Terry L. Gottfried

Trading relations in the perception of /r/‐/l/ by Japanese learners of English

The role of language-specific factors in phonetically based trading relations was examined by assessing the ability of 20 native Japanese speakers to identify and discriminate stimuli of two synthetic /r/-/l/ series that varied temporal and spectral parameters independently. Results of forced-choice identification and oddity discrimination tasks showed that the nine Japanese subjects who were able to identify /r/ and /l/ reliably demonstrated a trading relation similar to that of Americans. Discrimination results reflected the perceptual equivalence of temporal and spectral parameters. Discrimination by the 11 Japanese subjects who were unable to identify the /r/-/l/ series differed significantly from the skilled Japanese subjects and native English speakers. However, their performance could not be predicted on the basis of acoustic dissimilarity alone. These results provide evidence that the trading relation between temporal and spectral cues for the /r/-/l/ contrast is not solely attributable to general auditory or language-universal phonetic processing constraints, but rather is also a function of phonemic processes that can be modified in the course of learning a second language.

Musical experience and Mandarin tone discrimination and imitation

Previous work [T. L. Gottfried and D. Riester, J. Acoust. Soc. Am. 108, 2604 (2000)] showed that native speakers of American English with musical training performed better than nonmusicians when identifying the four distinctive tones of Mandarin Chinese (high‐level, mid‐rising, low‐dipping, high‐falling). Accuracy for both groups was relatively low since listeners were not trained on the phonemic contrasts. Current research compares musicians and nonmusicians on discrimination and imitation of unfamiliar tones. Listeners were presented with two different Mandarin words that had either the same or different tones; listeners indicated whether the tones were same or different. Thus, they were required to determine a categorical match (same or different tone), rather than an auditory match. All listeners had significantly more difficulty discriminating between mid‐rising and low‐dipping tones than with other contrasts. Listeners with more musical training showed significantly greater accuracy in their discrim...

Intelligibility of vowels sung by a countertenor

Ten American English vowels were sung in a /b/-vowel-/d/ consonantal context by a professional countertenor in full voice (at F0 = 130, 165, 220, 260, and 330 Hz) and in head voice (at F0 = 220, 260, 330, 440, and 520 Hz). Four identification tests were prepared using the entire syllable or the center 200-ms portion of either the full-voice tokens or the head-voice tokens. Listeners attempted to identify each vowel by circling the appropriate word on their answer sheets. Errors were more frequent when the vowels were sung at higher F0. In addition, removal of the consonantal context markedly increased identification errors for both the head-voice and full-voice conditions. Back vowels were misidentified significantly more often than front vowels. For equal F0 values, listeners were significantly more accurate in identifying the head-voice stimuli. Acoustical analysis suggests that the difference of intelligibility between head and full voice may have been due to the head voice having more energy in the first harmonic than the full voice.

Identification of coarticulated vowels

Previous explanations of vowel perception held that the most definitive information for vowel identity is the relatively constant formant frequencies in the steady-state portions of vowels. Perceptual studies indicate, however, that vowels spoken in syllables with labial stop consonants are identified more accurately than vowels spoken in isolation. The present study investigated the nature and scope of this consonantal context advantage in the perception of ten American English vowels spoken by adult male and female speakers. Vowels in /p/-vowel-/p/, /b/-vowel-/b/, /k/-vowel-/k/, /k/, /k/-vowel, and vowel-/k/ syllables were identified much more accurately than isolated vowels. This is consistent with the hypothesis that dynamic acoustic information due to the coarticulation in syllables is important for vowel identification. Identification of vowels in /g/-vowel-/g/, /g/-vowel, and vowel-/g/ syllables was not better than isolated vowels and was significantly poorer than for other consonantal contexts. Acoustical analyses were performed to determine whether poor production of vowels could account for perceptual errors. Misproduced vowel targets could not account for the overall pattern of identification performance. Phonological factors were also considered but were found to be inadequate to account fully for the results.

Relation of pitch glide perception and Mandarin tone identification

Second language (L2) learners differ considerably in their ability to learn new phonetic contrasts in the L2. For example, Gottfried and Suiter (1997) observed that many native speakers of American English were highly accurate in their identification of Mandarin Chinese tones, while others, with similar years of experience in learning Mandarin, were very poor. The current study compared listeners’ identification of Mandarin tones with their ability to perceive the pitch direction of sine‐wave glides. Sine waves of 400 ms with starting frequencies of 250 and 300 Hz were created with 0‐ to 50‐Hz sweeps up or down from the initial frequencies. Listeners who had no experience with Mandarin indicated whether the pitch went up, down, or remained the same. After testing on the sine waves, listeners identified the tone (high‐level, mid‐rising, low‐dipping, and high‐falling) of intact and silent‐center (with only the initial and final portions) Mandarin /l/‐vowel syllables. Analysis revealed that listeners with mu...

Effect of Speaking Rate on the Perception of Vowels

Three experiments examined the conditions under which the speaking rate of a context sentence affects vowel identification. In these experiments, listeners identified the vowel in synthetic /b/-vowel-/t/ syllables that varied systematically in the duration (temporal), the formant frequencies (spectral), or both the duration and formant frequencies (temporal-spectral) of the steady-state portion of the syllable. These syllables were embedded in two synthetic sentence frames, one with the temporal characteristics of a natural fast sentence and one with those of a slow sentence. For two vowel distinctions that are specified in natural speech by both temporal and spectral characteristics, /I/-/i/ and /e/-/ae/, listeners adjusted their identification of the vowels according to the sentence rate in all three conditions. Although there was a trend for the rate effect to be reduced in the temporal-spectral condition, the influence of sentential rate was never eliminated. By contrast, for a vowel distinction that is naturally specified primarily by spectral characteristics, /e/-/I/, there was no effect of sentence rate in any of the conditions. We conclude that when vowels are differentiated in natural speech by both temporal and spectral information, listeners obligatorily use the duration of the vowel to identify it and do so in relation to the rate of the sentence in which the vowel occurs.

Task variables in the study of vowel perception

Vowels produced in isolation by six speakers were identified less accurately than vowels coarticulated in /k/-vowel-/k/ syllables, whether tested by a key word task or by a rhyming task. Performance by naive listeners in the two tasks was highly correlated.

Perception of temporal versus spectral information in French vowels

Perceptual effects of orthogonal variations in temporal and spectral information differentiating French /o/ and /ɔ/were examined. Although both parameters contribute to acoustic differentiation of /o/ and /ɔ/, several phonetic and phonological characteristics of French lead us to expect that duration is a less important perceptual dimension in French than in languages like American English. Three 10‐step /kot/‐/kot/ continua were synthesized by systematically varying F1 and F2 of the vowel nuclei. The three continua differed in vowel duration: 200, 160, and 120 ms. Two perceptual tests, identification (labeling stimuli as /o/ or /ɔ/) and rating (judging vowel category goodness on a five‐point scale), were presented to two groups of listeners: native French and native American English learning French. Preliminary results indicate language group differences. Native French listeners showed little effect of vowel duration in either perceptual task. Americans generally showed displacements of the /o/‐/ɔ/ bound...

Training musicians and nonmusicians to discriminate Mandarin tones

Musicians and nonmusicians were compared in their ability to learn Mandarin tone contrasts. Participants were pretested to assess their accuracy in tone discrimination. Some were then given six sessions of articulatory and acoustic training over 3 weeks; others were given no training. At the end of 3 weeks, participants were again tested on their discrimination accuracy. Listeners judged whether the tones (tone 1, high level; tone 2, mid rising; tone 3, low dipping; tone 4, high falling) of two different Mandarin syllables were the same or different. In some trials, listeners discriminated tones in words spoken by two speakers; in other trials, they discriminated tones in words spoken by one speaker only. The pretest revealed that musicians are significantly more accurate in discriminating Mandarin tones than nonmusicians. Musicians and nonmusicians who received the training regimen performed more accurately than untrained listeners on the posttest discrimination. Musicians showed significantly less impro...

Production of Mandarin tone contrasts by musicians and non‐musicians

Previous research has shown that native speakers of American English who are musicians perform better than non‐musicians when identifying and producing the four phonemic tones of Mandarin. The present study corroborates these findings and analyzes acoustic properties of non‐natives’ tonal imitations. Listeners imitated Mandarin two‐syllable word phrases that varied the vowel (/li/, /la/, /lu/) and tone (high‐level, mid‐rising, low‐dipping, high‐falling) of the first syllable. Four native Mandarin speakers rated the musicians, on average, to be better in their imitation of Mandarin tone 4 (high‐falling) than non‐musicians. There were no significant differences between groups in how they were rated on the other three tones. Acoustical analyses revealed that non‐natives failed to match native speakers both in differences in initial F0 and in F0 contour (change from initial to final F0) across tones. Imitations by musicians did not show significant acoustic differences from non‐musicians, except for tone 4, w...