Investigating linguistic knowledge of a second language and its relationship to general language proficiency and individual learner differences in an EFL context

Abstract

A psycholinguistic modelling of language proficiency in terms of L2 linguistic knowledge (i.e., implicit and explicit knowledge) can contribute to both SLA theories and language testing practice. Yet to date few empirical studies have attempted this. The main reason is the lack of validated instruments for measuring implicit and explicit knowledge. The current study is intended to fill the this gap by examining three important issues: (1) whether the four tests (i.e., the Elicited Imitation Test, Timed Grammaticality Judgment Test, Untimed Grammaticality Judgment Test and Metalinguistic Knowledge Test) that R. Ellis (2009) showed can distinguish implicit and explicit knowledge of an L2 can also provide separate measures of these two types of L2 knowledge in a foreign language context (i.e. mainland China); (2) whether language proficiency can be conceptualised in terms of these two types of knowledge, and (3) whether four psychological factors (i.e., language analytic ability, language learning motivation, language anxiety and learner beliefs) can distinguish learners in terms of their implicit and explicit knowledge A total of 192 full-time university English majors participated in this study. 100 were first year students and 92 were third year students. They were English major students in a fouryear programme at a key university in Beijing, China. Four of the tests used in R. Ellis’s (2009) study were employed to distinguish implicit and explicit knowledge. The findings showed that R. Ellis’s model of implicit and explicit knowledge could account for the first year students’ results but that it failed to do so for the third year students. A factor analysis of the third year students’ test scores identified three rather than two factors. These were interpreted as follows: (1) the automated declarative knowledge needed for online production; (2) the non-automated declarative knowledge needed to judge ungrammaticality, and (3) the more automated declarative nowledge needed to judge grammatical sentences. The results for the third year students led to a re-evaluation of those for the first year students. It is proposed that both groups of learners were drawing on their declarative (explicit) knowledge and that various tests distinguished whether they were drawing on their automated declarative knowledge or their non-automated. This re-interpretation supports DeKeyser’s (2009) claims that older learners are unlikely to develop true implicit knowledge and instead have to rely on their explicit knowledge and that automated declarative knowledge is functionally equivalent to implicit knowledge. The findings of this study shed light on the relationship between general language proficiency and L2 linguistic knowledge. Two general proficiency tests were administered – the Oxford Placement Test (for the first year students) and the Test for English Majors, Band 4 (for the third year students). Scores representing the two factors identified for the first year students and the three factors for the third year students all entered into multiple regression analyses with the proficiency scores as the dependent variables. However, in both groups, it was the most explicit factor that proved the strongest predictor of proficiency. This suggests that these learners were primarily drawing on their non-automated declarative knowledge in the proficiency tests supporting Elder’s (2009) contention that proficiency tests by their very nature incline learners to make use of their explicit knowledge. Finally, the study examined the relationship between measures of individual difference factors and their scores on the battery of tests. Language anxiety was negatively correlated with measures of declarative knowledge supporting the findings of previous studies that have investigated this learner factor. However, somewhat surprisingly, very few other significant relationships were found although limited qualitative data derived from interviews with a small sample of the first and third year students suggested that individual learner differences were at work. Thus the thesis points to the need for more qualitative investigation of learner difference factors. This thesis concludes with a consideration of the theoretical and practical implications of the study and also points to several limitations (e.g., the battery of tests did not include the Oral Narrative Production Task used in R. Ellis’s study).