What do Timed Grammaticality Judgment Tests reveal within the framework of implicit and explicit knowledge?

Abstract

This study attempted to explore the construct validity of Grammaticality Judgment Tests (GJTs) on the basis of Loewen (2009). A particular focus was placed on a relationship between L2 learners' accurate performance on timed GJTs (expected to tap implicit knowledge) and their self-reported use of ‘feel’ in making judgments in terms of task stimulus (grammatical versus ungrammatical), and 17 grammatical structures. The participants were 90 Japanese learners of English from two different universities in Japan. They conducted timed (with 4-second response time) and untimed (without time constraints) GJTs. The data from a correlation analysis reveal that out of the 17 different target structures, the accurate performance of only one item Plural-s was significantly correlated with the reported use of ‘feel’ whereas neither grammatical nor ungrammatical items as a whole were. The investigation of Plural-s in terms of learning difficulty of implicit knowledge proposed by Ellis (2006) shows that Plural-s is relatively easy due to its simple Functional value and easy Processability. However, the accuracy order from the timed GJT reveals that the structure ranks 12th. With the help of the self-report analysis, it was found that there seem to exist various factors that affect the degree of difficulty such as sentence comprehension. Furthermore, although the use of ‘feel’ was reported subjectively, the results seem to highlight the importance of considering ideal response time in order to ensure that timed GJTs can tap implicit knowledge. It seems that a combined use of subjective and objective analysis might further contribute to the exploration of GJTs' validity issue.