How classifiers facilitate processing in L2 Chinese
Refereed conference paper presented and published in conference proceedings



摘要L2 processing has been argued to rely more on lexico-semantic and less on grammatical information than L1 processing (FelserEtAl2003, VanPatten2004). Chinese classifiers provide a unique opportunity to test this claim. Shape classifiers (e.g., tiáo, ‘long/narrow’) generally co-occur with nouns denoting objects with the relevant shape properties (e.g., shéngzi, ‘rope’). Yet most classifier classes also contain nouns not obviously matching these features (e.g., gǒu, ‘dog’; in ‘tiáo’ class); for these, classifier-noun co-occurrence constraints are largely agreement-like (similar to Det-N gender-agreement). Using the visual-world paradigm, Tsang and Chambers (2011; ‘T&C’) investigated to what extent native Cantonese listeners derive predictions from semantic vs. grammatical/form-class constraints on
classifier-noun co-occurrence. They found classifiers facilitated Cantonese speakers’ processing “primarily through their grammatical constraints”, with little distraction from competitors matching the classifier only in semantics but not form-class (e.g., shǒubiǎo ‘wristwatch’, which is long/narrow, but cannot co-occur with tiáo). If semantics plays a
stronger role in L2 processing, we predict larger competition effects from such competitors among L2 listeners.

We test this prediction in a Mandarin adaptation of T&C’s experiment. Participants listened to questions (1) while viewing scenes containing a (non-prototypical) target (e.g., ‘dog’), competitor and distractor (Fig1). Competitors differed by condition, representing either a member of the target classifier class (‘rope’; G+S+), a member of a different class with some of the target class’ semantic properties (‘wristwatch’; G-S+), or an entirely unrelated noun (píngguǒ ‘apple’; G-S-). In the L1 group (n=24), eye-gaze patterns (Fig2) in the G-S+ and G-S- conditions align, indicating no influence of classifier-semantics from class-inconsistent competitors. In the L2 group (19 L1-English advanced learners; data collection on-going), the G-S+ and G+S+ conditions align, suggesting L2ers are influenced by classifier-semantics regardless of grammatical noun-class constraints. These patterns (supported by mixed-effect regression models) provide support for greater reliance on lexicosemantics in L2 vs L1 predictive processing.
著者Grüter Theres, Elaine Lau, Wenyi Ling
會議名稱2nd International Symposium on Bilingual and L2 Processing in Adults and Children (ISBPAC)

上次更新時間 2018-29-11 於 14:31