Synthetic Continua Research Articles

Effects of syllable-final segment duration on the identification of synthetic speech continua by birds and humans

In an attempt to test whether experience with or knowledge of language is necessary to show typical speaking rate effects in the perception of speech, budgerigars (Melopsittacus undulatus) and humans categorized stimuli from the synthetic continua /ba/-/wa/ and /bas/-/was/, with both short and long syllable-final phonemes. This comparative approach aims to shed some light on whether knowledge of language has a role in rate normalization effects, such as using duration information as an indicator of speaking rate in human speech perception. Syllable-final phoneme durations were varied, and were either temporally adjacent to the initial target (CV series) or were nonadjacent (CVC series). The birds were always influenced by syllable-final duration variation in the present experiments and displayed greater boundary shifts than humans. In humans, there was a significant boundary shift observed in the CV series, but there were no effects of duration variation in the final segment in the CVC series. The results from the birds suggest that specialized speech-based principles may not be necessary for explaining findings of grouping speech or speechlike elements in perception.

Developmental Role of Static, Dynamic, and Contextual Cues in Speech Perception

The purpose of the current study was to examine the role of syllable duration context as well as static and dynamic acoustic properties in child and adult speech perception. Ten adults and eleven 4-5-year-old children identified a syllable as [ba] or [wa] (stop-glide contrast) in 3 conditions differing in synthetic continua. The 1st condition tested the potential existence of the syllable duration effect in young children, whereas the 2nd and 3rd conditions examined the developmental role of static and dynamic cues, respectively, as related to syllable duration context effects. In the 1st condition, the 1st and 2nd formant transition duration of stimuli varied from those appropriate for [ba] to those appropriate for [wa]. For the 2nd condition, a static burst was added to Condition 1 stimuli. For the 3rd condition, the dynamic transition frequency and transition duration for the first 3 formants varied as appropriate for [ba] and [wa]. In each condition, 3 syllable context durations of 105 ms, 170 ms, and 315 ms were tested. The results indicated that syllable duration context effects were present across all conditions for both adults and children. However, the adults and children did differ in the 3rd condition, in which both the transition frequency and the transition duration were altered. Thus, children used the dynamic formant transitions differently than adults when transition frequency was varied along with transition duration. These findings show that children have a bias toward formant transitions and indicate that young children of 4-5 years of age attend differently than adults to changes in dynamic cues such as formant transitions, as predicted by the developmental cue weighting shift model.

The development of phonemic categorization in children aged 6–12

The aim of this study was to assess the development of phonemic categorization across a range of phonemic contrasts: /g/-/k/, /d/-/g/, /s/-/z/ and /s/-/∫/. More specifically, we aimed to investigate the age at which children achieve adult-like competence, both in terms of their consistency in categorizing a range of phonemic contrasts, and in terms of their ability to categorize stimuli with limited acoustic cue information. Six-stimulus synthetic continua were created in which acoustic cues signaling these contrasts were manipulated singly or in combination. Stimuli were presented to 84 normally-hearing children aged between 6;0 and 12;6 years and 13 adult controls in the form of two-alternative forced-choice identification tests using an adaptive procedure. The gradients of the identification functions increased significantly in steepness between the ages of 6 and 12 but, by 12 years, children were still not, on average, categorizing the phonemic contrasts as consistently as adults. This study provides therefore further evidence that phoneme boundary sharpening occurs well into the second decade of life. Children were also less consistent than adults in categorizing continua containing limited acoustic cue information. Children aged 6–12 therefore appear to show less flexibility in their perceptual strategies than adults.

Effect of lexical status on children's and adults' perception of native and non-native vowels

The purpose of this study was to identify possible age-related changes in vowel perception and to assess lexical influences on both children's and adults' perception. More specifically, we sought to evaluate two proposals regarding the development of speech categories — namely, the category expansion and category definition hypotheses (e.g., Flege, 1992). In Experiment 1, 5-yr-old, 9-yr-old and adult monolingual, native speakers of English identified vowels from two synthetic continua in the nonword context /h_b/. Vowels on the “native” continuum ranged from English /I/ to /i/; those on the “foreign” continuum ranged from English /I/ to an unfamilar, foreign vowel /Y/. Young children's phoneme boundary extended further away from the /I/ endpoint on the foreign continuum than did older children's and adults' — a result opposite to that predicted by the category expansion hypothesis. However, in support of the category definition hypothesis, an age-related increase in the slopes of subjects' identification functions was observed, especially for the native continuum. In Experiment 2, the same vowel stimuli were presented in the contexts /b_b/ and /b_p/ to 5-yr-olds, 9-yr-olds and adults; thus, one endpoint for the native /I–i/ continua always formed a word (viz, “bib” or “beep”), whereas, for the foreign /I–Y/ continua, the only word endpoint was “bib”. It was again found that young children's phoneme boundary extended further away from the /I/ endpoint, when this vowel was not bounded by another native vowel. In addition, the slopes of their identification functions were steeper, and thus more like those of older listeners, especially when the endpoint stimuli were real words. The results suggest that despite similarities in the extent of native vowel categories for young and older listeners, young children's categories are still quite flexible. Moreover, developmental differences in how sharply defined category boundaries are may depend, in part, on variations in lexical knowledge.

Perception of stops in naturally produced V[l/r][b/d/g]V tokens: a reassessment of acoustic cues to place of articulation

This perceptual experiment reassesses a traditional assumption that the F3 transition is the primary cue for distinguishing [da] from [ga], evidence of which comes from perceptual experiments using synthetic continua [Mann, Percept. Psychophys. 5, 407–412 (1980); Fowler et al., ibid. 48, 559–570 (1990); Liberman, Speech (1996)]. Stimuli were tokens of /alba/, /alda/, /alga/, /arba/, /arda/, /arga/ produced three times by four speakers at three rates (fast, normal, slow), analyzed acoustically by Dainora et al. [J. Acoust. Soc. Am. 100, 2601 (1996)] and Hemphill et al. [CLS 33, 333–346 (1997)]. Subjects identified tokens in their original carrier sentences using a forced choice paradigm. Stimuli were blocked by speaker and rate; three repetitions of each were randomized within blocks. In five of the 216 tokens, for one or more subjects, the percept of the stop did not match the speaker’s intent. Analysis of these tokens showed a mismatch between formant transitions and burst spectra: formant transitions were characteristic of the stop intended by the speaker, while the burst spectra corresponded to the actual percept. These results suggest that when the burst spectra and formant transitions are in conflict, some listeners use the burst as primary for identifying place of articulation.

Effects of lexical status on native and non-native English-speaking adults’ vowel perception

Monolingual, English-speaking adults and native Chinese (Mandarin) speakers who learned English as a second language heard stimuli from two ‘‘native,’’ synthetic continua, in which the vowels ranged from English /i/ to /i/ in the context /b–b/ or /b–p/. Thus the end points of the first continuum constituted an English word and a nonword (‘‘bib’’ vs *‘‘beeb’’); the reverse held for the second continuum (*‘‘bip’’ vs ‘‘beep’’). These same subjects also heard stimuli from two ‘‘foreign’’ continua, where the vowels ranged from English /i/ to a foreign (non-English) vowel /Y/ in the contexts described above. Thus the end points of the first continuum corresponded to a word and a nonword (‘‘bib’’ vs *‘‘bYb’’); both end points of the second continuum corresponded to nonwords (*‘‘bip’’ vs *‘‘bYp’’). After training on end points, subjects’ identifications of the nine stimuli of a given continuum were examined to assess whether: the Chinese speakers, like native English speakers, exhibit a ‘‘lexical bias’’ effect for English vowels (from the native continua); vowel categories not bounded by another native/English vowel (as in the foreign continua) expand outward or become better defined with increased exposure to English and/or lexical status. [Work supported by NIH.]

Effects of first formant onset frequency on [-voice] judgments result from auditory processes not specific to humans

When F1-onset frequency is lower, longer F1 cut-back (VOT) is required for human listeners to perceive synthesized stop consonants as voiceless. K. R. Kluender [J. Acoust. Soc. Am. 90, 83-96 (1991)] found comparable effects of F1-onset frequency on the "labeling" of stop consonants by Japanese quail (coturnix coturnix japonica) trained to distinguish stop consonants varying in F1 cut-back. In that study, CVs were synthesized with natural-like rising F1 transitions, and endpoint training stimuli differed in the onset frequency of F1 because a longer cut-back resulted in a higher F1 onset. In order to assess whether earlier results were due to auditory predispositions or due to animals having learned the natural covariance between F1 cut-back and F1-onset frequency, the present experiment was conducted with synthetic continua having either a relatively low (375 Hz) or high (750 Hz) constant-frequency F1. Six birds were trained to respond differentially to endpoint stimuli from three series of synthesized /CV/s varying in duration of F1 cut-back. Second and third formant transitions were appropriate for labial, alveolar, or velar stops. Despite the fact that there was no opportunity for animal subjects to use experienced covariation of F1-onset frequency and F1 cut-back, quail typically exhibited shorter labeling boundaries (more voiceless stops) for intermediate stimuli of the continua when F1 frequency was higher. Responses by human subjects listening to the same stimuli were also collected. Results lend support to the earlier conclusion that part or all of the effect of F1 onset frequency on perception of voicing may be adequately explained by general auditory processes.(ABSTRACT TRUNCATED AT 250 WORDS)

Effects of lexical status on children’s and adults’ perception of native and non-native vowels

Monolingual, English-speaking 5-year-olds, 9-year-olds, and adults heard stimuli from two ‘‘native,’’ synthetic continua, in which the vowels ranged from English /i/ to /i/ in the context /b—b/ or /b—p/. Thus the endpoints of the first continuum constituted a word and a nonword (‘‘bib’’ vs *‘‘beeb’’); the reverse held for the second continuum (*‘‘bip’’ vs ‘‘beep’’). Other subjects heard stimuli from two ‘‘foreign’’ continua, where the vowels ranged from English /i/ to a foreign vowel /y/ in the contexts described above. Thus the endpoints of the first continuum corresponded to a word and a nonword (‘‘bib’’ vs *‘‘bYb’’); both endpoints of the second continuum corresponded to nonwords (*‘‘bip’’ vs *‘‘bYp’’). After training on endpoints, subjects’ identifications of the nine stimuli of a given continuum were examined to assess whether: children, like adults, exhibit a ‘‘lexical bias’’ effect for familiar vowels (from the ‘‘native’’ continua); vowel categories not bounded by another native vowel (as in the ‘‘foreign’’ continua) expand outward or become better defined with increasing age and/or lexical status.

Categorical perception of silent-center syllables

Work by Strange and colleagues [e.g., Strange etal., J. Acoust. Soc. Am. 74, 695–705 (1983)] has shown that CVC syllables without the steady-state vowel (‘‘silent-center syllables’’) are as accurately recognized as CVC syllables containing the steady-state vowel, suggesting that dynamic rather than static properties are more important in vowel perception. The present study expands this line of research by studying the identification of silent-center syllables as members of a vowel continuum. Two synthetic continua were presented to listeners for identification: A [bib–b(eh)b] continuum and a silent-center [bib–b(eh)b] continuum. The silent-center version was created by excising the steady-state portion of the original continuum, leaving the formant transitions intact. A comparison of the two identification functions revealed a significant shift, suggesting that full and silent-center syllables are not perceptually equivalent in this type of identification task. The results will be discussed in terms of theories of vowel perception.

Effects of first formant onset frequency on voicing judgments result from auditory processes not specific to humans

When Fl onset frequency is lower, a longer F1 cutback (VOT) is required for human listeners to perceive synthesized stop consonants as voiceless. K. R. Kluender [J. Acoust. Soc. Am. 90, 83–96 (1991)] found comparable effects of F1 onset frequency on the ‘‘labeling’’ of stop consonants by Japanese quail (Coturnixcoturnixjaponica) trained to distinguish stop consonants varying in F1 cutback. In that study, CV’s were synthesized with natural-like rising F1 transitions. Consequently, endpoint training stimuli (with 5- or 65-ms F1 cutback) differed in the onset frequency of F1 because a longer cutback resulted in a higher F1 onset. In order to assess whether earlier results were due to auditory predispositions or due to animals learning the covariance between F1 cutback and F1 onset frequency, the present experiment was conducted with synthetic continua having either a relatively low (375-Hz) or high (750-Hz) constant-frequency F1. Eight birds were trained to respond differentially to endpoint stimuli from three series of synthesized /Ca/’s varying in duration of F1 cutback. Second and third format transitions were appropriate for labial, alveolar, or velar stops. Despite the fact that there was no opportunity to use experienced covariation of F1 onset frequency and F1 cutback, quail exhibited reliably shorter ‘‘labeling’’ boundaries (more voiceless stops) for intermediate stimuli of the continua when F1 frequency was higher. Human performance with the same stimuli was like that for the birds. These results support the conclusion that the effect of F1 onset frequency may be adequately explained by general auditory processes. [Work supported by NIDCD/NIH Grant No. DC-00719.]

Perceived modulation magnitudes of second‐formant, center‐frequency variations

Speech signals convey information via rapid changes in formant frequencies. Listeners can reliably judge the relative “modulation magnitude” of frequency modulations [e.g., H. Fastl, Hear. Res., 59–69 (1982)]. Listeners were asked if they could also judge modulation magnitude for center‐frequency‐modulated formants by presenting pairs of 500‐ms stimuli with centered, “notchlike” modulations drawn from synthetic continua with varying notch duration (30–300 ms) or depth (50–500 Hz). The first, reference, stimulus (either a low‐ or high‐modulation extreme of a continuum) was assigned a reference magnitude (1 for low‐modulation‐reference trial blocks; 10 for high‐). Listeners assigned a relative “modulation strength” (1 to 10) to the second stimulus. Judged magnitude appears as an uneven perceptual surface in the magnitude × depth × duration space: Highest modulation strengths are heard for deep‐long stimuli; deep‐short stimuli have slightly lower judged magnitudes; shallow‐long modulations are much weaker; shallow‐short modulations are the weakest. The possible relation between perceived modulation magnitude and judgments of articulatory‐phonetic differences will be discussed. [Work supported in part by NIH‐NINCDS and the Louisiana Lions Eye Foundation.]

Enhanced amplitude of the first harmonic as a correlate of voicelessness in aspirated consonants

Acoustic analysis was undertaken to test the hypothesis that information signaling voicelessness era preceding aspirated consonant may reside in spectral characteristics associated with “breathy voice” at the onset of the following vowel. The study focused upon the enhancement of the first harmonic (H1) resulting from an increase in the open quotient of the voicing source waveform. The relative amplitudes of H1 and H2 were measured in vowels in CVd syllables pairing the consonants /p,t,k,h,b,d,g/ with the vowels /i,ɑ,u,e,Λ/. The syllables were spoken by three male native speakers of English. Harmonic amplitudes were measured from DFT spectra (computed without preemphasis) of the first few pitch periods of the vowel. In general, the amplitude of H1 was greater than that of H2 for vowels following voiceless consonants; the converse was true for voiced consonants. A perceptual study using synthetic continua (varying in VOT) is being conducted; the continua differ in relative amplitude of H1 and H2. Preliminary results indicate that for some subjects an enhanced H1 may contribute to a voiceless percept: The boundary between voiced and voiceless stops was shifted to smaller VOT values for stimuli in which the amplitude of H1 was greater than that of H2 at vowel onset. [Work supported by NIH.]

Vowel categorization by 3-year-old children

A new two-alternative forced-choice procedure was used to assess perceptual constancy for vowel categories in 3-year-olds. The stimuli were edited natural tokens of /i/ and /a/ produced by male and female adults and male and female children. Subjects were visually reinforced for making an appropriate pointing response to one of two televisions. Initial training of the pointing response consisted of pairing an adult male /i/ with one television and an adult male /a/ with the other. After reaching an 80% correct criterion on the training stimuli, subjects were presented with blocks of eight stimuli (four /i/s and four /a/s). For subjects in the vowel group, all the /i/s were assigned to one pointing response, and all the /a/s to the other. For subjects in the arbitrary group, two /i/s and wo /a/s were assigned to each pointing response. The vowel group showed nearly perfect perceptual constancy for vowel categories by ignoring differences in speaker; the arbitrary group performed near chance. The success of this 2AFC pointing procedure will be discussed with respect to additional data on vowel normalization and identification of tokens from various synthetic continua.

Developmental use of vowel duration in perception and production

This study was designed to investigate developmental use of vowel duration as a cue to postvocalic consonant voicing in both perception and production. The subject sample consisted of 10 three-year-olds, ten six-year-olds, and ten adults, with normal articulation and language skills and normal hearing sensitivity. For the perception study, steady-state vowel duration was varied in each of the three synthetic continua BIP-BIB, POT-POD, BACK-BAG. The percentage of responses judged as voiced was calculated by subject for each synthetic stimulus value and plotted against vowel duration, yielding 90 sets of identification data. The logistic function was fit to these data by maximum likelihood techniques. Two parameters, crossover and slope, were obtained from each fitted function and served as dependent measures for statistical analyses. For the production study, total vowel duration was measured in ten trials of each of the six test words, BIP, BIB, POT, POD, BACK, and BAG, for each speaker. Comparative analyses between the perception and production data revealed a refinement in the use of vowel duration, with an increase in age.