Adapting second language phonemic perception training to common instructional situations: A progress report

Abstract

Although current L2 pedagogy de‐emphasizes phoneme‐level pronunciation training, laboratory experiments demonstrate benefits from training non‐natives in perception of difficult target‐language phonemic contrasts. Specifically, evidence shows that: learners' perceptual performance improves (Jamieson & Morosan, 1986; Flege, 1995), improvements generalize to new talkers and words (Lively, Logan, & Pisoni, 1993), perceptual training triggers production improvements (i.e., without production training, Bradlow et al., 1997), and both perceptual (Lively et al., 1994) and production improvements (Bradlow et al., 1999) are maintained over several months. We are adapting perceptual training methods from such studies for use in common instructional situations, starting with Japanese students learning the English /l‐r/ contrast. This paper addresses three practical concerns. First, studies show that the perceptual training needs tokens with multiple kinds of variation. We discuss phonological variation in corpus design, sociolinguistic variation in talker recruitment, and inducement of within‐talker variations. Second the training must be usable both independently and in classrooms. We discuss the design of sessions short enough for inclusion as part of daily classroom activities. Third, the training must be computer‐controlled. We discuss necessary functions, available programs, and our choice. Finally, we briefly demonstrate the training, and give a sketch of interim results.