Difficulties of Japanese learners of English in identifying English /r/ and /l/ sounds and possible methods to teach them to differentiate between the two phonemes.

The Japanese language has only one of the two phonemes /r/ and /l/, which in articulation is in between the two. Therefore, since the two sounds bear no distinctive feature in differentiating between two words, it is difficult for Japanese people to hear the difference between pairs such as lead and read. The topic is a widely researched one, so there is an abundance of materials to be found. The aim of this bibliography is to find out how improvement on the perception and production of the above-mentioned sounds are achieved with the help of special trainings and what causes the difficulty for Japanese learners. Listed below are the articles I have read in the matter.

Strange, W., & Sibylla, D. (1984). Effects of discrimination training on the perception of /r-l/ by Japanese adults learning English. Perception & Psychophysics, 36(2), 131–145.

The study argues that although intensive conversational instruction (with native speakers, for example) is connected with improved perception of the minimal contrasts, perception is still not native-like, even for the most advanced students. It suggests that modification of phonetic perception of phonemes appears to be slow and effortful in case of adults, and it is characterized by great variation among individuals. In the case of Japanese learners, the difficulty is true for the production and the perception of the /r-l/ sounds, as well.

After that, the study elaborates on the details of a training session, which included eight native speakers of Japanese who received pretraining and posttraining tests of natural speech minimal pairs that contrast /r/ and /l/ in different positions, and categorical perception tests as well, which included synthetic speech series containing the two sounds in word initial position. During the trainings candidates had to listen to minimal pairs that contrast the two phonemes (such as ‘rock’ and ‘ lock’) multiple times, by the end of which sessions their abilities of perceiving have increased. The mean number of correct identifications on the minimal-pair pretests was 69% and on the speech series tests with word-initial /r-l/ was 64%.. The study concluded that while one of the subjects showed no improvement over the training sessions, the other candidates did show improvement, which meant this time 75% of their answers were correct on the exercises during the post-tests.

Bradlow, A. R., Pisoni, D. B., Akahane-Yamada, R., & Tohkura, Y. (1997). Training Japanese listeners to identify English /r/ and /l/: IV. Some effects of perceptual learning on speech production. Journal of the Acoustical Society of America, 101(4), 2299-2310.

The article investigates the effects of training in /r-l/ perceptual identification on the production of the same two phonemes by adult Japanese speakers. The work refers to authors of other works, so called ‘motor theorists’ (Liberman and Mattingly, 1989), who claim that listeners perceive speech in terms of their own articulatory gestures that would produce the perceived sound. Motorist theorists claim that there is a phonetic module that represents speech units in terms of articulatory gestures, and that this module mediates both speech perception and production. In accordance with this view, the authors of the present study carried out a test in which they trained subjects (native speakers of Japanese) in the perception of the /r-l/ sounds and then investigated whether it improved their production of the same two phonemes.

During the training candidates were exposed to minimal pairs similarly to the previous training described above, i.e. they had to listen to such words repeatedly and they had to pronounce them.  The results showed that mean 65% correct identification on pretests improved to an 81% correct identification rate in the posttests, which were conducted after 3-4 weeks of training. Right after the posttest and right before the pretests recordings of the candidates producing words containing the two phonemes were made, and these records also testified a substantial improvement in their ability of producing the sounds. The improvement in production can be attributed to the improvement in perception. However, results also showed that only subjects who performed better on the perception tests underwent a significant improvement on the production tests. Candidates with poor performance on the perception tests did not show better results on either the perception posttests or the production tests. The cause of such individual differences still needs research, however.

Lively, S. E., Logan, J. S., & Pisoni D. B. (1993). Training Japanese listeners to identify English /r/ and /l/. II: The role of phonetic environment and talker variability in learning new perceptual categories. Journal of the Acoustical Society of America, 94, 1242–1255.
Retrieved from http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3509365/

This study is a report based on a training and series of tests, carried out similarly to the previous ones. Native speakers of Japanese were pretested and posttested and during the exercises they had to identify the /r/ and /l/ phonemes in different positions in different words. However, the authors managed to find evidence and conclusions that were not discovered or highlighted in earlier but similar tests.

Important observations of the conductors were the following. They found that the degree to which candidates managed to identify the phonemes was dependent upon the voice of the speaker. That is to say, if a new speaker started to pronounce the same words to them during the tests, they were less likely to recognize the phonemes /r/ and /l/ than in cases when an old speaker who already used to pronounce the words for them during previous tests did so with new words. The study also proved results of earlier studies which claimed when speakers acquire new phonetic constraints, they develop abstract, content-invariant units, such as phonemes in their long-term memory. Studies have shown that some mental representations seem to serve as the ‘best’ representatives of the category. Good category members are close to the prototype of the category and ‘poor’ category members are further away. This means that ‘better’ category members are better perceived and seem to have a more stable representation in the long-term memory. The conductors of the tests thought that training listeners with tokens that contrast /r/ and /l/ in initial position would allow them to form a prototype in their heads that could be applied to other phonetic environments. However, this generalization ignored the fact that phonemes have different spectral and durational differences during their realizations in different phonetic environments. Results have shown that the sounds in final singleton position had the longest contrast in duration; therefore, it was the easiest position for listeners to differentiate them in. Accordingly, the most difficult position was initial singleton position in which the contrast in duration was the shortest between /r/ and /l/. The study highlighted many different factors and positions influencing the degree of identification, but it failed to suggest possible training methods that work equally in every situation.