DISERTASI dan TESIS Program Pascasarjana UM, 2019

Ukuran Huruf:  Kecil  Sedang  Besar


Eka Fadilah


This study aims to investigate the effect of Task-Based Language Instruction (TSLI) and two corrective feedback constructs: situated recast (SR) and negotiated feedback (NF) embedded in the TSLI on the learners’ knowldge of grammatical feature: be-passive voice construction. TSLI alone embraces task-based activities (i.e., pre-TSLI and During TSLI) in which pre-task encompasses the discussion of the form, use, and function of be-passive voice, while during-task entails the learners to perform some language utterances or outputs (e.g., procedure of doing or making something, dictoglos, enhanced-information gap game). On the other hand, the two constructs of corrective feedback are inserted during TSLI activities. While the former is seen from a cognitive-interactionist paradigm, the latter is viewed from a socio-cultural paradigm. Additionally, the former put in a flexible-move strategy via raising intonation and segmenting the part of utterances that differs from the implicit (pure) recast maintaining covert strategy, the latter embraced implicit-explicit strategy employing the quality of the lecturer and learners’ negotiation of form-meaning mapping under Zone Proximal Development (ZPD).

Utilizing a quasi experimental design, four intact-class encompassing 127 leraners of Indonesian Economics Higher-School (STIESIA/Sekolah Tinggi Ilmu Ekonomi Indonesia) were assigned to four groups involving one control group and three experimental groups. The former group was provided with pre- and post-test only, while the rests were supplemented with delayed-test and supplied with different interventions: one group was provided with task-supported language activities without any corrective feedback supervised (TSLI only), another group was provided and supplied with such activities supplemented with situated recast (TSLI+SR), and the other group was supplemented with those activities and negotiated feedback (TSLI+NF). Two testing instruments utilized were Error Correction Test (ECT) used as measuring explicit knowledge and Elicited Imitation Test (EIT) realized as measuring implicit knowledge.

A mixed-design (within-between) repeated measure analysis of variance (RM-ANOVA) was utilized to gauge the groups’ pre, post, and delayed tests. This mixed-design model was applied to investigate the development of both individuals within the groups and the comparisons between groups. prior to analyzing the data, some statistical assumptions were set up: group homogeneity, normal distribution, and sphericity indicating no violations of such assumptions relied on the level of significance (p>.05). Furthermore, to detect the statistical difference and interaction of the groups and tests, Wilk’s Lambda was employed. This measurement entails that when p-value