On the reliability of grammaticality judgement tests in second language acquisition research

The use of grammaticality and metalinguistic judgement tests in second language acquisition (SLA) research has been the subject of considerable scrutiny over the past decade (see, for example, Chaudron, 1983; Birdsong, 1989; Ellis, 1991; Cowan and Hatasa, 1994; Gass, 1994; Davies and Kaplan, 1998).Grammaticality judgement (GJ) test data in research design are used to make inferences about the syntactic structures and rules that constitute learners’ linguistic competence, in the Chomskian sense of the word. One criticism levelled at the use of this type of test, however, is that they are not reliable measures of linguistic competence. The present study compared GJ test data with dehydrated sentence (DS) test data, an assessment tool commonly used in the L2 (second language) classroom. Data were collected from three levels (second, fourth and sixth semester) of adult L2 learners of Spanish about verb movement (Vmovement). The results from the comparison of the two tests indicated that GJ data are reliable measures of linguistic knowledge.