Search

Compensatory shortening has been reported to occur with increased utterance complexity; for example, durations of segments decrease as the number of syllables in a phrase increases. But compensatory shortening is not always found, and it has been suggested that its occurrence is related to differences in the nature of the speech tasks required of a speaker, for example, producing citation-form speech and oral reading [T. H. Crystal and A. S. House, J. Acoust. Soc. Am. 88, 101-112 (1990)], and may also vary with the language studied. The origin of this timing pattern (or its absence) is of particular interest in understanding speech production, as a reflection of planning unit size; if its presence varies with speech task, so must the planning unit. This paper reports results of a study of several speech tasks (list reading, repeating a recorded exemplar, paragraph reading) on the extent of compensatory shortening. [Work supported by St. John's University and NIH DC-00121 to Haskins Laboratories.] [ABSTRACT FROM AUTHOR]

Each of two methods was used to determine the onset time of EMG activity associated with speech gestures in two data sets. Each data set contained between 10 and 20 tokens (repetitions) of several utterances. Onset times were specified in relation to a reference point in the speech signal. In one method, measurements were made from the digitized waveforms for the individual tokens. Average onset time for each utterance type was calculated by averaging measurements from its tokens. In the other method, the ensemble average of the EMG signals from tokens of each utterance type was calculated after carefully aligning them at the reference point. Onset time was then measured for each utterance type from the ensemble-average waveform. Onset-time measurements from ensemble averages were consistently earlier than corresponding averages of individual-token measurements. For some utterances, the ensemble-average onset time was earlier than any individual-token measurement. These results confirm the value of ensemble averaging for reducing the inherent noise in EMG signals associated with speech gestures. [Work supported by NINCDS and BRSG.] [ABSTRACT FROM AUTHOR]

The extent of coarticulation is of central interest to theories of speech production. Knowing the anticipatory and carryover limits of coarticulation will allow us to specify the constant features of a set of speech sounds. Most earlier work has reported on observations of the articulator movement or acoustic result of coarticulation [e.g., B. E. F. Lindblom, J. Acoust. Soc. Am. 35, 1773-1781 (1963); S. E. G. Öhman, J. Acoust. Soc. Am. 39, 151-168 (1966); P. F. MacNeilage and J. L. DeClerk, J. Acoust. Soc. Am. 45, 1217-1233 (1969); R. Daniloff and K. Moll, J. Speech Hear. Res. 11, 707-721 (1968); T. J. Gay, J. Phonetics 2, 255-266 (1974)], The present study is an attempt to define the extent and effects of coarticulation on EMG activity recorded from the tongue and lips for VCV and CVC utterances, and to correlate this activity with the resulting acoustic signal. Preliminary results suggest that carryover effects are more extensive than anticipatory effects. [Supported by NIDR Grant DE-01774.] [ABSTRACT FROM AUTHOR]

Considerable evidence attests to the phenomenon of F0 declination during the course of an utterance. Recently, the question of whether upper articulators might show declination-type effects has been raised for the jaw [E. V.-Bateson and C. A. Fowler, J. Acoust. Soc. Am. Suppl. 1 84, S128 (1988)]. In previous studies, lower velar peaks were noted in sentence-final than -initial syllables, prompting a systematic investigation of velar declination. Thus peak velar height for early and late syllables in natural and reiterant-speech sentences of from three to nine syllables has been examined. Velar position and acoustic data were collected from three speakers of American English. Differences in velar height were examined with respect to number of syllables and acoustic duration. Possible interactions between declination-type processes and factors such as stress were also studied, since stress has been shown to have a strong influence on velar movement [R. A. Krakow, J. Acoust. Soc. Am. Suppl. 1 82, S17 (1987); J. Vassiere, Phonetica 45, 122-139 (1988)]. [Work supported by NIH Grants DC-00121 and RR-05596 to Haskins Labs.] [ABSTRACT FROM AUTHOR]

One hypothesis about the temporal control system of speech postulates two levels of control, a motor planning level and a motor execution level [J.-F. Bonnot, J. d'Acoust. 2, 287-296 (1989)]. This study of four ataxic dysarthric and four unimpaired speakers was designed to explore the role of cerebellar function in speech and identify some of the linguistically and physiologically determined aspects of speech timing. Preliminary results reveal a systematic but nonuniform timing disruption: a tendency for compensatory shortening effects to be preserved, even in the presence of segmental lengthening by the dysarthric speakers, and an absence of terminal lengthening effects. These results suggest that compensatory shortening is planned and sufficiently robust to survive the temporal disruption at the motor execution level, but that terminal lengthening arises from the execution of a motor sequence, rather than being part of the motor plan. [Work supported by Ministère de Recherche et Technologie, France, St. John's Univ. William Paterson College, and NIH Grants DD-00121 and RR-05596 to Haskins Labs.] [ABSTRACT FROM AUTHOR]

Studies of coarticulation have provided evidence of lip rounding activity during consonants immediately preceding a rounded vowel, although reports differ as to the extent of this anticipatory activity. One would naturally expect to find this rounding reflected in the acoustic spectra of the consonants, and to have perceptual salience. To establish the extent of such acoustic and perceptual effects, we conducted studies using two speakers, one who uses lip gestures for alveolar consonants in both rounded and unrounded vowel environments, and another who uses them only in rounded vowel environments. In the first study, spectral analyses of 520 V1CnV2 utterances from two speakers yielded little evidence of V2 rounding gestures effecting an unfounded V1, while there is evidence that such lip rounding gestures influence fricative noise immediately preceding a rounded or unrounded V2. The results indicate that the acoustic effect is not perceptually salient for a speaker for whom lip rounding is inherent in alveolar consonant production. [This work was supported by grant NS-13617 and RR-05596.] [ABSTRACT FROM AUTHOR]

We have used an articulatory synthesizer to study the effects of independent and systematic changes in velar position and velar port size on vocal tract transfer functions of vowels. We have also examined the acoustic and perceptual effects of combining these components of natural nasalizing gestures. Lowering the velum, in the absence of nasal coupling, has small effects, primarily on F2 and F3 frequencies. On the other hand, opening the port without lowering the velum has the greatest influence on F1 frequency, shifting it upward or downward, as necessary, to produce a concentration of spectral energy at about 400 Hz. As expected, the combined effects of lowering the velum and opening the velar port are more heavily influenced by nasal coupling than by changes in velar position. We have also found differences in the degree of perceived nasality associated with velar port areas from 0 to 84 mm2, as a function of the oral vowel (0 mm2) quality. [Work supported by NINCDS.] [ABSTRACT FROM AUTHOR]

A central preoccupation of theorists since the beginning of the modern explosion in speech production research has been formulating the underlying rules which allow speakers to concatenate message units and overall temporal properties of utterances (Ohman, 1966, 1967; Lindblom, 1963). We argue that, in general, when stress and speaking rate was varied, relational invariance is preserved over substantial changes in the absolute duration of articulatory events, and that rules which preserve temporal structure can be used to describe coarticulatory changes. [This work was supported by NIH Grant NS-13870 to Haskins Laboratories.] [ABSTRACT FROM AUTHOR]

The interposition of a constant between vowels has been shown to affect systematically the formant frequencies of those vowels. The nature of these effects appears to depend on the extent to which the vowels and consonants share the tongue and lips as articulators. Traditional articulatory descriptions suggest two alternatives concerning the status of [h] and of [?] as consonants. On the one hand, since they are classified as consonants, we would expect them to produce the same coarticulatory effects as other fricatives and stops. On the other hand, since they are described as nonlingual/labial in their articulation, we would expect them to produce minimal coarticulatory effects. Indeed, results of some recent studies of speech production suggest that neither [h] nor [?] interrupts V-to-V articulations in the way that other consonants do (Harris and Bell-Berti, 1982). The present study employed spectrographic analysis of VV, VhV, V?V, and VCV sequences. Results revealed that there was minimal coarticulation between [h] or [?] and the surrounding vowels. Indeed, the formant structure of the vowels in VhV and V?V sequences were more similar to those of VV sequences than to those of other VCV sequences. Since consonants are often portrayed as the units that delimit syllables, these results raise the question of functional homogeneity within the class of consonants. [Research supported by NINCDS.] [ABSTRACT FROM AUTHOR]

Attempts to model speech production processes based on abstract static units, such as syllables, phonemes, or features, fail to account for a number of observations of natural speech. For example. a phonetic feature such as lip rounding has been described as being anticipated varying numbers of consonant segments before the vowel for which it is specified. Such data conform to predictions made with models assuming a CnV syllable structure. Such models, however, fail to predict nasality anticipated in vowels preceding nasal consonants. On the other hand, feature-based models that apparently account for observations of anticipated nasality fail to predict other coarticulatory phenomena, such as the suppression of lip-rounding or tongue-fronting activity during the production of an intervening consonant presumably containing no conflicting articulatory gesture. A model assuming that speech is specified segmentally with fixed timing relationships between the end and beginning of successive segments (and variable durations of dynamic gestures) can eliminate these, and other, discrepancies between predicted and observed behavior. [Work supported by NINCDS and BRSG grants.] [ABSTRACT FROM AUTHOR]

Recent studies have revealed that although durations of sentences produced by a group of ataxic dysarthric subjects were significantly longer than those of a control group (sometimes twice as long), and that durations of segments occurring early in sentences were significantly longer in the speech of the dysarthric subjects, there was an absence of final lengthening [Bell-Berti et al., J. Acoust. Soc. Am. Suppl. 1 88, S128 (1990)]. One explanation offered for the lack of final lengthening is that the speech was already so slow (and the individual segments so long) that further lengthening was not possible. The present study of three normal speakers was undertaken to examine the possibility that the magnitude of the final lengthening is related to speaking rate and the extent of final lengthening. Indeed, at very slow rates, utterance-final segments may even be shorter than they would in sentence-medial positions. [Work supported by St. John's University and NINDCD Grant DC-00121 to Haskins Laboratories.] [ABSTRACT FROM AUTHOR]

Recently, Bateson and Fowler [E. V.-Bateson and C. A. Fowler, J. Acoust. Soc. Am. Suppl. 1 84, S128 (1988)] and Bell-Berti and Krakow [F. Bell-Berti and R. A. Krakow, J. Acoust. Soc. Am. 89, 1916 (A) (1991)] have shown that declination effects are not restricted to laryngeal-respiratory behavior. Bateson and Fowler observed a decline in the extent, peak velocity, and duration of jaw opening and closing gestures, and, in this paper, a decline in peak velar height over the course of a sentence was observed. This study reports evidence of declination in another articulator, the lower lip. Consistent with previous findings for the velum, preliminary results show a decline in the height of the lip when comparing the first to the last peak in each sentence. Like the velum, the lower lip shows a decline within bisyllabic words, and across adjacent and nonadjacent bisyllabic words. Here, the co- occurring patterns of declination in these two articulatory subsystems are examined and how these patterns interact with stress is shown. [Work supported by NIDCD Grant No. DC-00121.] [ABSTRACT FROM AUTHOR]

Velar position for a phonetic segment is determined by both the nature of the segment and the phonetic environment in which it occurs. Thus, velar position for consonants is higher in a close-vowel than an open-vowel environment. This observation led to an experiment designed to provide data for quantifying the temporal extent of carryover and anticipatory influences of vowels on consonants. Velar position was determined from frame-by-frame analysis of motion pictures of the velum of a native speaker of English. The 27 experimental utterances, repeated from 4 to 7 times each. were disyllables having either /aCni/ or /iCna/ patterns, with Cn containing various combinations of /t/ and /s/. Consonant-string durations were measured from digitized waveforms of the speech signal. Preliminary analysis of the data reveals that the velum is higher at the temporal mid-point of a consonant string for utterances whose second vowel is /i/, suggesting that anticipatory influences predominate at that time. [Work supported by NINCDS and BRSG.] [ABSTRACT FROM AUTHOR]

EMG data from studies of lip rounding and jaw lowering activity reveal that, for a given speaking rate and stress pattern, the onset of EMG activity associated with these gestures occurs at a time before vowel onset which is unaffected by the acoustic duration or the number of consonants in the preceding string. The apparent contradiction between these conclusions and those of previous studies of anticipatory coarticulation may arise at least in part from neglect of the effects of clustering on consonant duration. [Supported by NINCDS grants to the Haskins Laboratories.] [ABSTRACT FROM AUTHOR]

The position of the velum for any phonetic segment is determined by at least two factors: the nature of the segment itself and the phonetic environment in which the segment occurs. Given a constant phonetic environment, velar elevation increases, and velopharyngeal port size decreases, through the series of segment types: nasals-open vowels-close vowels-obstruents. It is commonly known that velar elevation for vowels, in English, is affected by adjacent nasals. More general interactions between vowels and consonants both nasal and obstruent, have not been described. Velar elevation was measured for utterances including nasal and obstruent consonants and either close or open vowels. Generally, velar elevation varied as predicted above. In addition, vowel quality was found to affect velar position during adjacent consonants: The velum was lower for both nasals and obstruents in an environment of open vowels than in an environment of close vowels. [This work was supported by NINCDS Grants NS-13617 and NS Fellowship-05332, and NIDR Grant DE-01774.] [ABSTRACT FROM AUTHOR]

EMG studies of the American English vowel pairs /i-[small_capital_eye]/ and /e-[variant_greek_epsilon]/ reveal two different production strategies: some speakers appear to differentiate the members of each pair primarily on the basis of tongue height; for others the basis of differentiation appears to be tongue tension. To determine if these production differences correspond to perception differences, two vowel identification tests were given to the EMG subjects. Subjects were asked to label the members of a seven-step vowel continuum, /i/ through /[small_capital_eye]/. In one test each item had an equal probability of occurrence. The other was an anchoring test: the first stimulus, /i/, was heard four times as often as any other. Compared with the equal-probability test labelling boundary, the boundary in the unequal-probability test was displaced toward the anchor stimulus. However, the magnitude of the shift was greater for subjects using a production strategy based on tongue height than for subjects using tongue tension to differentiate the vowels. [This work was supported by NIH Grants NS-13617, NS-05332, HD-01994, and RR-05596 to Haskins Laboratories; and SUNY Research Foundation.] [ABSTRACT FROM AUTHOR]

The anticipation of articulatory features, in particular, lip rounding in anticipation of a rounded vowel, has been reported to occur as many as four segments before the segment for which the feature is specified. Recently we reported [F. Bell-Berti and K. S. Harris, J. Acoust. Soc. Am. 60, S63(A) (1976)] that the motor commands for the rounding gesture for /u/ begin a fixed time before voicing onset for the vowel. This timing is unaffected by the number of preceding consonant segments or the location of syllable boundaries in the preceding string. Thus, the initiation of lip rounding appears to be linked to other features of the vowel articulation, much as movement of the velum has been shown to be linked to lingual articulation [R. D. Kent, P. J. Carney, and L. R. Severeid, J. Speech Hearing Res. 17, 470-488 (1974)]. Analysis of these and additional data lead us to postulate a system of motor organization which is segment rather than feature based. [Work supported by NINCDS grant NS 13870, NIDR grant DE 01774, and NINCDS postdoctoral fellowship 1 F32 NS 05332-01A1.] [ABSTRACT FROM AUTHOR]

The extent of both anticipatory and carryover coarticulation is of central importance to theories of speech organization. One articulatory gesture whose anticipation has been studied repeatedly is vowel lip rounding. Daniloff and Moll [J. Speech Hear. Res. 11, 707-721 (1968)], reporting the results of a cinéfluorographic study of lip rounding, describe vowel lip rounding as being initiated with the beginning of a preceding consonant string, for strings up to four segments long. Preliminary results of an EMG experiment in which /u/ was preceded by a one- or two-consonant string show activity for the rounding gesture beginning at the same time before the vowel onset for all conditions regardless of the position of the syllable boundary. Parallel results were obtained for carryover coarticulation. [ABSTRACT FROM AUTHOR]

A proposed model of velopharyngeal function in speech, based on electromyngraphic (EMG) data from velopharyngeal musculature for open and close vowels, oral stops, and nasal stops, states that the levator palatini is the primary effect or of velopharyngeal port closure and that its activity for oral segments varies directly with oral cavity impedance [F. Bell-Berti, Ph.D. thesis, City University of New York (1973)]. An EMG study of the velopharyngeal musculature was conducted, to test the model. The experimental utterances included oral fricative sod frictionless continuant consonants in addition to the previously studied oral and nasal stop consonants. The model predicts that levator palatini EMG activity will be greatest for oral stop consonants, and will decrease through the series fricative consonants, frictionless consonants, high vowels, low vowels, sod that it will be lowest for nasal consonants, to facilitate nasal coupling. [Supported by NIDR grant DE 01774.] [ABSTRACT FROM AUTHOR]

It has been noted that the height of the velum for oral speech segments varies with the degree of oral cavity constriction for these segments [e.g., J.F. Lubker, Cleft Palate J. 5, 1-18 (1968); F. Bell-Berti, Ph.D. thesis, City University of New York (1973)]. A cinéfiberscopic study was undertaken to describe the velar height relationships between oral stop, fricative, and frictionless consonants, and the effects of phonetic environment on the achieved velar height for each segment. Frame-by-frame measurements of velar height were made from motion pictures made of the nasal surface of the velum of one speaker of American English, speaking a set of nonsence disyllables containing oral stop, fricative, frictionless, and nasal stop consonants in a constant vowel environment. Both voiced and voiceless oral stop and fricative consonants were used, as well as a format that included both oral-consonant-nasal-consonant and nasal-consonant-oral-consonant contrasts. The data are examined and discussed with reference to earlier velar height and electromyographic data. [Supported by NIDR grant DE 01774.] [ABSTRACT FROM AUTHOR]

Feature spreading and coproduction models make fundamentally different assumptions about the nature and organization of speech motor control, and yet each model is supported by some, but not all, of the existing empirical data. This has led some researchers to conclude that speakers probably use alternative strategies at different times. This study suggests that the identification of coarticulatory influences requires the concurrent identification of intrinsic articulatory characteristics of the segment. Moreover, the evidence for feature spreading or variable coarticulation strategies derives from the misidentification of such intrinsic characteristics as context effects. This velar coarticulation study used a controlled comparison between CVnN and CVnC minimal pairs, where C is an oral consonant, Vn is any number of vowels, and N is a nasal consonant. Vocalic string duration was manipulated by varying the number of segments and speech rate, allowing us to alter the time between the onsets of vocalic and subsequent consonantal gestures. Velar lowering occurred in CVn sequences, whether or not a nasal consonant followed, and similar vocalic gestures were observed across minimally contrastive environments with and without the nasal consonant. Moreover, velar lowering for the nasal consonant began in close temporal proximity to the nasal murmur. These results strongly support the coproduction model and provide insight into previously conflicting reports.

The anticipation of articulatory features, in particular lip rounding in anticipation of a rounded vowel, has been reported to occur as many as four segments before the segment for which the feature is specified. In the data presented here, we find that the moter commands for the rounding gesture for /u/ begin a fixed time before the onset of the vowel. This timing is unaffected by the number of consonant segments in the preceding string. Thus, the initiation of lip rounding appears to be linked to other features of the vowel articulation.

According to some theories, anticipatory coarticulation occurs when phones for which a feature is unspecified precede one for which the feature is specified, with consequent migration of the feature value to the antecedent phones. Carryover coarticulation, on the other hand, is often attributed to "articulatory sluggishness." In this paper, EMG evidence is provided that this formulation is inadequate, since the beginning of EMG activity associated with vowel lip rounding is independent of measures of the acoustic duration of adjacent lingual consonants. We suggest that the vowel-rounding gesture simply co-occurs during predictable intervals with portions of preceding and following lingual consonant articulations.

The purpose of this letter is to explore some reasons for what appear to be conflicting reports regarding the nature and extent of anticipatory coarticulation, in general, and anticipatory lip rounding, in particular. Analyses of labial electromyographic and kinematic data using a minimal-pair paradigm allowed for the differentiation of consonantal and vocalic effects, supporting a frame versus a feature-spreading model of coarticulation. It is believed that the apparent conflicts of previous studies of anticipatory coarticulation might be resolved if experimental design made more use of contrastive minimal pairs and relied less on assumptions about feature specifications of phones.