An online study of L2 relative clause processing: Evidence from self-paced reading in Persian learners of English
Author
Summary, in English
This study examines relative clause (RC) processing in Persian learners of English and native speakers of English to explore whether or how different task demands, referential context information with three potential RC antecedents along with the variables noun type (definite, indefinite), RC length (short, long) and RC type (extraposed, non-extraposed) affect their processing, using both an online non-cumulative self-paced reading task and an offline questionnaire. In the self-paced reading task, the online processing of RC attachment resolution was examined when participants read temporarily ambiguous sentences with RCs preceded by one clause or two clauses containing three NPs followed by comprehension questions to explore L2 RC attachment preferences. In the offline task, participants were asked to read ambiguous sentences and choose the antecedent which was most plausible to them for the RCs. Moreover, the Language History Questionnaire (LHQ) 3, the LexTALE test, a digit span task, and a reading span task were also administered to investigate whether measures from these tasks predicted the online processing outcome.
The results showed that Persian learners’ and native speakers’ online L2 RC processing was not significantly affected by the experimental conditions, but group effects were observable. There were significant differences between groups when reading sentences with extraposed RCs (region 4). The Persian learners read sentences with definite short non-extraposed RCs (region 2) faster than indefinite short non-extraposed RCs, and so did native speakers. Moreover, the Persian learners’ overall reading times were longer than native speakers’ reading times, whether with extraposed RCs or not. In contrast, in the offline task the Persian learners and the native speakers behaved differently in attachment preference. The Persian learners preferred high attachment and were sensitive to lexical-semantic dependencies in line with the predictions of the Competition model (CM) and Shallow Structure Account (SSA) which underline L1 influence. The native English speakers’ RC ambiguity resolution, on the other hand, was affected by the recency principle, relying more on structural dependency and word order. However, the native speakers’ tendency to high attachment was considerable. The results also revealed that for both groups, online L2 RC processing and observed attachment preferences are not robustly directly modulated by working memory, however, by a simple heuristic processing activated by semantic memory-based information in which earlier good-enough linguistic representations are mostly preferred to achieve a state of Cognitive Equilibrium (resolved or minimized ambiguity), such that language users are more likely to prefer high attachment than to prefer low attachment in RC attachment processing.
The results showed that Persian learners’ and native speakers’ online L2 RC processing was not significantly affected by the experimental conditions, but group effects were observable. There were significant differences between groups when reading sentences with extraposed RCs (region 4). The Persian learners read sentences with definite short non-extraposed RCs (region 2) faster than indefinite short non-extraposed RCs, and so did native speakers. Moreover, the Persian learners’ overall reading times were longer than native speakers’ reading times, whether with extraposed RCs or not. In contrast, in the offline task the Persian learners and the native speakers behaved differently in attachment preference. The Persian learners preferred high attachment and were sensitive to lexical-semantic dependencies in line with the predictions of the Competition model (CM) and Shallow Structure Account (SSA) which underline L1 influence. The native English speakers’ RC ambiguity resolution, on the other hand, was affected by the recency principle, relying more on structural dependency and word order. However, the native speakers’ tendency to high attachment was considerable. The results also revealed that for both groups, online L2 RC processing and observed attachment preferences are not robustly directly modulated by working memory, however, by a simple heuristic processing activated by semantic memory-based information in which earlier good-enough linguistic representations are mostly preferred to achieve a state of Cognitive Equilibrium (resolved or minimized ambiguity), such that language users are more likely to prefer high attachment than to prefer low attachment in RC attachment processing.
Department/s
- General Linguistics
- Master's Programme: Language and Linguistics
Publishing year
2022
Language
English
Full text
Document type
Student publication for Master's degree (two years)
Topic
- Languages and Literatures
Keywords
- relative clause
- online L2 processing
- second language comprehension
- attachment preference
- self-paced reading
- heuristic processing
- good-enough approach
- working memory span.
Supervisor
- Marianne Gullberg (Professor)