Over/generalization effects in the production of English weak and full vowels in unstressed syllables after vowel reduction training: evidence for L2 sound learning in progress

Poznan, Poland

In second language acquisition, those processes by which the learner is able to transfer what s/he has learnt to other settings where the linguistic rule applies (generalization) or where the rule does not apply (overgeneralization) are taken as evidence of language learning in progress. More specifically, context, lexical and speaker variability has led phonetic training studies to look at performance in environments other than those used in the training in order to test acquisition robustness (Rochet, 1995; Bradlow et al., 1997). Additionally, transfer to contexts where the L2 phonological process does not apply may be evidence of the activation of interlanguage phonological rules (Major, 1987). The present study analysed the production of weak and strong vowels in unstressed syllables by Spanish learners of English. The items were presented in minimal syllable (sherbet-alphabet) and word pairs (Upton-upturn) to two experimental groups who received specific training on vowel reduction by means of either perception or production tasks. Improvement in the production skills in both experimental groups transferred to two generalization contexts: embedding sentences and novel words. Furthermore, a significant deterioration of full vowel performance in unstressed syllables in the post-test was observed. Results will be discussed in terms of over/generalization evidence in the acquisition process and the need to further explore learning robustness in aspects such as retention over time and level of attainment.