Stability of grammaticality judgments in German-English code-switching

Code-switching, a bilingual language phenomenon, which may be defined as the concurrent use of two or more languages by fluent speakers is frequently misunderstood and stigmatized. Given that the majority of the world's population is bilingual rather than monolingual, the study of code-switching provides a fundamental window into human cognition and the systematic structural outcomes of language contact. Intra-sentential code-switching is said to systematically occur, constrained by the lexicons of each respective language. In order to access information about the acceptability of certain switches, linguists often elicit grammaticality judgments from bilingual informants. In current linguistic research, grammaticality judgment tasks are often scrutinized on account of the lack of stability of responses to individual sentences. Although this claim is largely motivated by research on monolingual strings under a variety of variable conditions, the stability of code-switched grammaticality judgment data given by bilingual informants has yet to be systematically investigated. By comparing grammaticality judgment data from 3 groups of German-English bilinguals, Group A (N=50), Group B (N=34), and Group C (N=40), this thesis investigates the stability of grammaticality judgments in code-switching over time, as well as a potential difference in judgments between judgment data for spoken and written code-switching stimuli. Using a web-based survey, informants were asked to give ratings of each code-switched token. The results were computed and findings from a correlated groups t test attest to the stability of code-switched judgment data over time with a p value of .271 and to the validity of the methodologies currently in place. Furthermore, results from the study also indicated that no statistically significant difference was found between spoken and written judgment data as computed with an independent groups t test resulting in a p value of .186, contributing a valuable fact to the body of data collection practices in research in bilingualism. Results from this study indicate that there are significant differences attributable to language dominance for specific token types, which were calculated using an ANOVA test. However, when using group composite scores of all tokens, the ANOVA measure returned a non-significant score of .234, suggesting that bilinguals with differing language dominances rank in a similar manner. The findings from this study hope to help clarify current practices in code-switching research.