Syntactic Processing in Second Language Production Susanna Flett

Syntactic Processing in Second Language Production Susanna Flett Holly Branigan, Martin Pickering, & Antonella Sorace School of Philosophy, Psychology and Language Sciences University of Edinburgh

L 2 sentence level production: • Research focuses on L 2 lexical level • What about phrasal/sentence level? • What ? – Which structures available – How to form them • When ? – Frequency of L 2 structure – Semantic or pragmatic constraints on usage

Models of L 1 production (Roelofs, 1992, 1993; Pickering & Branigan, 1998) • Syntactic knowledge in lexicon • Combinatorial information: • Phrasal in nature • Linked to specific lexical items • Shared between different lexical items • Same architecture in L 2? – (de Bot, 1992; Truscott & Sharwood Smith, 2004) • Based on on-line behavioural evidence syntactic priming • Tendency to re-used previously processed structure

Typical priming study • Hear or read a sentence (prime) • Syntactic structure varies: – Active: “One of the fans punched the referee” – Passive: “The referee was punched by one of the fans” • Then describe unrelated picture (target) Priming Effect: Passive target more likely after passive prime

Syntactic priming • Found with a variety of structures • (e. g. , active/passive; dative PO/DO; word order; ‘that’ complementiser. . . ) • Not due to lexical, semantic or rhythmic overlap • (Bock, 1989; Bock & Loebell, 1990) • Indicates abstract syntactic representations • Stronger effect if lexical overlap • (e. g. , Pickering & Branigan, 1998; Cleland & Pickering, 2003) • Tool to study sentence level production • Well established in L 1 speakers

L 2 syntactic processing • Do L 2 speakers acquire: – Abstract syntactic representations? – Syntactic processing similar to L 1? – L 2 structural preferences? • • L 2 Priming? Stronger than in L 1? Change with proficiency? Change with experimental context?

Experiments • L 1 and L 2 Spanish (English L 1) • • 1) Actives/Passives – dialogue 2) Actives/Passives – computerised 3) SV/VS – unergative verbs 4) SV/VS – unaccusative verbs

Experiments 1 & 2 • English and Spanish have actives and passives – – John built the house The house was built by John Juan construyó la casa La casa fue construida por Juan • Spanish passive grammatical but uncommon • Active/Passive • Same/Different verb • (Branigan et al. , 2000; Pickering & Branigan, 1998)

Experiment 1 • Picture description game • Dialogue with native Spanish confederate • (e. g. , Branigan, Pickering & Cleland, 2000) • Sit opposite each other • Confederate follows script • Spoken primes

Experiment 1

Different verb in prime and target: Participant hears: “El autobús persigue el tren” OR “El tren es perseguido por el autobús”

Different verb in prime and target: Then see: Decide if previous sentence matches picture or not

Different verb in prime and target: Then see: Describe picture out loud…

Same verb in prime and target: Participant hears: “La guitarra destruye el televisor” OR “El televisor es destruido por la guitarra”

Same verb in prime and target: Then see: Decide if previous sentence matches picture or not

Same verb in prime and target: Then see: Describe picture out loud…

Experiment 1 • Participants: – L 1 Spanish (n=12) – Intermediate (n=12) and advanced (n=12) L 2 Spanish (L 1 English) • Scoring: Actives, Passive or Other produced in each condition

Experiment 1: Percentage of passive targets % Passives Group Prime Type

Experiment 2 • Social influence – pressure to conform to native speaker interlocuter? • Experiment 2: – Monologue – Primes and pictures on computer – Visual primes • Identical pattern of results, attenuated L 2 priming

Implications • • L 2 abstract representation Linked to specific verbs lexical boost Shared in comprehension and production L 2 more susceptible to priming than L 1 • Passives exist in English and Spanish • Shared across languages • (e. g. , Hartsuiker, Pickering and Veltkamp, 2004) • Prime a new structure?

Experiments 3 & 4: word order • Spanish allows SV and VS order • Juan llegó ‘Juan arrived’ • Llegó Juan ‘*Arrived Juan’ • Preference determined by: – lexical verb class – discourse structure • Difficult for L 2 learners to use appropriately

Intransitives • Two intransitive verb groups – (e. g. , Perlmutter, 1978) – Unaccusatives: Arrive; enter; leave; fall • (argument is theme or patient, base-generated in object position) – Unergatives: Shout; dance; speak; laugh • (argument is an agent, generated in subject position) • Semantic differences, syntactically represented – (e. g. , Levin and Rappaport Hovav, 1995)

Lexical preferences in neutral context: • Unergatives take SV order • Mi hermana gritó (‘my sister shouted’) • Unaccusatives prefer VS order • Llegó mi hermana (‘arrived my sister’) • Early L 2 (English L 1) use only SV • Increasing sensitivity to preferences • Then over-generalise VS (Hertel, 2003; Lozano, 2004)

Experiments 3 & 4: • Word order is primable – (e. g. , Hartsuiker & Westenberg, 2000; Hartsuiker, Kolk & Huiskamp, 1999) • How will priming and lexical preferences interact in L 1 and L 2 speakers?

Method: • Primes presented on computer – SV or VS order – Same or different verb • Expt 3: Unergatives (shout-type) • Expt 4: Unaccusatives (arrive-type) • L 1 Spanish (n=20) • L 2 Spanish (L 1 English; n=24)

Different verb in prime and target: Participant reads: “El pingüino baila” OR “Baila el pingüino”

Different verb in prime and target: Then see: Decide if previous sentence matches picture or not

Different verb in prime and target: Then see: Describe picture out loud…

Experiment 3 (unergatives): proportion of VS % VS order Group Prime Type

Experiment 4 (unaccusatives): proportion of VS % VS order Group Prime Type

% VS order Comparing Expts: 3 and 4

Summary of findings • • • Word order priming in L 1 and L 2 Spanish Stronger when verbal repetition Lexical preferences affect L 1 priming L 2 priming same for both verb groups L 2 speakers more willing to use less frequent structures (passives, VS) • L 2 > L 1 priming only when structure dispreferred in L 1

Structural preferences in L 1 and L 2 • Priming stronger for lower-frequency structures • (Bock, 1986; Hartsuiker & Kolk, 1998; Hartsuiker & Westermann, 2000; Scheepers, 2003) • But dispreference of structures? – L 1 dispreference resist priming • (e. g. , Pickering, Branigan & Mc. Clean, 2002, no heavy NP shift priming in English ; Hartsuiker & Kolk, 1998, passives in Dutch) – L 2 speakers: find passive and unerg-VS more acceptable

Conclusions • Syntactic representation and processing similar in L 1 and L 2 (á la de Bot, 1992) – Syntactic priming – Lexical overlap boost • L 2 speakers not sensitive to preferences – can prime dispreferred structure • L 1 speakers floor effect can’t be overcome

Future Research? • How would effects vary for: – Proficiency: beginner and near-native L 2? – L 1 and L 2 speakers based in Spanish environment? – Structures equally acceptable in both languages?

Thank you for listening • susanna. flett@ed. ac. uk