Abstract
One of the most important second language acquisition (SLA) trends of the past 15 years is applying cognitive linguistic (CL) analyses and principles to teaching traditionally challenging areas of English. This chapter continues in this vein by presenting a CL-based, effects-of-instruction study of English phrasal verb constructions (PVCs). The multiple meanings of PVC have largely been considered arbitrary and thus a particular challenge for L2 learners. Analyses aimed at finding systematicity have tended to focus exclusively on the contribution of the multiple meanings of the prepositions in PVCs and assumed a single meaning for the verb. However, using a CL-inspired framework (Goldberg, 2005; Langacker, 2008; Tyler & Evans, 2003), Mahpeykar and Tyler (2014) noted that verbs in PVC constructions also have multiple meanings. Their corpus-based analysis established that considering the contribution of the polysemy networks of both verbs and the prepositions allows for a systematic representation of the multiple meanings of PVCs. After presenting this framework, we discuss the results of a small-scale study which investigated using the CL framework as the basis for classroom instruction. The participants were enrolled in two, low intermediate classes in an intensive English programme; 11 participants received CL-based instruction on the multiple meanings of four PVCs; 14 received non-cognitive instruction. The intervention consisted of three hours of instruction, which combined interactive, teacher-fronted PowerPoint presentations interspersed with student centred, task-based instruction. All participants took a pre-test, immediate post-test and delayed post-test; each test item included a 'guessed'/`didn't guess' confidence rating. Accuracy results showed both groups improved (CL gain scores were 2.5 points, control were 0.9) but there were no significant between-group differences. However, the confidence ratings showed that the CL group's 'didn't guess' scores went from 1.5 to 9.1, while the control group's went from 6.3 to 9.8; a repeated analysis of variance (ANOVA) showed that the CL group's confidence gains by the post-test were significantly greater (p = 0.000); this gain was maintained for the delayed post-test.