Building an L2 lexicon, the acquisition of verb classes relevant to causativization in English by speakers of Hindi-Urdu and Vietnamese

grantor: University of TorontoThis study examines the L2 acquisition of constraints on English causativization in order to gain insight into the organization of the L2 lexicon, the route of L2 lexical acquisition, and the possible interaction between the L1 and L2 lexicons. Lexical organization is investigated via an analysis of the syntactic production and judgments of learners on the premise that verbs displaying similar grammatical behaviour generally do so because they share meaning components. It was hypothesized that (i) the ability to generalize beyond the input would improve with lexical proficiency, and (ii) the learner's L1 would influence the acquisition of verbs in those L2 semantic classes where there were translation equivalents of existent L2 items in the L1. In order to investigate lexical transfer, the performance of Hindi-Urdu speakers at three levels of lexical proficiency was compared with that of Vietnamese speakers at corresponding proficiency levels since there are marked differences between these languages in terms of causativization. The hypotheses were tested through four tests in English: a picture-based production task, a picture-based multiple-choice task, a grammaticality judgment test involving high-frequency verbs, and a grammaticality judgment test involving low-frequency, previously unknown verbs. Results indicated that on the second judgment task, the advanced learners in each language group had significantly higher overall scores than the elementary learners, suggesting that advanced learners make better connections than elementary learners between newly encountered L2 verbs and previously known verbs that are semantically related to them. Significant increases between the elementary and advanced levels were also witnessed on the picture-based tests; since these two tests centred on known verbs, these increases were seen as indications of growing accuracy across semantic classes with increasing proficiency. The validation of many of the transfer-related sub-hypotheses suggests that there is at least some sharing of semantic information between the L1 and L2 verb lexicons. Specifically, the findings highlight the existence of varying degrees of "transferability" across verb sub-classes and suggest that transfer plays a role in the ease or difficulty involved in shedding overgeneralized features of the L2.Ph.D