Exploring a vocabulary test and a judgment task as diagnoses of early and late bilinguals' L2 proficiency

Abstract

This psycholinguistics study delved into the adequacy of a measure of vocabulary size the Vocabulary Levels Test (VLT) as a predictor of Brazilian Portuguese-English speakers and Spanish Heritage Language (HL) speakers capacity to access grammatical representations when using their non-dominant language. Considering the scenario with two types of bilinguals (L2 learners and HL), two types of knowledge (implicit and explicit), and two types of proficiency measures (vocabulary size and sentence judgment), we conducted an exploratory study designed in two experiments. In the first experiment, we compared participants performances in the VLT (English version), an overall abilities test (Oxford Placement Test OPT), and a Speeded Acceptability Judgment (SAJ) task in English, in which a ceiling of 8 seconds was set for each judgment call. In the second experiment, we compared participants performances in the VLT (Spanish version), the Spanish Placement Test (SPT), a SAJ task in Spanish, in which a ceiling of 6 seconds was set for each judgment call, and a self-assessment test on participants language abilities. Moreover in the second experiment, we employed a Bilingual History Questionnaire (BHQ) to operationalize HL speakers language use/dominance. For both experiments, our SAJ task stimuli were composed of 56 sentences, and 16 of them contained grammatical violations. There were two types of sentence violations applied to 8 sentences each: argument structure realization violations involving unergative verbs in transitive syntax, and explicit morpho-syntactic violations involving long-distance dependencies (Wh-movement) and subject-verb agreement. In both experiments, the VLT, the OPT/SPT and the SAJ tasks were submitted to the Receiver Operating Characteristic (ROC) curve procedures for the assessment of their sensitivity and specificity as diagnostic tests of language proficiency. Results for both experiments show that only those participants who classified as high proficiency in the VLT and the OPT/SPT were capable of detecting grammatical violations. We interpret the results as indicating that a measure of vocabulary size is a predictor of both fluency in lexical access and fluency in grammatical knowledge access of Second Language (L2) and HL speakers. Moreover, ROC curve analysis from both experiments revealed that the VLT and the SAJ task are adequate instruments for language proficiency diagnosis, since they are able to differentiate two groups of proficiency.