Developing a unified feature-based model for L2 lexical and syntactic processing

Abstract

Research on lexical processing shows that lexical representations of L2 speakers are less developed, so frequency and vocabulary size affect the way they use lexical information. Specifically, reduced access to lexical features hinders the processing system of L2 speakers from working efficiently, having an impact on their ability to build syntactic structures in a native-like manner. The present research project aims to construct and test a unified model that explains how lexical and sentence processing interact. First, it develops and validates a productive vocabulary task for L2 Italian to measure vocabulary size. The task, called I-Lex, is based on the existing LEX30 for English, and uses frequency to determine lexical knowledge. Then, adopting the formalism of Head-Driven Phrase Structure Grammar, a framework that associates all the information relevant to the grammar with the lexicon, the research project develops a model that explains the effects of lexical access on syntactic processing. The model is tested in two empirical studies on L2 speakers of Italian. The first study, using an Oral Elicited Imitation task, and the I-Lex productive vocabulary task investigates the effects of frequency and vocabulary size on cleft sentences. The second study, using the same productive vocabulary task and a Self-paced Reading task, investigates frequency and vocabulary effects on relative clauses. The results reveal that frequency and vocabulary size interact with the ability of L2 speakers to process both cleft and relative clauses, providing evidence that accessing lexical features is a crucial stage for processing syntactic structures. Based on the results, a feature-based lexical network model is constructed. The model describes how lexical access and the activation of structural links between words can be described using the same set of lexical features. In the last chapter, the model is applied to the results of the two studies.