Reiko A. Yamada

The effects of experimental variables on the perception of American English /r/ and /l/ by Japanese listeners.

The effects of variations in response categories, subjects’ perception of natural speech, and stimulus range on the identification of American English /r/ and /l/ by native speakers of Japanese were investigated. Three experiments using a synthesized /rait/-/lait/ series showed that all these variables affected identification and discrimination performance by Japanese-subjects. Furthermore, some of the perceptual characteristics of /r/ and /l/ for Japanese listeners were clarified: (1) Japanese listeners identified some of the stimuli of the series-as/w/.(2). Apositive correlation between the perception of synthesized stimuli and naturally-spoken stimuli was found. Japanese listeners who were able to easily identify naturally spoken stimuli perceived the synthetic series categorically but still perceived a /w/ category on the series. (3) The stimulus range showed a striking effect on identification consistency; identification of /r/ and /l/ was strongly affected by the stimulus range, the /w/ identification less so. This indicates that Japanese listeners tend to make relative judgments between /r/ and /l/.

Effect of extended training on /r/ and /l/ identification by native speakers of Japanese

It has been reported (1) that being exposed to the American English (AE) speaking environment in adulthood has less of an effect on the perception of AE /r/ and /l/ sounds for native speakers of Japanese than experience at younger ages, and (2) that laboratory training of those sounds on adult‐Japanese has little effect. Recently, however, Logan et al. [J. S. Logan et al., J. Acoust. Soc. Am. 89, 874–886 (1991)] and Lively et al. [J. Acoust. Soc. Am. (1993)] showed that identification training with natural tokens produced by multiple talkers was effective in improving Japanese listeners ability to identify /r/ and /l/. In the present paper, the effect of extended training was examined. Subjects were adult native speakers of Japanese with no experience in living abroad. The training procedure was identical to Lively et al.’s procedure, except that subjects were trained for 45 sessions and were given two midterm tests. The results showed that (1) most of the subjects, even those whose score was around 60% i...

Testing the importance of talker variability in non-native speech contrast training

In contrast to results of training with stimuli produced by 5 talkers, Lively et al. [ 1242–1255 (1993)] reported that Japanese adults trained to perceive English /r/ and /l/ with stimuli produced by a single talker failed to improve from pretest to post‐test, or to generalize to novel stimuli. That study was extended by training 5 groups of subjects each with a different talker, and by examining the retention of learning after 3 and 6 months. The previous results were partially replicated: Although all subjects showed significant learning during training, subjects in 3 of the 5 groups did not show significant improvement in a pretest–post‐test comparison, did not generalize well to new stimuli, and did not show good retention in 3‐ and 6‐month follow‐up tests. Subjects in two of the five groups improved significantly from pretest to post‐test, generalized well to new stimuli, and showed retention comparable to that of subjects trained with multiple talkers. The results indicate that while multiple‐talker...

Training Japanese listeners to identify English /r/ and /l/: A replication and extension

Monolingual native speakers of Japanese were trained to identify English /r/ and /l/ using a modified version of Logan, Lively, and Pisoni’s [J. S. Logan et al., J. Acoust. Soc. Am. 89, 874–886 (1991)] high variability training procedure. Both the talker’s voice and the phonetic environment were varied during training. Subjects improved in their ability to identify /r/ and /l/ from the pre‐test to the post‐test and during training. Generalization accuracy depended on the voice of the talker producing the /r/–/l/ contrasts: Subjects were significantly more accurate when words were produced by a familiar talker than when they were produced by an unfamiliar talker. Three months after the conclusion of training, subjects were given the post‐test and the tests of generalization again. Surprisingly, accuracy decreased only slightly on each test, even though no training or exposure to /r/ and /l/ occurred during the 3‐month interval. These results demonstrate that the high variability training paradigm is effect...

Effects of perceptual learning on the production of a non‐native contrast

Previous research has shown a significant correlation between the perception and production of English /r/ and /l/ by Japanese speakers [Yamada et al., Proc. ICSLP94, 2023–2026 (1994)]. The present study further investigated this perception‐production link by examining the transfer of training in perception to production of a non‐native contrast. Twelve monolingual Japanese speakers were trained to perceive the English /r/–/l/ contrast using a high‐variability training program [Lively et al., 2076–2087]. Recordings were also made of the trainees’ productions of English /r/‐/l/ minimal pairs before and after the perceptual training. These pretest and post‐test recordings were then evaluated perceptually by American listeners who were presented with pairs of tokens in an A‐B test format. As expected, subjects’ accuracy on the perceptual task improved by about 16% after 45 training sessions. More importantly, subjects’ productions at pretest and at post‐test were distinguishable by American listeners. More p...

Talker variability and the identification of American English /r/ and /l/ by Japanese subjects

Experiments are described that designed to investigate how Japanese subjects develop criteria for ‘‘R’’–‘‘L’’ decisions. Subjects were asked to identify English words beginning with /r/ and /l/ in two conditions: blocked (all words produced by one talker) and mixed (words produced by several talkers). Subjects performed more accurately in the blocked condition than in the mixed condition. This indicates that, as is generally true for native‐language perception, stable talker characteristics facilitate non‐native speech perception. Subjects also exhibited preferences for ‘‘R’’ or ‘‘L’’ responses for some talkers in one condition, but the opposite or no preference in the other. Together, these results indicate that in the blocked condition, subjects adopted talker‐specific criteria. However, when the talker changed randomly in the mixed condition, subjects were not able to adjust their criteria for each talker, but apparently based their criteria upon the range of all /r/s and /l/s they heard. Thus subjects...

Effects of speech style on the perceptual assimilation of American English vowels by Japanese speakers.

Of specific interest in this study was the extent to which speech style (citation form versus sentence form) affected perceptual assimilation of AE vowels by Japanese listeners. Four adult male speakers produced 11 AE vowels in /hVba/ bisyllables in both citation form (lists) and in the sentence ‘‘I hear the /hVb/ on the tape.’’ Response categories, designated in Katakana symbols, included five short vowels, five long vowels, and eight vowel combinations (e.g., ou, ei, ja:). Twenty‐four Japanese listeners selected the response category which was most similar to the /hV/ syllable they heard and rated its ‘‘goodness of fit.’’ Results indicated that, in general, AE vowels were assimilated to the Japanese vowel category closest in articulatory vowel space. However, intrinsic duration information was utilized much more consistently when the syllables were embedded in a carrier sentence. That is, intrinsically long (or diphthongized) AE vowels were assimilated to long‐vowel or two‐vowel response categories more...

Perception and production of the English /r/–/l/ contrast by Japanese speakers: Relationship between performance and learning in the two domains

This study investigated the relationship between performance in perception and production, and the relationship between degrees of learning in the two domains within individual subjects. Eleven monolingual Japanese adults participated in an /r/–/l/ perceptual training program. Both perception data (minimal pair identification scores) and production data (recordings of /r/–/l/ minimal pair productions) were collected before and after perceptual training. Pre‐ and post‐test production data were then evaluated by native speakers of American English in a minimal pair identification task. Results showed considerable individual variation in all aspects of the perception–production relationship. Subjects varied widely in pretest levels of performance in both perception and production, as well as in improvement in perception and production. In general, subjects who performed well in the perception pretest also had good productions at pretest; however, subjects who were low performers in the perception pretest var...

A new masked spectrum representation applied to English /r/–/l/ dissimilarity measurement

This paper proves that a dynamic cepstrum is effective not only in automatic speech recognition, but also in explaining the speech perception mechanism. Talker dependency on the perception of American English /r/–/l/ in Japanese listeners has been reported [J. S. Logan et al., J. Acoust. Soc. Am. 89, 874 (1991)]. In order to interpret this phenomenon with regard to the acoustical properties of the stimuli, talker dependency is assumed to be caused by the acoustical dissimilarity of the stimuli. This paper applies a dynamic cepstrum to measuring the dissimilarity of the stimuli. The dynamic cepstrum is a new spectral representation for automatic speech recognition that incorporates the time‐frequency characteristics of forward masking [K. Aikawa et al., J. Acoust. Soc. Am. 92, 2476(A) (1992)]. This parameter enhances formant shifts and suppresses stationary spectral features. In a perception experiment, identification tests for English /r/–/l/ minimal pairs uttered by five talkers were conducted. The dynam...

Perceptual characteristics of English syllable‐initial /r,l/ for Japanese listeners

This study investigates the perceptual characteristics of American English /r,l/ for Japanese listeners using synthesized stimuli. Five major findings are obtained. (1) The Japanese listeners identify the stimuli using a variety of acoustic cues, and their response patterns are strongly influenced by acoustic features of the stimuli. In contrast, the American listeners can identify /r/ and /l/ as long as a primary cue remains, even under the condition where some of the acoustic cues are missing. (2) As the Japanese listeners tend to perceive some stimuli as /w/ more than American listeners do, perception experiments with /w/ as well as /r/ and /l/ for a choice of identification better clarify the perception mode of the Japanese listeners. (3) A positive relationship between the identification ability of the natural /r,l,w/ spoken by native Americans and that of the synthesized /r,l,w/ is found for the Japanese listeners. (4) Contextual effects in words are very strong for the Japanese listeners when tryin...