Does training to perceive L2 English vowels in one phonetic context transfer to other phonetic contexts?

Résumé

Implicit in many pronunciation curricula and materials is the belief that training learners how to perceive and produce speech categories in one context will generalize to novel contexts. In the second language (L2) speech perception literature, however, many hold the opposite view. Flege’s  (1995) Speech Learning Model has been the catalyst for research demonstrating that knowledge of a sound category in one phonetic context does not predict knowledge of the same category elsewhere.  This view has more recently been supported by Thomson (2012), Thomson, Nearey and Derwing (2009), and Thomson and Isaacs (2009).  The present study extends this line of research by investigating English vowel learning by 15 L1 Spanish speakers in 12 separate phonetic contexts.  Following the High Variability Phonetic Training paradigm (Logan, Lively & Pisoni, 1991), participants were trained to perceive ten English vowels during 40 computer-based training sessions, over a period of approximately two months (with nearly 8000 training tokens per learner).  While learners were taught the same vowels during each session, the phonetic context in which vowels were presented varied. Results indicate that while training improves learners’ vowel perception within specific contexts, generalization of learning from one context to another does not occur. These findings appear to have major implications for pronunciation curriculum, indicating, for example, that instructors need to provide learners with vowel instruction in all possible phonetic contexts. Recommendations for further research will also be discussed.