5. A processing problem or a representational problem? L2 acquisition of syntax-semantics interface in the Chinese ba construction by English-speaking learners
by Liu, Tongkun
Abstract: Second language (L2) learners’ grammars seem always more or less divergent from the natives’ grammar of the same language, even when L2 learners have reached a very advanced proficiency level. Interface Hypothesis (Sorace & Filiaci, 2006; Sorace, 2011) addresses the reason for this phenomenon by arguing that the divergent part of L2 grammars from the native grammar is interfaces where syntax and other cognitive domains (e.g. semantics) interact, and it is interface that can be permanently vulnerable in L2 grammars.
The Mandarin Chinese ba construction is an ideal test ground for this hypothesis. In the ba construction [FinP NP1 [Fin’ Fin [vP <NP1> [v’ BA [VP NP2 [V’ V XP]]]]]], ba is a phonetically realised little v (Huang, 2007). For the verb, which is syntactically C-commanded by ba, it always has [affected, resultative] semantic properties at the same time (Huang et al, 2009; Sun, 2015), otherwise the ba construction will be unacceptable. e.g.:
A. Zhangsan ba Lisi da-si-le. (Zhangsan BA Lisi beat-kill-LE) [+affected, +resultative]
B. ?Zhangsan ba Lisi da-rao-le. (Zhangsan BA Lisi beat-disturb-LE) [+affected, -resultative]
C. ?Zhangsan ba Lisi kan-jian-le. (Zhangsan BA Lisi watch-see-LE) [-affected, +resultative]
D. ??Zhangsan ba Lisi mo-fang-le. (Zhangsan BA Lisi model-imitate-LE) [-affected, -resultative]
However, if there is no ba in the numeration, then the verb will not be C-commanded by ba but raises to v in the above syntax structure, the structure will be realised as a corresponding canonical subject-verb-object Chinese sentence (e.g. Zhangsan mo-fang-le Lisi), in which there is no [affected, resultative] restrictions on the verb at all. This means that the semantic restrictions on the verb only exist when v is filled by ba and the verb is directly C-commanded by ba, hence a syntax-semantics interface at the verb in the ba construction.
This article reports an experimental study on whether English-speaking learners have such syntax-semantics knowledge in their L2 Chinese grammars, and if they are sensitive to the semantic restrictions on the verb in the ba construction in on-line sentence processing. An acceptability judgement task (AJT) and a self-paced reading (SPR) task were adopted with stimuli of ba sentences such as the above and their corresponding canonical sentences (e.g. Zhangsan mo-fang-le Lisi) as controls. In the SPR task, following up clause were added to the ba construction to capture possible spill-over effects. Comprehension questions were also used after each stimulus to make sure participants paid attention to the stimuli. Twenty-four Chinese native speakers and 83 English-speaking learners from intermediate to very advanced levels participated the experiments. The experiments find that only very advanced learners behaved native-like in the AJT, while advanced and high-intermediate groups showed great optionality in their judgement; intermediate group only had the syntactic knowledge of the ba construction but no semantic knowledge on the semantic restrictions. Although very advanced learners behaved native-like in the AJT, they, like other groups, failed to be congruent with the natives’ processing patterns of the ba construction in the on-line processing SPR task. The results suggest that internal interface properties are hard to acquire by L2 learners but not impossible, and the vulnerability at interfaces seem to be a processing problem rather than a representational problem in L2 grammars.
6. Covert Subjects and Objects Resulting from TP Ellipsis in English Speakers’ L2 Chinese: Evidence of the Declarative/Procedural Model
by Xu, Lilong
Abstract: Subjects are obligatorily overt in English, whereas not only covert subjects but also covert objects are allowed in Chinese. It has recently been argued that the gaps in the subject and object position in an affirmative reply to a Chinese yes-no question should be viewed as the result of movement and TP ellipsis under a verbal identity condition (Simpson, 2014). To affirmatively answer a Chinese yes-no question when the verb in the yes-no question and answer are identical, both the subject and object in the answer can be covert (see example 1 B1); Its counterpart in English is ungrammatical (see example 2). However, when the verb in the answer is not identical to the verb in the question but is synonymous with it, neither the subject nor the object can be covert (see example 1 B2) (see “the verbal identity condition”, Holmberg, 2015). This study investigates whether English native speakers who learn Chinese as their second language (L2) show developmental progress in their L2 Chinese covert subjects and objects.
A cross-modal self-paced reading task (SRT) and an acceptability judgement task (AJT) were used. The results of AJT showed that all participant groups rated covert arguments under the verbal identity condition (like example 1 B1) as being significantly more natural than those under the non-verbal identity condition (like example 1 B2). The results of SPT showed that beginners were not sensitive to errors related to the violation of the verbal identity condition either at the critical regions or at the following regions. However, intermediate learners, advanced learners and Chinese native speakers were sensitive to them and the sensitivity to the violation spilled over to the following two regions. The differences found in beginners ' performance in the on-line and off-line tasks suggest that derivations, such as movement and ellipsis, are not accessible in L2 online processing until intermediate or advanced levels. This finding can be accounted for with the declarative/procedural model (Ullman, 2006).