PhD in Second Language Studies Dissertation Defense
Cross-linguistic activation in Korean L2 learners’ processing of remention bias in English
Chair: Theres Grüter
Friday November 9, 11:00 a.m.–2:00 p.m.
Moore Hall, Room 258
Effects of cross-linguistic activation in L2 learners have been demonstrated abundantly at the word level (e.g., Dijkstra, 2005; Prior, Degani, Awawdy, Yassin & Korem, 2017; Van Assche, Duyck & Brysbaert, 2013), yet less is known about the consequences of cross-language activation at the word level on processing at the sentence and discourse level. This dissertation investigates if and how Korean learners of English are affected by the strength of referential biases associated with certain interpersonal predicates in Korean in their reference choices and processing in English. The study also tests potentially modulating roles of translation priming, L2 proficiency, and L2 learning experience, whose effects on cross-language activation remain an issue of ongoing investigation.
Remention bias is a well-known phenomenon whereby certain verbs appear to create biases to remention either its subject or its object in a causal dependent clause (Garvey & Caramazza, 1974; Hartshorne, 2014). Importantly, some English remention bias verbs have no lexical translation equivalents in Korean and can only be translated with a syntactic causative (SC) construction involving explicit causative marking (e.g., ‘surprise’ànolla-key-ha, be surprised-RESULT-do). Experiment 1 tested whether Korean SC predicates lead to stronger remention biases than non-SC predicates. Results from written sentence-completion tasks in Korean and English, respectively, showed stronger subject bias with SC than non-SC predicates among native Korean speakers, yet similar biases for the English translation equivalents of these predicates among native English speakers. Experiment 2 further explored whether the stronger bias with SC versus non-SC predicates in Korean affects Korean-speaking L2 learners’ sentence completions in English. The study also probed for potential effects of translation priming by having L2 learners complete a translation task either preceding or following the sentence-completion task. Results indicated that the strength of a verb’s referential bias in Korean affected learners’ reference choices in English. This effect emerged independent of the presence or absence of translation priming. Experiment 3 tested whether the results from Experiment 2 could be replicated with a different set of remention bias verbs with more uniform argument structures. The results of Experiment 3 not only replicated the effects of cross-linguistic activation in L2 referential choices but also showed that these effects emerged regardless of learners’ proficiency or learning experience (immersed vs. instructed). Based on the effect of cross-linguistic activation in L2 learners’ (offline) referential choices observed in Experiments 2 and 3, Experiment 4 used the visual world eye-tracking paradigm to investigate whether the effect extends to online processing. Results showed that while L2 learners used remention bias information during real-time listening, their use of the information was delayed compared to native speakers. Yet no robust evidence was found that either proficiency or cross-linguistic activation interacted with L2 learners’ use of remention bias.
Overall, the results from this study indicate that the effect of cross-linguistic activation goes beyond the word or construction level and influences Korean-speaking L2 learners’ referential choices at a discourse level. These effects were robust and replicable in two offline tasks, and emerged irrespective of the presence of translation priming, L2 proficiency, and L2 learning experience. These effects are assumed to arise through the mental models created under the inﬂuence of cross-linguistic activation at the word and construction level during L2 learners’ production of written discourse continuations. In the visual world eye-tracking task, by contrast, no clear effects of cross-linguistic activation emerged, potentially due to L2 listeners’ delayed use of remention bias in real-time processing.