Self-rated language proficiency as a context-dependent judgment

While self-ratings of proficiency are widely used in bilingualism research and clinical settings, previous work has found that they often lack reliability within and across bilinguals, across languages, tasks and over time. Yet little is known about the potential effect of conducting these ratings early versus late in a testing session, and specifically, before or after objective tasks of proficiency. We examine this question in a sample of unbalanced Hebrew-English bilinguals (N = 186), who completed self-ratings of proficiency in their two languages both before and after they completed a multiple-choice vocabulary test in one of their languages. Importantly, we manipulated the difficulty of the task (easy vs. difficult words) to further probe the role of context in self-judgment of proficiency. Findings revealed that self-ratings were influenced by recent experience: regardless of language, when the vocabulary task was difficult, self-ratings in the language of the task lowered. Moreover, when the difficult task was in the native language, the self-ratings in the second language also decreased, even though the second language was not tested, and post- rather than pre-task self-ratings tended to be more strongly correlated with vocabulary scores. Overall, these findings show that judgments of self-proficiency are dynamic, context-dependent inferences shaped both by prior beliefs of proficiency and immediate experience, underscoring the need to consider individual and task factors when collecting self-reported measures.