Consonant Release Research Articles

Production and perception of the consonantal voicing distinction in speech without VOT: An acoustic study using electric larynx speech.

Acoustic features and perceptual cues for the voicing distinction of intervocalic stops were investigated using electric larynx speech. Alaryngeal talkers who speak American English with an electric larynx of the transcervical type produced multiple repetitions of /schwa CVC/ syllables. As the sound source of the electric larynx was continuously generated throughout an utterance, systematic control of voice onset time (VOT) was not available for the intervocalic stops in this speech. Statistical analyses based on the talkers’ intention revealed that duration and amplitude of high-frequency noise following the consonant release were increased for the voiceless stops than for the voiced cognates. The F1 frequency value of the following vowel was higher for the voiceless stops than for the voiced cognates. The major finding of the perceptual experiment was that duration and amplitude of the above-mentioned high-frequency noise were highly correlated with the perception of ‘‘voicelessness.’’ Articulatory efforts made by the electric larynx talkers and the perceptual effectiveness of the efforts will be discussed. [Research supported, in part, by NIH Grant to Haskins Laboratories.]

Models for voicing and aspiration in stop consonants

A theoretical analysis of sound production for voiceless and voiced stop consonants has been carried out. The analysis includes the effects of active and passive (in response to intraoral pressure changes) expansion or contraction of the pharyngeal volume, active and passive changes in glottal configuration, the generation of periodic glottal vibration and aspiration noise, and the generation of transient and frication noise sources at the consonantal release. Both monopole and dipole components of the turbulence noise sources are considered. The model is used to calculate the absolute sound‐pressure levels and spectra for the various components of the radiated sound for labial, alveolar, and velar stops. The predictions of the model are in reasonable agreement (within a few dB) with data from spoken syllables. Procedures for estimating the characteristics of turbulence noise sources still need to be refined, particularly the distribution of the sources downstream from a glottal or supraglottal constriction. The model can be used to infer details of the time course of supraglottal and laryngeal constrictions based on data from fine‐grained acoustic analysis of utterances. [Research supported in part by NIH Grant DC00075.]

Perceptual Evidence Against Internal Structure in American Sign Language Syllables

Syllables in spoken languages have been argued to have an internal structure which may cluster the consonants and vowels into subgroups. The most commonly used subgroups are the Onset (initial consonants) and Rhyme (vowel and remaining consonants), with the further possibility of dividing the Rhyme into Nucleus (the vowel and any on- or off-glides) and the Coda (the final consonants). Although linguistic debate continues about the exact formulation of such syllable models, the evidence is clear that listeners identify rhythmic beats at a particular point inside the syllable, namely at the release of the initial consonants into the following linguistically stressed vowels (referred to as the syllable peak, assumed to be part of the Nucleus). To determine whether such loci would be identified for syllables in American Sign Language, deaf native signers, hearing native signers, and hearing subjects unfamiliar with sign language were asked to tap to videotaped signed stimuli. Analysis of the tap locations in signed syllables revealed a relatively flat distribution across the syllable for all three groups and for individual subjects. The absence of syllable peak perception is discussed in relation to the motoric characteristics of sign production.

Articulatory and acoustic differences between palatal and velar stops

Many experimental studies have shown that velar consonants have a more forward place of articulation in front vowel contexts than in back vowel contexts. On some accounts, such velar fronting is basically the same process as the secondary articulation of palatalization, and furthermore, both “fronted” and “palatalized” velars are equated with the “palatal” cononants of other languages. In this study, velar stops (in English, Russian, Czech, Hungarian) are compared with palatal stops (in Czech, Hungarian) and with palatalized velar stops (in Russian) in different vowel contexts. X-ray and palatographic data from the literature show that palatalized velars are very much like velars in high front vowel context, but that both of these are very different from palatals. Acoustic analysis comparing spectra of consonant release and the following vowel agrees with the articulatory data. It also makes clear that languages with palatals also have fronted velars, differing in the distribution of energy across frequencies. These results indicate that palatalization of velars is phonetically equivalent to fronting, but that neither fronted nor palatalized velars are like the palatal consonants.

Acoustic properties and perception of stop consonant release transients.

This study focuses on the initial component of the stop consonant release burst, the release transient. In theory, the transient, because of its impulselike source, should contain much information about the vocal tract configuration at release, but it is usually weak in intensity and difficult to isolate from the accompanying frication in natural speech. For this investigation, a human talker produced isolated release transients of /b,d,g/ in nine vocalic contexts by whispering these syllables very quietly. He also produced the corresponding CV syllables with regular phonation for comparison. Spectral analyses showed the isolated transients to have a clearly defined formant structure, which was not seen in natural release bursts, whose spectra were dominated by the frication noise. The formant frequencies varied systematically with both consonant place of articulation and vocalic context. Perceptual experiments showed that listeners can identify both consonants and vowels from isolated transients, though not very accurately. Knowing one of the two segments in advance did not help, but when the transients were followed by a compatible synthetic, steady-state vowel, consonant identification improved somewhat. On the whole, isolated transients, despite their clear formant structure, provided only partial information for consonant identification, but no less so, it seems, than excerpted natural release bursts. The information conveyed by artificially isolated transients and by natural (frication-dominated) release bursts appears to be perceptually equivalent.

Release mechanisms for stop and nasal consonants

Theoretical models have been developed to account for the rapidly changing acoustic events at the release of stop and nasal consonants with different places of articulation. The model for stop consonants predicts the level and spectrum of the initial transient, the burst of noise, aspiration noise, and formant transitions, and shows differences in the contribution of each of these components over the initial 10–20 ms for labial, alveolar, and velar places of articulation. For nasal consonants, the abrupt change in spectrum at the release can be explained by rapid movement of the zeros of the transfer function of the oral-nasal system, and by a somewhat slower movement of the poles. The pattern of movement of the zeros and hence the nature of the spectrum change depend on the consonantal place of articulation. Spectra data for spoken stop and nasal consonants, obtained with fine time resolution, show general agreement with the theoretical predictions, although the lack of quantitative data on dynamic and kinematic articulatory parameters makes detailed comparison difficult. [Work supported in part by a grant from NINCDS.]

Voice source variations during consonant-vowel transitions

Studies of the voice source during speech have been concerned mostly with variations in fundamental frequency and amplitude. However, the source variations also comprise the spectral characteristics and the harmonics-to-noise ratio. The present study reports detailed measurements of voice source properties during transitions between consonants and vowels. Oral air flow, oral air pressure, and transillumination were recorded from two subjects producing reiterant speech with the vowel /æ/ and different voiced and voiceless consonants; the flow signal was inverse-filtered to obtain an estimate of the glottal pulse. Results indicate a breathy phonation at the release of voiceless consonants as indicated by an open quotient close to 1. Peak flow during each glottal pulse is thus high at voicing onset following a voiceless consonant and decreases, with an undershoot, during the first 20 pitch periods. The source pulse is also skewed to the left during the first pitch periods following voiceless consonants. Source variations following voiced consonants are less pronounced. [Work supported by NINCDS Grants NS-13617 to Haskins Laboratories.]

Formant onset duration: A secondary cue for identifying the place of articulation in stop consonants

Formant onset duration (FOD) is suggested as a possible alternative to voice onset time (VOT), to characterize the place of articulation in stop consonants. The FOD is defined as the duration from the start of the burst (denoting consonantal release) to the start of the formant structure (denoting vowel onset). In many languages, VOT is the most salient cue distinguishing between voiced and voiceless stop phonemes. Although VOT has been shown to increase in duration from bilabials to alveolars to velars [Lisker and Abramson, Word 20; Zue, Tech. Rep. 523, MIT], attempts to use VOT as a cue to place of articulation [Kewley‐Port, J. Acoust. Soc. Am. 73, 322–335 (1983)] were not successful. In a recent study of stop consonants using the partially smoothed Wigner distribution [Garudadri et al., J. Acoust. Soc. Am. Suppl. 1 82, S55 (1987)], FOD and F2 of the following vowel were used along with the spectral shape during the burst, to resolve context dependencies due to coarticulation effects. The FOD increases from bilabials to alveolars to velars and appears to be related to the inertia of the articulators. Unlike VOT, FOD is always positive and corresponds to the vocal tract rather than the source of excitation. Evidence from English, Telugu, and French stop consonants supporting these ideas is presented. [Work supported in part by NSERC, Canada.]

Acoustic properties of isolated stop consonant release transients

This study focuses on the initial component of the stop consonant release burst, the release transient. In theory, the transient, because of its impulselike source, should contain much information about the vocal tract configuration at release, but it is usually weak in intensity and difficult to isolate from the accompanying frication in natural speech. For this investigation, a human talker produced isolated release transients of /b,d,g/ in nine vocalic contexts by “mouthing” these syllables. He also produced the corresponding CV syllables with regular phonation for comparison. Spectral analyses showed the isolated transients to have a clearly defined formant structure, which was not seen in natural release bursts whose spectrum was dominated by the frication noise. The formant frequencies varied systematically with both consonant place of articulation and vocalic context, with the latter, “coarticulatory” influence actually exceeding the former. The transient thus provides an acoustic snapshot of the vocal tract immediately after the release. [Work supported by NICHD.]

The stop–glide distinction: Acoustic analysis and perceptual effect of variation in syllable amplitude envelope for initial /b/ and /w/

Amplitude change at consonantal release has been proposed as an invariant acoustic property distinguishing between the classes of stops and glides [Mack and Blumstein, J. Acoust. Soc. Am. 73, 1739-1750 (1983)]. Following procedures of Mack and Blumstein, we measured the amplitude change in the vicinity of the consonantal release for two speakers. The results for one speaker matched those of Mack and Blumstein, while those for the second speaker showed some differences. In a subsequent experiment, we tested the hypothesis that a difference in amplitude change serves as an invariant perceptual cue for distinguishing between continuants and noncontinuants, and more specifically, as a critical cue for identifying stops and glides [Shinn and Blumstein, J. Acoust. Soc. Am. 75, 1243-1252 (1984)]. Interchanging the amplitude envelopes of natural /bV/ and /wV/ syllables containing the same vowel had little effect on perception: 97% of all syllables were identified as originally produced. Thus, although amplitude change in the vicinity of consonantal release may distinguish acoustically between stops and glides with some consistency, the change is not fully invariant, and certainly does not seem to be a critical perceptual cue in natural speech.

Acoustic analysis and perceptual effect of variation in the amplitude envelopes of /b/ and /w/

Relative energy change at consonantal release has been proposed as an acoustic property, invariant across vowel contexts and speakers, to distinguish between stops and glides [M. Mack and S. E. Blumstein, J. Acoust. Soc. Am. 73, 1739–1750 (1983)]. We followed the procedures of Mack and Blumstein to measure distributions of energy in /bV/ and /wV/ syllables, spoken by two speakers before five vowels. Results for one speaker matched those of Mack and Blumstein; results for a second speaker did not. In a subsequent perceptual experiment, amplitude envelopes of /bV/ and /wV/ syllables containing the same vowels were crossed. Envelope crossing had little effect on the syllables of one speaker: Nearly all were perceived as originally produced. For the second speaker /w/ to crossing also had little effect, but /b/ to /w/ crossing elicited more than 75% /b/ responses, particularly before /a/. Thus although relative energy change at consonantal release may separate /b/ from /w/, acoustically and perceptually, in some contexts, for some speakers, it is evidently not invariant. [Work supported by NICHD.]

On the role of the amplitude envelope for the perception of [b] and [w

This study investigated the role of the amplitude envelope in the vicinity of consonantal release in the perception of the stop-glide contrast. Three sets of acoustic [b-w] continua, each in the vowel environments [a] and [i], were synthesized using parameters derived from natural speech. In the first set, amplitude, formant frequency, and duration characteristics were interpolated between exemplar stop and glide endpoints. In the second set, formant frequency and duration characteristics were interpolated, but all stimuli were given a stop amplitude envelope. The third set was like the second, except that all stimuli were given a glide amplitude envelope. Subjects were given both forced-choice and free-identification tasks. The results of the forced-choice task indicated that amplitude cues were able to override transition slope, duration, and formant frequency cues in the perception of the stop-glide contrast. However, results from the free-identification task showed that, although presence of a stop amplitude envelope turned all stimuli otherwise labeled as glides to stops, the presence of a glide amplitude envelope changed stimuli labeled otherwise as stops to fricatives rather than to glides. These results support the view that the amplitude envelope in the vicinity of the consonantal release is a critical acoustic property for the continuant / noncontinuant contrast. The results are discussed in relation to a theory of acoustic invariance.

Variable perceptual potency of the initial 20 ms time window

The major focus of this experiment was examination of the contribution or “perceptual potency” of the initial 20 ms “time window” associated with voiceless stop consonant release. An electronic gating procedure was used to achieve consonant burst elimination; the remaining aspiration + vocalic transition + vowel (A + VT + V) segments of six meaningful voiceless stop + vowel CV and eve syllables (/pi, ti, ki, pap, tap, kap/) were presented to 30 adult female listeners for identification. Each stimulus syllable was originally produced in sentence-initial position within a constant carrier phrase at slow, moderate, and fast rates of articulation. Listeners were presented the segments from each target syllable both with the accompanying carrier phrase and in “isolation”. Although no advantage for carrier phrase context was observed, the perceptual potency of the 20 ms time window was not constant but varied significantly as a function of both type of syllable construction and rate of articulation.

Effect of burst amplitude on the perception of stop consonant place of articulation.

We have examined the effects of the relative amplitude of the release burst on perception of the place of articulation of utterance-initial voiceless and voiced stop consonants. The amplitude of the burst, which occurs within the first 10-15 ms following consonant release, was systematically varied in 5-dB steps from -10 to +10 dB relative to a "normal" burst amplitude for two labial-to-alveolar synthetic speech continua--one comprising voiceless stops and the other, voiced stops. The distribution of spectral energy in the bursts for the labial and alveolar stops at the ends of the continuum was consistent with the spectrum shapes observed in natural utterances, and intermediate shapes were used for intermediate stimuli on the continuum. The results of identification tests with these stimuli showed that the relative amplitude of the burst significantly affected the perception of the place of articulation of both voiceless and voiced stops, but the effect was greater for the former than the latter. The results are consistent with a view that two basic properties contribute to the labial-alveolar distinction in English. One of these is determined by the time course of the change in amplitude in the high-frequency range (above 2500 Hz) in the few tens of ms following consonantal release, and the other is determined by the frequencies of spectral peaks associated with the second and third formants in relation to the first formant.

Fundamental frequency perturbation caused by voiced and voiceless stops in Japanese

The study investigates the effect of voicing of stops on the fundamental frequency (F0) of the following vowels in Japanese, particularly, how this effect interacts with phonemic accentual patterns. A speaker of Standard Japanese recorded/Can/syllables where C was/p,t,k,b,d,g,m,n/. F0 was measured for each glottal period after the consonantal release, or after the vowel onset in the case of aspirated stops. The results showed a dip in F0 around 40 ms after the release of voiced stops, while F0 generally stayed higher after voiceless stops. To examine the interaction between consonantally‐induced F0 variations and accentual patterns, the F0 contours in C1VC2V words with /t,d,n/ as C1 with either high‐low or low‐high accent were analyzed. In the high‐low accent, the low initial F0 after a voiced stop caused F0 to peak later in time than after a voiceless stop. This pattern was not observed in the low‐high accent, where the initial dip was followed by a fast increase in F0 after a voiced stop, while the pitc...

Knowledge‐based recognition of voiceless stop consonants

Recent experiments [M. Bush et al., Proc. 1983 IEEE ICASSP] have demonstrated the ability of a trained spectrogram reader to identify initial stops in /CVb/ syllables from a table of numerical acoustic measurements with approximately 80% accuracy. This paper discusses an automatic system for discriminating between the voiceless plosives (/p,t,k/)which is based on the features and rules identified in these experiments. Ten binary features are extracted from two linear prediction spectra which are computed during the 35 ms following the consonant release. Typical features include “back‐k‐release‐spectrum” and “compact‐release‐spectrum.” The features are detected by examining the frequencies and amplitudes of the local maxima and minima of the two LPC spectra, in a manner motivted by the actions of the human spectrogram reader. A simple statistical classifier is used to combine the outputs of the ten feature detectors. The classifier was trained on the 108 /p,t,k/ tokens of the multi‐speaker corpus used in the spectrogram reading experiment. When tested on the training data, the system achieved 96% correct recognition. When tested on two additional data sets of similar composition, the system achieved scores of 94% and 92%, respectively.

The b/w contrast and considerations of syllable duration

Recent research suggests that a set of acoustic properties are ambiguous with regard to their categorization as /b/ or /w/ depending on the duration of the following vowel [J. Miller and A. Liberman, Percept. Psychophys. 25, 457–465 (1979)]. Such evidence causes difficulty for a theory of acoustic invariance [S. Blumstein and K. N. Stevens, J. Acoust. Soc. Am. 66, 1001–1017 (1981)], since identification of properties at the onset appears to be dependent on the context of later-occurring acoustic material. We repeated earlier experiments [J. Miller and A. Liberman, Percept. Psychophys. 25, 457–465 (1979)] with some variation. While maintaining continua of stimuli between /b/ and /w/, our stimuli were made more natural by manipulating the spectrum of voicing prior to consonant release. The consequence was weak prevoicing and an abrupt onset at the /b/ end, and stronger prevoicing and a gradual onset at the /w/ end. Results showed that the tendency for identification of mid-valued stimuli to be affected by the following vowel duration was reduced. These findings suggest that earlier results may be explained by the structure of the synthetic stimuli and by properties of auditory processing. [Work supported in part by NSF and by NINCDS.]

Parameter interaction in the perception of French plosives.

This investigation takes as its point of departure observations on the production of French fortis versus lenis plosive consonants. In a series of perception tests the role of different parameters in the fortis versus lenis distinction found in production (e.g. duration of the pre-consonantal vowel, the consonantal closure, voicing during the closure and intensity of the consonantal release noise) was investigated. Apart from durational and intensity parameters, vowel quality differences (depending on the fortis versus lenis character of the following consonant) are also shown to play a part in phoneme identification. Furthermore it is shown that in obstruent clusters like /tp, dp/ versus /tg, dg/ not only do phonetic parameters like those mentioned above influence perception, but that also the phonemic identity of /p/ versus /g/ as fortis versus lenis contributes to the hearers' decision about the identity of the first consonant in the cluster. Phonetic as well as contextual factors are shown to interact in a complex manner. Implications of the results for speech perception theory are discussed.

Time-varying features as correlates of place of articulation in stop consonants.

Running spectral displays derived from linear prediction analysis were used to examine the initial 40 ms of stop-vowel CV syllables for possible acoustic correlates to place of articulation. Known spectral and temporal properties associated with the stop consonant release gesture were used to define a set of three-time-varying features observable in the visual displays. Judges identified place of articulation using these proposed features from running spectra of the syllables /b,d,g/paired with eight vowels produced by three talkers. Average correct identification of place was 88%; identification was better for the male talkers (92%) than the one female talker (78%). Post hoc analyses suggested, however, that simple rules could be incorporated in the feature definitions to account for differences in vocal tract size. The nature of the information contained in linear prediction running spectra was analyzed further to take account of known properties of the peripheral auditory system. The three proposed time-varying features were shown to be displayed robustly in auditory filtered running spectra. The advantages of describing acoustic correlates for place from the dynamically varying temporal and spectral information in running spectra is discussed with regard to the static template matching approach advocated recently by Blumstein and Stevens [J. Acoust. Soc. Am. 66, 1001-1017 (1979)].

Speech evoked activity in the auditory radiations and cortex of the awake monkey

To determine whether phonetic features of human speech are reflected in activity patterns of the auditory cortex and its thalamic afferents, concurrent recordings of multiple unit activity (MUA) and averaged evoked potentials (AEP) to 3 synthetic syllables: /da/, /ba/ and /ta/, were performed in awake monkeys. Using clicks, responses from thalamocortical axons and cortical cells were differentiated on the basis of their response latency, spatial distribution, and relationships to AEP components. Voice onset time was reflected in MUA time-locked to consonant release and voicing onset, and phase-locked to the syllables' fundamental frequency. Place of articulation was reflected in discriminative ‘on’ and phase-locked responses occuring to the formant transitions of the syllables. Duration of the voiced formant transitions was represented by an accentuation of the phase-locked responses occurring to this period. Activity of thalamocortical fibers and cortical cells differed. Thalamocortical fibers were more responsive to speech sounds, as well as responding more frequently with a phase-locked response pattern. Cortical cells responded with sustained activity to a greater degree. Responses to identical portions of the vowels were biased by the preceding consonant. The spatial extent and timing of the responses demonstrate that speech sounds are processed along parallel, but not synchronous, channels. Relevance to human psychoacoustical phenomena is discussed.