VALIDATING GRAMMATICALITY JUDGMENT TESTS

Abstract

Several previous factor-analytic studies on the construct validity of grammaticality judgment tests (GJTs) concluded that untimed GJTs measure explicit knowledge (EK) and timed GJTs measure implicit knowledge (IK) (Bowles, 2011; R. Ellis, 2005; R. Ellis & Loewen, 2007). It has also been shown that, irrespective of the time condition chosen, GJTs’ grammatical sentences tap into IK, whereas their ungrammatical ones invoke EK (Gutiérrez, 2013). The current study examined these conclusions by employing two more fine-grained measures of IK: that is, a self-paced reading task and a word-monitoring task. The results of a confirmatory factor analysis revealed that manipulating GJTs’ time conditions and/or the grammaticality of the sentences does not render them distinct measures of EK and IK. The current work shows that GJTs are too coarse to be measures of IK, and that the different types of GJTs measure different levels of EK.