THE EFFECTIVENESS OF ERROR CORRECTION DURING ORAL INTERACTION: EXPERIMENTAL STUDIES WITH ENGLISH L2 LEARNERS IN THE UNITED KINGDOM AND SAUDI ARABIA

The current classroom experiments examined the effects of two types of oral corrective feedback (CF), recast and metalinguistic information, during oral production tasks on the learning of English modals (will, can and must). These techniques were compared to an intervention with identical oral production tasks but in which CF was not provided. The study also investigated the extent to which instructional setting (EFL in Saudi Arabia and ESL in the United Kingdom) and learners' attitudes towards CF mediated the effect of CF on learning. Pre-intermediate adult learners of English as a second language (ESL) in the United Kingdom (UK; n=36) and of English as a foreign language (EFL) in Saudi Arabia (SA; n=64) were randomly assigned to one of three conditions: metalinguistic information; recast, or task only (no CF). Participants had four sessions within these conditions, over four consecutive weeks, delivered by a trained and experienced EFL teacher (the author). In the UK context, learners’ knowledge was measured, in pre, post and delayed posttests, using a free oral picture description, a timed grammaticality judgment and a written gap fill. In the SA context knowledge was measured by similar written gap fill and the same picture description, but also an elicited imitation, and un-timed grammaticality judgment with explicit knowledge probes. In both contexts, an exit questionnaire (similar to Sheen’s, 2006) was administered after each test to check awareness of the target feature being tested. In addition, an attitudinal questionnaire was used to measure the possible role of participants’ attitude towards error correction and grammatical accuracy. Tests of normality were used to decide whether parametric or non-parametric statistical tests were required. Equivalence between groups at pre-test was checked to determine whether actual scores or gain scores (or ANCOVAs, if parametric tests) should be analysed. The results suggested that both metalinguistic information and recasts can be beneficial for the development of English modals, though effectiveness was influenced by the outcome measures used, the length of time between intervention and test, and the context (UK and SA). Recast and metalinguistic information were generally found to be beneficial in most measures regardless of contexts. In most measures, task only group in the UK had no significant gains but in SA had significant gains. The study indicated that learners had an equal preference for recast and metalinguistic information CF in the EFL context but preference for recast was pronounced in the ESL context.