p < 0.01 Midline WindowLanguage Context Cognate status F(1,25) 300-50036.05**31.01**0.21 300-40017.51**18.57**0.53 400-50041.66**35.33**2.32 300-3507.85*12.79**2.46 350-40022.50**20.80**0.03 400-45022.62**35.45**2.01 450-50055.89**29.03**1.82
33
Results ERP experiment: Main effect: language of the sentence collapsed ? Language switching effect ?
34
Results ERP experiment: Cognates or noncognates in Dutch or English sentences Language switching effect Language switching effect Independent of lexical characteristics of the targets: cognates vs. Independent of lexical characteristics of the targets: cognates vs. noncognates, but language effect is smaller for cognates noncognates, but language effect is smaller for cognates
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.