Feedback in Intelligent Computer-Assisted Language Learning and Second Language Acquisition: A study of its effect on the acquisition of French past tense aspect using an Intelligent Language Tutoring System

Abstract

Questions surrounding the impact of feedback in response to learner error are of interest in the fields of both Second Language Acquisition (SLA) and Intelligent Computer-Assisted Language Learning (ICALL). Current empirical SLA research seeks to ascertain what feedback types have a statistically significant positive impact on the process of acquiring a second language. Similarly, research in ICALL focuses on testing Intelligent Language Tutoring Systems (ILTSs) generally as well as the effectiveness of the feedback that they deliver. Despite this common interest in feedback, to date there has been no significant interdisciplinary research involving the two fields. The experiment reported here seeks to bridge this gap. Using a purpose-built ILTS, we tested the effect of two types of feedback on the acquisition of French past tense aspect among anglophone learners. Inspired by previous work in SLA, Explicit Inductive (EI) and Input Processing (IP) feedback were tested against a control group using a pre test/post test design. The learners completed a transformation and a grammaticality judgment task. For the transformation, they were presented with texts in the present tense and asked to re-write them in the past tense. For the grammaticality judgment, they had to rank the grammaticality of each sentence in a set of texts. In response to errors, EI feedback interpreted the aspectual meaning of the learners' answer and explicitly told them that it was not the most natural according to the context. In order to encourage formmeaning mapping, IP feedback asked the learners to match their erroneous answer to its interpretation. Two interpretations were presented: one was the target interpretation and the other matched the learner's answer. Having made their choice, they were then told whether it was correct as well as which interpretation was in fact target-like. The quantitative evaluation of the effectiveness of the EI and IP feedback was not statistically significant. We argued that this was due to a combined effect of learner level, target structure and feedback.