second language proficiency places cognitive constraints on sentence processing noriko hoshino...

24
Second Language Proficiency Places Cognitive Constraints on Sentence Processing Noriko Hoshino Department of Psychology The Pennsylvania State University

Post on 21-Dec-2015

218 views

Category:

Documents


0 download

TRANSCRIPT

Second Language Proficiency Places Cognitive Constraints

on Sentence Processing

Noriko Hoshino

Department of Psychology

The Pennsylvania State University

I can’t answer your question right now.

I have to concentrate on

driving.

Why was it so hard for a beginning driver to answer a question in her L2 while driving?

Small working memory capacity? Low L2 proficiency?? L1 and L2 have very different syntactic

structures???

Cognitive Resources and Sentence Processing

Within-Language Research (Just & Carpenter, 1992; Hartsuiker et al., 1999)

Individuals with limited cognitive resources are less sensitive to semantic information during on-line sentence processing than those with a high level of cognitive resources.

Bilingual Research (Hasegawa et al., 2003)

More computation and activation is required to process the L2 than the L1.

Research Question Does L2 proficiency place cognitive constraints

on sentence processing?

Examine the production of subject-verb agreement in monolingual and bilingual speakers Is the ease of access to conceptual number in the

production of subject-verb agreement modulated by L2 proficiency?

Conceptually Distributive Number Single-ReferentThe baby on the blankets

Distributive-ReferentThe label on the bottles

Single

Distributive

Mag

nit

ude

of N

um

ber

Mis

mat

ch

Magnitude of Number Mismatch = Agreement Error Rate for Mismatch - Match Control

Single

Distributive

Mag

nit

ude

of N

um

ber

Mis

mat

ch

Sensitive to conceptualas well as grammatical number

Sensitive to grammatical number alone

The baby on the blankets (singe-referent mismatch)The baby on the blanket (match control)The label on the bottles (distributive-referent mismatch)The label on the bottle (match control)

Cognitive Resources and Subject-Verb Agreement(Hartsuiker et al., 1999)

Mag

nit

ude

of N

um

ber

Mis

mat

ch

Broca's Aphasics

Single

Distributive

Mag

nit

ude

of N

um

ber

Mi s

mat

chControls

Single

Distributive

• Sensitivity to conceptual information during the process of subject-verb agreement is constrained by the degree of available computational resources.

Single

Distributive

Mag

nit

ude

of N

um

ber

Mis

mat

ch

Single

Distributive

Mag

nit

ude

of N

um

ber

Mis

mat

ch

Sensitive to conceptualas well as grammatical number

Sensitive to grammatical number alone

Predictions If individual differences in working memory capacity and in L2

proficiency have similar cognitive constraints on the process of subject-verb agreement…

High span monolingualsHighly proficient bilinguals

Low span monolingualsLess proficient bilinguals

Participants

56 English Monolinguals Higher Span (n=26) Lower Span (n=27)

Groups based on a reading span task (adopted from Waters & Caplan, 1996)

41 English-Spanish Bilinguals 30 Spanish-English Bilinguals

L1&L2

L1&L2

L1

Materials Design: Four 64-item lists in English and in Spanish 32 Critical Items:

The author of the novels (single-referent target) The author of the novel (number match control)

The drawing on the posters (distributive-referent target) The drawing on the poster (number match control)

32 Fillers: Include plural head nouns Examples:

The roads to the stores The roads to the store The rooms in the apartments The rooms in the apartment

Sentence Completion Task

+

The author of the novels

famous

+

Self-Paced

600 ms

1800 ms

RT

(Beep)

Scoring

Correct The uniform for the soldiers is white.

Agreement Error The uniform for the soldiers are white.

Eight other scoring categories

English Monolinguals by Reading Span Groups

0

2

4

6

8

10

12

14

16

Lower Span

Mag

nitu

de o

f N

umbe

r M

ism

atch

(%

)

Single

Distributive

Magnitude of Number Mismatch = Agreement Error Rate for Mismatch - Match Control

Single Target: The author of the novels Distributive Target: The drawing on the posters

Match Control: The author of the novel Match Control: The drawing on the poster

(n = 26) (n = 27)0

2

4

6

8

10

12

14

16

Higher Span

Mag

nitu

de o

f N

umbe

r M

ism

atch

(%

)

Single

Distributive

English-Spanish Bilinguals

0

2

4

6

8

10

12

14

16

Spanish (L2)

Mag

nitu

de o

f N

umbe

r M

ism

atch

(%

)

SingleDistributive

Magnitude of Number Mismatch = Agreement Error Rate for Mismatch - Match Control

Single Target: The author of the novels Distributive Target: The drawing on the posters

0

2

4

6

8

10

12

14

16

English (L1)

Mag

nitu

de o

f N

umbe

r M

ism

atch

(%

)

SingleDistributive

(N = 41) (N = 41)

Language History: English-Spanish vs. Spanish-English Bilinguals

English-Spanish

Bilinguals

Spanish-English

BilingualsAge (years) 22.2 25.5L1 Rating

(10-point scale) 9.4 9.5L2 Rating

(10-point scale) 6.5 8.3Immersion

(months) 4.3 57.4

• Spanish-English bilinguals were more proficient in L2 than English-Spanish bilinguals.

Spanish-English Bilinguals

0

2

4

6

8

10

12

14

16

English (L2)

Mag

nitu

de o

f N

umbe

r M

ism

atch

(%

)

Single

Distributive

Magnitude of Number Mismatch = Agreement Error Rate for Mismatch - Match Control

Single Target: The author of the novels Distributive Target: The drawing on the posters

0

2

4

6

8

10

12

14

16

Spanish (L1)

Mag

nitu

de o

f N

umbe

r M

ism

atch

(%

)

Single

Distributive

(N = 30) (N = 30)

Summary English monolinguals with higher reading span were

sensitive to the conceptual number of the subject phrase during the process of subject-verb agreement, whereas those with lower span were not.

Less proficient bilinguals showed sensitivity to conceptual number only in their L1, whereas highly proficient bilinguals were sensitive to conceptual number in both languages.

Discussion Individual differences in working memory capacity

and in L2 proficiency have similar cognitive constraints on the process of subject-verb agreement.

Semantic information comes to play during syntactic processing such as the process of subject-verb agreement only when individuals have available processing resources.

Although the availability of processing resources is a critical factor to influence the sensitivity to the conceptual number, are there any language constraints on the process of subject-verb agreement?

Examine the performance of Japanese-English bilinguals whose L1 does not have subject-verb agreement If the absence of a rule for subject-verb agreement in

the L1 affects performance in the L2, then Japanese-English bilinguals, even those who are highly proficient in English, may fail to demonstrate sensitivity to conceptual number.

Recent within-language (Bock, 2004) and bilingual studies (Nicol & Greth, 2003; Van Hell & Mensies, 2004) suggest that cross-linguistic differences per se do not determine sensitivity to conceptual number. However, within-language lexical and syntactic constraints may contribute to the processing of agreement.

Language History:Spanish-English vs. Japanese-English Bilinguals

Spanish-English

Bilinguals

Japanese-English

BilingualsAge (years) 25.4 29.0AOA (years) 10.9 11.9L1 Rating

(10-point scale) 9.7 9.6L2 Rating

(10-point scale) 8.6 6.9Immersion

(months) 54.0 60.2LDT Acc

Nonword (%) 86.0 81.2LDT AccWord (%) 94.5 91.8

Although L2 ratings for Japanese-English were lower than for Spanish-English bilinguals, Japanese-English bilinguals seem to have rated themselves as less proficient than they actually were.

Japanese-English Bilinguals

0

2

4

6

8

10

12

14

16

English (L2)M

agni

tude

of

Num

ber

Mis

mat

ch (

%)

SingleDistributive

(N = 19)

• Possibility 1: Significant structural differences between L1 and L2 may affect the ability to compute subject-verb agreement like a native speaker.

• Possibility 2: Distinctive structural differences may impose higher demands on processing resources in L2.

Single Target: The author of the novels Distributive Target: The drawing on the posters

Magnitude of Number Mismatch = Agreement Error Rate for Mismatch - Match Control

In ongoing research, we are examining the relation between cognitive resources, L2 proficiency, and cross-language similarity.

Acknowledgements

Paola Dussias Judith Kroll

Judith Pirela Raul Rios James Burns Natalie De Rosa

Grant Support:NSF Grant BCS-0418071 and NIMH Grant RO1MH62479 to Judith F. Kroll

Travel Support:Research and Graduate Studies Office, Department of Psychology, and Language Acquisition Graduate Organization, Penn State University