Second Language Word Association: Processes, Methodologies and Models

This thesis examines processing in the bilingual mental lexicon by way of word association (WA) studies. The research reported here was designed to address three sets of key issues. Firstly, these studies attempt to establish whether or not WA methodology is a viable means of exploring the bilingual mental lexicon. Research questions to be addressed here concern the validity of current categorization schemes (i.e., whether they comprehensively account for all WA response data) and whether post-task interviews are necessary or useful in disambiguating responses. The second set of issues to be addressed here arise from the focus on cognitive processes in the lexicon. Research questions here are designed to address gaps in the literature concerning whether researchers are justified in conceptualizing subjects' responses as evidence of underlying cognitive styles. These questions will be addressed by implementing underused and never before utilized methods. Methods employed include restricted association tasks and an unconventional priming manipulation intended to alter response types (as opposed to altering speed of response, the conventional measure for inferring that priming has occurred). The final set of issues to be addressed here concerns the identification of the determinants of WA behaviour. Based on the results of the current studies, I will present a "dynamic" model of the WA process. The model attempts to account for the interplay among respondents' cognitive styles, features of the presented cues, and the influences of the experimental methodologies within which they meet. In general, results of the studies showed that WA research is a viable means of investigating processes in the mental lexicon and that aspects of current WA methodologies (e.g. response categorization schemes) were up to the task. The findings also showed that, in some instances, subjects respond in accordance with what appears to be an underlying response preference, or cognitive style. In other cases, however, preferred response types can be altered by experimental manipulation. This suggested that WA responses are determined by interaction between types of cues, response preferences, and the methods employed. In the final chapter, a dynamic model of the word association process (DMWA) is introduced. Based on the conclusions described above, the DMWA describes and predicts WA behaviour in terms of the interaction of cue properties, characteristics of the respondents, and WA methodology.