In real-time sentence processing, the human language processor draws on multiple sources of information and applies them in a rapid fashion. The time pressure of language processing requires highly efficient use of this information. Ideally, second language (L2) learners should also be able to employ all the sources of information as efficiently as first language (L1) speakers, but in reality, it is often difficult. Capitalizing on typological difference between Chinese, English, and Japanese, this dissertation research examines the extent to which L2 learners are able to use different sources of information in online sentence processing and what underlying mechanisms (e.g., L1 transfer, inadequate representation, or cognitive limitations) cause processing difficulties.
This research focuses on lexical-semantic, contextual, and morphosyntactic sources of information, which are the major sources of information that the human language parser exploits in making parsing decisions. The project includes four main experiments with intermediate and advanced learners of Chinese, whose L1 is either English or Japanese, as well as native controls. Participants in each experiment complete one of two online tasks -- comprehension-focused self-paced reading or acceptability-judgment self-paced reading -- in addition to an associated offline task.
This project is a pioneering study in (L2) Chinese sentence processing. Most of the current theories on L1 and L2 sentence processing have been formulated in the context of languages with inflectional morphology. As a language with virtually no inflectional morphology, Chinese plays a special role in resolving theoretical controversies about the interaction of semantics, context, and morphosyntax during sentence processing. Studying how L1 and L2 speakers process Chinese sentences will help researchers obtain a much richer picture of human sentence processing in general, and L2 sentence processing in particular.
Our research project investigates how second language (L2) learners process L2 sentences in real time, with a focus on how L2 learners of Chinese process Chinese sentences. It seeks to reveal how L2 learners exploit different sources of information and what processing difficulties they may or may not have. The specific research issues include (a) the extent to which L2 learners are able to make use of different sources of information (semantic, contextual, and morphosyntactic) in online sentence processing and (b) the underlying mechanisms (e.g., first language [L1] transfer, inadequate representation, or cognitive limitations) that cause L2 processing difficulties. In researching these issues, the project also explores how to measure "deep-level" (i.e., morphosyntactically detailed or fully specified) L2 processing and whether L2 learners can ultimately develop native-like processing abilities. The research includes five main experiments: one in English comparing native English speakers with Chinese learners of English, and four in Chinese comparing native Chinese speakers with both English and Japanese learners of Chinese. Each experiment contains two processing tasks: (a) acceptability-judgment self-paced reading (AJSPR) and (b) comprehension-focused self-paced reading (CFSPR). AJSPR requires participants to judge the acceptability of the sentence after reading it; this task therefore engages participants, both native speakers and L2 learners, in immediate and detailed grammatical processing. CFSPR, by contrast, requires participants to answer a comprehension question after reading the sentence; this task therefore encourages (especially) L2 learners to practice "shallower" or less detailed grammatical processing and thus may attract L2 learners away from exercising detailed or fully specified processing. Comparisons of L2 learners’ results from these two online tasks can help us determine the different processing strategies they use under different task conditions. Experiment 1 assesses whether AJSPR can elicit some deep-level processing that CFSPR cannot reveal. This is done by examining L1 and L2 processing of English subject-verb number agreement in an AJSPR task in comparison to a CFSPR task. Experiment 2 tests whether L2 learners are good at processing lexical-semantic information in Chinese. It is done by comparing L1 and L2 participants’ sensitivity to local (im)plausibilities caused by temporary objects that satisfy or violate verbs’ selectional restrictions. For example, the verb liaojie "understand" can take both human and nonhuman objects while titie "considerate" can only take human objects, and so titie gongchang "be considerate to the factory" in titie gongchang-de gongren "be considerate to the factory workers" is locally implausible. Experiment 3 examines whether L2 learners can use contextual information in processing Chinese sentences in which discourse contexts are either supportive or unsupportive of object omissions. Experiment 4 examines whether L1 transfer can cause difficulty in L2 processing of morphosyntactic information. The specific morphosyntactic element is the Chinese plural marker -men, an element similar to the plural morpheme -s in English in terms of its grammatical function of marking NP plurality but different in terms of its inability to co-occur with a quantifier expression (e.g., *san-ge xuesheng-men ‘three-classifier student-plural’). Experiment 5 compares L1 and L2 processing of negation-aspect interactions in Chinese to test whether adult L2 learners whose L1 lacks such morphosyntactic knowledge can ultimately acquire it and are able to use it in online sentence processing. Several important findings have resulted from our research. First, L2 learners are able to make use of different sources of information (semantic, contextual, and morphosyntactic) in online sentence processing as their proficiency rises. This gives us the hope that L2 learners can ultimately develop native-like processing abilities. Second, L2 learners are not across-the-board good at processing semantic or contextual information, and they are able to engage in fully-specified morphosyntactic processing. Such results challenge deficit theories of L2 acquisition and processing. Third, L2 sentence processing is subject to the influence of task demands and L2 proficiency, and L2 processing difficulties are not necessarily indicative of deficit representations but rather attributable to L2 learners’ L1 interference and cognitive limitations. In addition to these findings, our research establishes the structure-focused AJSPR as an appropriate tool for measuring deep-level L2 processing, and this provides an effective way of assessing L2 deep-level processing. From a broad and practical perspective, our research has led to a better understanding of (native and nonnative) sentence processing in Chinese. We have learned that both L1 speakers and L2 learners make use of morphosyntactic information in sentence processing in Chinese, a language well known for its scarcity of inflectional morphology. This informs us that in certain grammatical domains, morphosyntax plays an indispensable role in Chinese sentence processing.
