One of the most vexing aspects of learning English conversation for students in Japan seems to be mastering the phoneme /l/, and this study sets out to describe the linguistic phenomenon in detail. Just over 100 first-year college students in four conversation classes were recorded speaking in pairs over a threemonth period-in pre-, mid-, and post-course tests. The /l/ frequency errors and linguistic environments were tabulated. It was found that just 10.9% of all /l/s uttered in all

more »

... s were deemed unacceptable or nonnativelike. Students made significantly more errors on the pre-course test than on the mid-and post-course tests. But, when just two words which also appear in the Japanese language with katakana pronunciationspolicy and complex-were eliminated from the pre-course test tabulations, the progress made by students during the course was not statistically significant. A final interesting finding was that /l/ was the hardest to pronounce when between vowels (eg., hello) or in consonant clusters (eg., play). The classroom implications for teachers are described. BACKGROUND When trying to learn English conversation, few linguistic phenomena seem to give students in Japan more anxiety than the phoneme /l/. If not learned as children, it is often hard for native Japanese speakers to & Flege, 2001). The first is the fact that the Japanese language does not have an alveolar /l/ or a rhotic /r/, but rather has a flap [ɺ] which is between an /l/ and /r/. The English /l/ requires contact with the alveolar ridge and raising of the tongue dorsum, especially when syllable-final. The Japanese [ɺ] has the tongue briefly touching the roof of the mouth just behind the alveolar ridge. With English /r/, the tip of the tongue never touches the roof of the mouth. A possible second factor working against Japanese learning /l/ is that people who begin to learn English at a later age-perhaps past the age of 12-often have much more difficulty, because the brains of adolescents and adults seem to lose "neuroplasticity" (Lenneberg, 1967) . Known as the critical period hypothesis, or the sensitive period hypothesis, this posits that children are much better at learning pronunciation and grammar -38 -than adults. Other factors affecting /l/ acquisition are English-learning experiences and motivation. Despite these obstacles, a lot of research also shows that even older Japanese students of English can learn to pronounce /l/ well. Flege et al. (1995) found that 10 out of 12 Japanese learners of English who had lived abroad for at least a dozen years produced /l/ as well as native speakers, and produced /l/ significantly better than a group which had less than 3 years of foreign residence. Yamada (1995) also found that some Japanese citizens who had lived in the US could discriminate between /l/ and /r/ as well as native English speakers. Larson-Hall (2006) found that Japanese abroad improved their /l/ pronunciation even with shorter time residences. With respect to the perception of /l/, research shows that perception ability can be significantly improved when students get minimal-pair training (eg., rock/lock) while listening to multiple speakers (MacKain et al.,1981; Iverson et al., 2003; Lively et al., 1993; Logan et al., 1991; Uther et al., 2008 ). Lively et al. (1994 , for example, found that a three-week training focusing on minimal pairs significantly improved students' abilities to distinguish /l/ from /r/. The students retained some perception abilities three months after training, and a bit less ability six months after training. However, when listening to speakers different from the ones they had learned from during training, the perception accuracy went down. With respect to production of /l/, there are reports of effective training dating back to Goto (1971) . More recently, Hattori (2009) found that native Japanese speakers "approached a 100% accuracy ceiling" when producing /l/ after they received one-on-one training, including explicit instructions and feedback in the form of spectrograms showing their own production. Bradlow et al. (1997) found that there is a link between perception and production. They showed that perceptual learning generally resulted in better production. In fact, in that study the students from Japan could improve /l/ production without any production practice! Hattori (2009) also found a significant correlation between the ability to identify the difference between /l/ and /r/ and the ability to produce them. Interestingly, Hattori also found that the students were better at producing /l/ than /r/. The best teaching methods for pronunciation focus on form while engaging in meaningful communication (Lyster,