Lexical representations in bilingualism: syntax and phonology Rob Hartsuiker
Language integration To what extent are the two languages of a bilingual represented in separate or shared stores? Usually addressed for words
Kroll & Stewart (1994) L1 L2 Concepts
Separate lexicons? Increasing evidence for integrated lexicons: Comprehension: Spivey & Marian (1999), Van Heuven et al. (1998) Production: Colomé (2001), Costa et al. (1999), Hermans et al. (1998)
What about other aspects of language? Syntax: are syntactic representations shared between languages or kept separate? Phonology: are L2 words represented in terms of L2-phonology, or does L1-phonology play a role?
Part 1: Is syntax separate or shared between languages? The case of actives and passives in English and Spanish: The taxi chases the truck El taxi persigue el camión The truck is chased by the taxi El camión es persiguido por el taxi
A priori arguments Separate syntax Shared syntax Representational: Constructions only superficially identical Computational: Reduction of number of options taken into account Shared syntax Parsimonious storage: only one rule, instead of several.
Syntactic Priming Tendency to repeat the syntactic structure of previously processed sentences. E.g., Bock (1986). Repeat sentences and describe pictures…
Syntactic Priming Subject repeats a sentence: 5 people carried the boat or The boat was carried by 5 people Subject describes a picture: The alarm clock wakes up the boy The boy is awoken by the alarm clock
Syntactic priming (cont.) More passives following passive primes (Bock, 1986) Occurs in several languages In writing and in speaking For many types of constructions Cannot be attributed to non-syntactic factors Seems stronger in dialogue games
Syntactic priming in bilinguals If syntax is shared between languages: priming when prime is in one language, but production in another. If syntax is separate: no effect.
Experiment Subjects: 24 native speakers of Spanish, living in an English-dominant culture (22 mnths average) Materials: 32 pictures, best described with an active or passive, and 32 sets of prime sentences Syntactic primes: always in Spanish Responses: always in English
Conditions Active: El taxi persigue el camión Passive: El camión es perseguido por el taxi Intransitive: El taxi accelera OVS: El camión lo persigue un taxi
Experimental set-up El taxi persigue el camión
Scoring Active: The A verbs/is verbing the T Passive: The T is (being) verbed by the A Others: Everything else (e.g. ‘the truck is chased’; ‘there is a car chase, etc.) Proportion of passives out of actives and passives is dependent var.
Results Main effect of prime type Simple contrast: only passive condition differs from intransitive (baseline) Simple contrast: only passive condition differs from active condition.
Discussion Cross-linguistic priming: But asymmetric… Languages only moderately related But asymmetric… However, clearest symmetric effects with DATIVES (gave the man a book / gave a book the man) Intransitives are in the active voice. Prime for actives??
Proposal for monolingual lexical-syntactic representation Concepts Lemmas Active HIT TOW Passive V
Proposal for BIlingual lexical-syntactic representation Concepts Lemmas Active HIT TOW Passive GOLPEAR REMOLCAR V
Implications Code-switching: “I saw a red perro” facilitated because concepts and syntax are contact points Syntactic borrowing: “zij stemden de resolutie” (voter le resolution) Possible, because syntax from both languages activated
Part 2: A rows is a rose.. But is a road a load? Languages differ in their phonological systems. Japanese: /l/ vs /r/ absent Spanish: /b/ vs /v/ and /I/ vs/i/ To what extent are L2-words influenced by L1-phonology (even if you have mastered L2-phonology?)
Some informal observations… Spelling errors are frequently homophones of the intended word Too vs two Their vs there Bilingual spelling errors are sometimes <<‘omofones>>
Engrish
Engrish (2)
More ‘omophone errors Englitalian: ‘Buckingam’ Englitalian: ‘teach that staff’ Englitalian: ‘If you need anything for Carlo or so let me know, we leave so close…’ Dutchish: ‘Does, we conclude’ Spanglish: ‘I feel more estable’ EngRish: ‘As I hate getting tunned’ Scottish: ‘I got to literally run home to back for weekend in Inverness’
What about reading? Do ‘omophones lead to problems in L2-reading? If so, evidence that phonological representations for L2-words are influenced by L1-phonology.
Semantic categorization (Van Orden, 1987) A FLOWER More false positives for homophones (ROWS) of correct exemplar (ROSE) than for spelling controls (ROBS) Slower correct rejections for homophones than for spelling controls ROWS 1500 ########### 2000 N Y
Framework ROSE R O S E /r/ /o:/ /z/ /ro:z/ Semantic Representation Sublexical Orthographic Units Sublexical Phonological Units Phonological Lexicon Orthographic Lexicon Semantic Representation
Japanese phonological lexicon? Semantics Orth. lexion Ph. lexicon /*l-r* o: d/ ROAD LOAD
Participants Monolingual speakers of English Highly proficient* Japanese/English bilinguals Highly proficient* Spanish/Englilsh bilinguals * As determined by phonological discrimination pretests
Conditions Experimental Control H SEA/SEE SET A/U BAG/BUG BEG B/V BAN/VAN CAN L/R LOAD/ROAD TOAD
Procedure Semantic Categorization Every Category twice A street Load A street Toad Correct response for (h)omophones and controls always NO.
RT effect sizes (Z-transformed)
RTs: (h)omophony vs controls English: main effect of (h)omophony interaction with phon. contrast Japanese: main effect of (h)omophony Spanish: main effect of (h)omophony
English controls (N = 20) RT (ms) Exp Control Effect sig H1/H2 686 644 42 * A/U 668 660 8 ns B/V 640 660 -20 ns L/R 673 633 40 * ERRORS (%) H1/H2 11.9 2.5 * A/U 5 2.5 ns B/V 1.25 3.75 ns L/R 1.88 1.25 ns
Japanese/English bilinguals (N = 17) RT (ms) Exp Control Effect sig H1/H2 1213 1164 49 * A/U 1366 1121 245 * B/V 1289 1140 149 * L/R 1264 1091 173 * ERRORS (%) H1/H2 29.4 5 * A/U 33.8 10.6 * B/V 27.5 8.8 * L/R 20.6 13.1 t1: p =.062; t2: ns
Spanish/English bilinguals (N = 17) RT (ms) Exp Control Effect sig H1/H2 1166 1063 103 t1: p = .067 A/U 1195 1127 68 ns B/V 1135 1103 32 ns L/R 1082 1020 62 ns ERRORS (%) H1/H2 14.4 3.4 * A/U 10.6 6.3 ns B/V 7.5 7.5 ns L/R 4.4 7.5 ns
Summary English: Interference from homophones and /l/ - /r/ ‘omophones (not in errors) Japanese: (h)omophone interference Spanish: (h)omophone interference, but not for individual contrasts, not on errors.
Why /l/-/r/ interference? Cf. Lukatela et al. (2001) (masked priming in lexical decision): ZEA - SEA > VEA - SEA LOAD - ROAD > TOAD - ROAD ?
‘Omophony and bilingualism Words tend to activate ‘omophones more than similarly spelled words. But representations for ‘omophones do not overlap (completely), at least in the Spanish group.