Upload
holly-peters
View
213
Download
0
Embed Size (px)
Citation preview
ARE BILINGUALS LIKE TWO MONOLINGUALS IN ONE PERSON?
EVIDENCE FROM RESEARCHIN SENTENCE PROCESSING
Eva M. Ferná[email protected]
Queens College & Graduate Center ▪ CUNY
CUNY Academy ▪ Junior Faculty SeriesNovember 25 ▪ Rosenthal Library, Room 230 ▪ Queens College
COLLABORATION & SUPPORT
Dianne Bradley & Janet Fodor CUNY Graduate Center
Elaine KleinQueens College & Graduate Center, CUNY
Javier Sainz & Lola Oria-MerinoUniversidad Complutense de Madrid
RISLUS: Research Institute for the Study of Language in an Urban Society
CUNY Graduate Center
BILINGUAL PROCESSING
How do bilinguals process their two languages?
using strategies similar to those of monolinguals? with similar timing to that of monolinguals? with similar accuracy when the task involves it? with both written and acoustic stimuli?
¿ Bilingual (Lx, Ly) = Monolingual (Lx) + Monolingual (Ly) ?
A BILINGUAL IS…
a person who can communicate efficientlyin two codes, Lx & Ly
a person who has: underlying competence in Lx and Ly underlying differentiation of Lx and Ly
Lx
Ly
Who did you say that _ left?
Who did you say _ left?
¿Quién dijiste que _ se marchó?
¿Quién dijiste _ se marchó?
TWO COMPONENTS OR ONE?
TWO GRAMMARS
evidence: grammaticality judgments that differbetween Lx & Ly
requirement: grammaticality difference
rule in Lx rule in Ly
TWO PROCESSORS
evidence: processing preferences that differbetween Lx & Ly
requirement: processing difference
strategy in Lx strategy in Ly
MONOLINGUAL PERFORMANCE
SENTENCES(parser Lx)
Lx Ly
SENTENCES(parser Ly)
IF CROSS-LINGUISTICDIFFERENCES:
Mary saw a giftfor a boy …
María vio un regalo para un niño …
BILINGUAL PERFORMANCE
SENTENCES(parser Lx)
Lx Ly
SENTENCES(parser Ly)
TWO PARSERS?
STRATEGIES DEPEND ON THE LANGUAGE OF THE STIMULUS
María vio un regalo para un niño …
Mary saw a giftfor a boy …
BILINGUAL PERFORMANCE
Lx Ly
OR ONE?
SENTENCES(parser Lx)
SENTENCES(parser Ly)
UNIFORM STRATEGIES, WITH STIMULUS IN EITHER LANGUAGE;type of strategy depends on individual speaker variables
María vio un regalo para un niño …
Mary saw a giftfor a boy …
PARSING PRINCIPLES
MINIMAL ATTACHMENT(“Build the simplest structure”)
LATE CLOSURE / RECENCY PREFERENCE(“Attach locally”)
MINIMAL ATTACHMENT
Mary saw… Mary saw a gift for a boy…S
VP
V
saw
NP
a gift for a boy
NP
Mary
Mary saw a gift for a boy would be a good idea.
VP
would be a good idea
S
building complex structure = processing cost
LATE CLOSURE, in English
Mary saw a gift for a boy…
NP
PP
P
for
NP
a boy
NP
a gift
Mary saw a gift for a boy in a box.
PP
in a box
attaching non-locally = processing cost
LATE CLOSURE in English… y en español
María vio un regalo para un niño…
NP
PP
P
para
NP
un niño
NP
un regalo
María vio un regalo para un niño en una caja.
PP
en una caja
attaching non-locally = processing cost
LATE CLOSURE, RECENCY PREFERENCE
ATTACH LOCALLY
... a gift to a boy in a box
in many languages with many constructions
no interesting predictions for bilinguals:• bilinguals and monolinguals will all
prefer local attachments
EXCEPTION: N1 of N2 RC
… N1 of N2 RC
the relative clause (RC) attachment ambiguity structurally ambiguous: RC could attach to N1 or N2
An assassin shot the maid of the actress …
who was on the balcony.
Un asesino disparó a la criada de la actriz …
que estaba en el balcón.
EN:
SP:
N1 N2
N1 N2
QUESTIONNAIRE STUDIES
An assassin shot the maid of the actress who was on the balcony.
Who was on the balcony? the maid the actress
low attachment(N2) preference
high attachment(N1) preference
AMBIGUOUS TARGETS:
QUESTIONNAIRE STUDIES
Mary lent her favorite sweater to her best friend Susanne.
Who borrowed a sweater? Mary Susanne
An assassin shot the maid of the actress who was on the balcony.
Who was on the balcony? the maid the actress
AMBIGUOUS TARGETS:
UNAMBIGUOUS FILLERS:
SPANISH [high] ≠ ENGLISH [low]
LOW ATTACHMENT
ENGLISH, et a few al. Arabic Norwegian Romanian Swedish ??
Un asesino disparó a la criada de la actriz que estaba en el balcón.
An assassin shot the maid of the actress who was on the balcony. MONOLINGUALS…
HIGH ATTACHMENT
SPANISH, et al. Afrikaans, Dutch Brazilian Portuguese Bulgarian, Russian Croatian French German Greek ??
Un asesino disparó a la criada de la actriz que... HIGH in SP
An assassin shot the maid of the actress who… LOW in EN
BILINGUAL SENTENCE PROCESSING
BILINGUALS…
HIGH ATTACHMENT LOW ATTACHMENT
Babble babble in either language N1 P N2 RC… HIGH if SDOM
Babble babble in either language N1 P N2 RC… LOW if EDOM
LANGUAGE DEPENDENT PROCESSING:depending on the language of the stimulus?
LANGUAGE INDEPENDENT PROCESSING:same strategies, no matter the language;
type of strategy based on individual speaker variables?
HIGH ATTACHMENTin Spanish
LOW ATTACHMENTin English
HIGH ATTACHMENTif Spanish-dominant
LOW ATTACHMENTif English-dominant
CROSS-LINGUISTIC DIFFERENCES:
Ultimate preferences are the result ofinitial attachments Spanish parser English parser
Ultimate preferences are the result ofpost-syntactic processing Spanish parser = English parser departure from (early) low attachment due to semantics
(meaning), pragmatics (use), prosody (segmentation)…
WHY?
MONOLINGUAL PERFORMANCE
… la criada de la actriz que …
SENTENCES(parser Lx)
POST-SYNTAXPROCESSING(pragmatics,prosody? Lx)
… the maid of the actress that …
SENTENCES(parser Ly)
POST-SYNTAXPROCESSING(pragmatics,prosody? Ly)
(universal parser)
(universal parser)
the maid (N1) was on the balcony! the actress (N2) was on the balcony!
EXPERIMENTAL DESIGN
SUBJECTS monolingual & bilingual
MATERIALS English & Spanish
TASKS speeded “on-line” task (early processing) &
unspeeded “off-line” task (later processing)
SUBJECTS
Monolingual Speakers of…
American English
(USENG)
Castillian Spanish(CSPA)
N = 64 N = 64
Bilingual Speakers (from NYC)…
Dominant in English (EDOM)
Dominant in Spanish (SDOM)
N = 40 N = 40
Self-Rated Proficiencydifference Eng – Spa,
listening/speaking reading/writing
– 0.75– 1.15
+ 0.72+ 0.77
The journalist interviewed the coach of the gymnast…
… the coach of the gymnasts that was …
… the coaches of the gymnast that was …
… the coach of the gymnast that was …
MATERIALS
Ambiguous, questionnaire:
Disambiguated, self-paced reading:
Matrix with N of/de N in post-verbal position:
Ambiguous, questionnaire:
Disambiguated, self-paced reading:
Matrix with N of/de N in post-verbal position:El periodista entrevistó al entrenador del gimnasta…
… el entrenador de los gimnastas que estaba ...
… los entrenadores del gimnasta que estaba …
… el entrenador del gimnasta que estaba …
MATERIALS
SELF-PACED READING TASK
EARLY PROCESSING Read DISAMBIGUATED sentences
presented in 2 frames followed by comprehension questions INDIRECT measure of preferences which is faster, a forced low or a forced high
attachment?
… the coaches of the gymnast / that was …
… the coach of the gymnasts / that was …
The journalist interviewed the coach of the gymnaststhat was signing autographs during the competition.
forced high
forced low
Was the coach signing autographs during the competition?
QUESTIONNAIRE TASK
LATER PROCESSING Read AMBIGUOUS sentences
typed on one line followed by question about the attachment DIRECT measure of preferences which is chosen more frequently,
N2 or N1?
The journalist interviewed the coach of the gymnast that was sick.
Who was sick? the coach the gymnast
The dog bit the mailman and barked at the cat.
Who bit the mailman? the dog the cat
… the coach of the gymnasts
-150
-100
-50
0
50
100
150
USENG CSPA
ON-LINE READING TIMES: MONOLINGUALS
that was signing autographs during the competition.FRAME 2FRAME 1The journalist interviewed the coaches of the
gymnast
… the coaches of the gymnast
low attachmentpreference:low faster
high attachmentpreference: high faster
-150
-100
-50
0
50
100
150
USENG CSPA
ON-LINE READING TIMES: MONOLINGUALS
main effect of Site:F1 (1,72) = 7.77, p < .01F2 (1,20) = 6.15, p < .05
Language Site n/s
25
50
75
USENG CSPA
OFF-LINE PREFERENCES:MONOLINGUALS
… the coach of the gymnast that was signing autographs…
Who was signing autographs? the coach the gymnastthe coach the gymnast
low attachmentpreference
high attachmentpreference
25
50
75
USENG CSPA
OFF-LINE PREFERENCES:MONOLINGUALS
main effect of Language:F1 (1,44) = 5.48, p < .025
F2 (1,10) = 56.05, p < .001
ON-LINE READING TIMES:BILINGUALS
-150
-100
-50
0
50
100
150
ENG SPA
EDOM SDOM
-150
-100
-50
0
50
100
150
ENG SPA
EDOM SDOM
main effect of Site n/s:F1, F2 < 1
Site Language n/sSite Dominance n/s
Site Dominance Language n/s
MONOLINGUALS
-150
-100
-50
0
50
100
150
USENG CSPA
ON-LINE READING TIMES:MONOLINGUALS & BILINGUALS
2200
2400
2600
2800
3000
3200
English Materials
Spanish Materials
Rel
ativ
e C
laus
e, O
vera
ll R
Ts
(mse
c)
USENG & CSPA EDOM SDOM
SUBJECTS
Monolingual Speakers of…
American English
(USENG)
Castillian Spanish(CSPA)
N = 64 N = 64
Bilingual Speakers (from NYC)…
Dominant in English (EDOM)
Dominant in Spanish (SDOM)
N = 40 N = 40
Self-Rated Proficiencydifference Eng – Spa,
listening/speaking reading/writing
– 0.75– 1.15
+ 0.72+ 0.77
25
50
75
ENG SPA
EDOM SDOM
25
50
75
ENG SPA
EDOM SDOM
OFF-LINE PREFERENCES: BILINGUALS
MONOLINGUALS
25
50
75
USENG CSPA
Dominance Language n/s
main effect of Dominance:F1 (1,40) = 9.04, p < .005
F2 (1,20) = 59.36, p < .001
EARLY PROCESSING Low attachment in English and Spanish monolinguals
Bilinguals slower than monolinguals
No attachment preferences in English/Spanish bilinguals
LATER PROCESSING Differences in monolingual English (low) and Spanish
(high)
Language independent processing in bilinguals
Strategies associated with those of monolinguals in the bilinguals’ dominant language
BILINGUAL SENTENCE PROCESSING:1 + 1 = 1
Do bilinguals process input as if they were monolinguals of each of their languages?
NO
BILINGUAL SENTENCE PROCESSING
BILINGUALS…
HIGH ATTACHMENT LOW ATTACHMENT
Babble babble in either language N1 P N2 RC… HIGH if SDOM
Babble babble in either language N1 P N2 RC… LOW if EDOM
LANGUAGE DEPENDENT PROCESSING:depending on the language of the stimulus?X X X X
LANGUAGE INDEPENDENT PROCESSING:same strategies, no matter the language;
type of strategy based on language dominance
HIGH ATTACHMENTif Spanish-dominant
LOW ATTACHMENTif English-dominant
RECENT DEVELOPMENTS
CONVERGING EVIDENCE?
Brazilian Portuguese & English bilinguals
off-line questionnaire BP L1 or EN L1
BP L1 bilinguals: high in both languages EN L1 bilinguals: low in both languages
(Maia & Maia, 2001)
RECENT DEVELOPMENTS
CONVERGING EVIDENCE?
Spanish & English bilinguals
off-line questionnaire early acquirers of Lx & Ly;
late acquirers of EN L2 or SP L2
early acquirers: no preference late acquirers:
• EN L2: low in EN, high in SP• SP L2: low in EN & SP
(Dussias, 2001)
RECENT DEVELOPMENTS
CONVERGING EVIDENCE?
Spanish & English bilinguals
on-line self-paced reading, materials only in SP early acquirers of Lx & Ly;
late acquirers of EN L2 or SP L2
early acquirers: no preference late acquirers:
• EN L2: high in SP• SP L2: trend to high in SP
(Dussias, 2001)
RECENT DEVELOPMENTS
CONVERGING EVIDENCE?
speakers of Greek as L2
on-line self-paced reading, materials only in GK late acquirers of GK, L1 speakers of SP, GE, RU
all L2 learner groups: no preference
(Papadopoulou, 2002)
GENERAL CONCLUSIONS
Relative Clause Attachment Preferences Similarity between English and Spanish in
early processing Departure from low attachment preference in
later phases of processing
Bilingual sentence processing Evidence of language-independent strategy
use Strategies resemble those of monolingual
speakers of a bilingual’s dominant language
REMAINING PROBLEMS
insensitive “on-line” task did we miss the early low attachment preference in the bilinguals? or do bilinguals not engage in structurally-based processing strategies?
a mystery, what drives cross-linguistic differences grammar? (unlikely, given these results) person-based variable: lexical frequencies? tuning? prosody?
circumstantial idiosyncrasies of bilinguals corroborate with evidence from other bilingual populations
focus on language dominance: other variables? manner and age of acquisition frequency of language use literacy, primary language of education etc.
THANK YOU!
Please send questions and comments to:
Eva Ferná[email protected]
download a copy of this presentation at:
http://www.qc.edu/~efernand/papers/emf_25nov02.ppt