INTRAORAL AIR PRESSURE AS A F E E D B A C K CUE IN CONSONANT PRODUCTION ROBERT A. PROSEK
Walter Reed Army Medical Center, Washington, D. C. ARTHUR S. HOUSE
Institute for Defense Analyses, Princeton, New Jersey The effects of oral-sensory deprivation on the production of consonants was studied using narrow phonetic transcriptions and measurements of intraoral air pressure and duration. The speech materials were 20 bisyllabic words produced both in isolation and in sentences, and sentences that included words with 34 stop variants. These utterances were produced by four normal talkers and by the same talkers when deprived of oral sensation. The state of oral-sensory deprivation was induced by a series of mandibular, infraorbital, and palatal injections of 2~ xylocaine. The talkers also scaled levels of effort used to produce the syllables /pa/ a n d / h a / , both with and without the anesthetic. In general, after the administration of the anesthetic, the characteristic tongue carriage of the talkers was shifted posteriorly, the rate of speech was slower, and there were minor imprecisions in articulation consisting primarily of alterations in lip and tongue activity. In addition, consonants were produced with slightly greater intraoral air pressures and longer durations. The talkers had no difficulty in scaling levels of effort in either the normal or the anesthetic conditions, and maintained a linear relationship between effort and intraoral air pressure in both conditions. The results suggest that the talkers used more effort in producing speech in the anesthetic condition and are untenable with the idea that intraoral air pressure constitutes an important feedback parameter in controlling articulation. The speech produced by the talkers while anesthetized (that is, while without sensation in the mouth) was reasonably precise and must be postulated to have been under the control of a pressure-sensing system other than a closed feedback loop. Recent interest in oral sensation and perception has led to the development of models of speech production that specify certain oral-sensory mechanisms that m a y operate to control articulation. MacNeilage (1970), for example, has hypothesized that the motor system for speech includes two major control components; an open-loop component provides phonological information, and a closed-loop component provides moment-to-moment information from the peripheral speech mechanism. MacNeilage has suggested that the closed-loop information is supplied by the gamma-efferent system that provides information on instantaneous muscle length and the rate of change of muscle length. This information is used to achieve control of articulator movement that is dependent on the state of the muscle. Other authors, however, have proposed alternative oral-sensory mechanisms 133
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
that may provide the closed-loop information. Perkell (1969) has suggested that, during consonant production, intraoral air pressure is sensed in the oral cavity and monitored with myotactic feedback. In particular, he has proposed three modes of intraoral air pressure increase that provide information relative to three consonant classes: high intraoral air pressures are associated with voiceless fortis consonants; intermediate pressures are associated with voiced lenis consonants; and low intraoral pressures are associated with sonorants. A number of studies of the aerodynamic properties of consonant production have provided data that are compatible with Perkell's model. These studies have attempted to answer two questions about intraoral air pressure during consonant production. First, can consonants be distinguished on the basis of differences in intraoral air pressure? Second, are talkers aware of changes in intraoral air pressure? Several investigations have provided data that answer the first question (Subtelny, Worth, and Sakuda, 1966; Arkebauer, Hixon, and Hardy, 1967; Brown and McGlone, 1969; Netsell, 1969; Brown et al., 1970; Malecot, 1955, 1966, 1968, 1970). These studies show that voiceless fortis consonants are produced with higher peak intraoral air pressures than voiced lenis consonants; for most subjects, the intraoral air pressure developed during the repetition of words and sentences is reliably produced with little variation; and consonants occurring before a stressed vowel have greater peak intraoral air pressures than the same consonants occurring after a stressed vowel, or before and after an unstressed vowel. In fact, Malecot (1970) has suggested that intraoral air pressure is the feedback cue used by talkers to control the production of fortis and lenis consonants, since the differences in their intraoral air pressures are reliable and occur in any context. On the other hand, Lisker (1970) has claimed that consonants are not adequately distinguished on the basis of intraoral air pressure. Lisker's data may be said to demonstrate that intraoral air pressure cannot be the sole feedback cue for consonant production. There is little experimental evidence to answer the second question concerning sensitivity to changes in intraoral air pressure. Malecot (1966) reported that the difference limen for intraoral air pressure is 1 cm H20, or less, and that the difference in peak intraoral air pressure between voiceless fortis consonants and voiced lenis consonants was greater than 2 cm H20. MaIecot has interpreted these findings as evidence that intraoral air pressure can be monitored during speech production. Another bit of evidence has been supplied by Ringel, House, and Montgomery (1967), who reported that the effort used to produce initial consonants /p/ and /b/ was related to the intraoral air pressure by a power function with a slope of approximately one. This result demonstrated that when a talker is asked to change his effort, the intraoral air pressure also changes. It is possible that one of the cues used by the talkers to change their effort was intraoral air pressure. In summary, there is some evidence that intraoral air pressure may be
134 Journal o[ Speech and Hearing Research
18 133-147 1975
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
monitored directly to control consonant production. Differences in peak intraoral air pressure and in consonant duration have been reported between voiceless fortis and voiced lenis consonants in all contexts in which the opposition occurs. Normal talkers are sensitive to intraoral air pressure differences that are less than the differences observed between fortis and lenis consonants in speech. Furthermore, changes in intraoral air pressure are related linearly to changes in effort. The purpose of the present study was to determine the changes in intraoral air pressure and consonant duration that occur when talkers are deprived of oral sensation. Comparisons between the normal (that is, control) and the sensory-deprived (that is, anesthetized) conditions were made for consonants produced in isolated bisyllabic words and in sentences. In addition, the subjects estimated the magnitude of the effort used to produce the consonants/p/ and /b/. This task provided information on the perceptual changes in effort brought about by anesthesia. Finally, all utterances were transcribed to describe the articulatory changes brought about by eliminating sensation in the oral structures. METHOD Speech Material
Twenty bisyllabic words were used in the experiment. Half the words were verbs in which the second syllable received the stress, and half were nouns that differed from the verbs only in stress pattern. The words provided six stops and four fricatives as the initial phonemes of the second syllables. The list of words and the phonemes studied in each word are shown in Appendix A. To examine the effects of oral-sensory deprivation on the production of consonants in utterances that are longer than two syllables, a list of sentences, containing the words described above, was prepared. Subtelny et al. (1966), Malecot (1970), and Brown et al. (1970) have reported that differences in intraoral air pressure between voiceless fortis and voiced lenis consonants are observable in sentences, but the absolute values may not be as great as those observed in isolated words. The sentences used in the present study are listed in Appendix B. In addition to the sentences in Appendix B, a list of 13 sentences was prepared to provide a wide variety of allophonic variants of the stop consonants under study. These variants represent articulations that differ from simple stops in their closure, release, and voicing characteristics, but which occur frequently in connected speech. These sentences have been included to provide data on the effects of oral-sensory deprivation on more complicated articulations. The types of stop variants examined and the word in which the variant appears are delineated in Appendix C; the number in parentheses following each word refers to the sentence number in which the word appears. The words are shown embedded in sentence contexts in Appendix D.
PrtosEK, HousE: Intraoral Air Pressure 135
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
Experimental Conditions and Subjects The speech materials were produced in a control and an experimental mode by four young adult talkers. Each talker had normal speech and hearing, normal oral structures, and no history of sensory or motor disturbances. In the control condition (normal speech), the talkers were able to use all their usual sources of oral-sensory information. In the experimental condition, the oral structures were anesthetized to eliminate sensation. In both conditions the talkers were able to hear themselves normally. The state of oral-sensory deprivation (the anesthetic condition) was achieved by a series of nerve block injections of 2~ xylocaine. The injection series consisted of (1) bilateral mandibular injections that eliminated sensation in the lower lip and teeth, the anterior two-thirds of the tongue, the skin of the cheek, and the mucous membrane of the mouth and lower gingiva; (2) bilateral infraorbital injections eliminating sensation in the upper lip and teeth; (3) bilateral posterior palatine iniections that eliminated sensation in the mucous membrane of the velum and uvula; and (4) medial nasopalatine injections that eliminated sensation in the mucous membrane of the alveolar ridge and the anterior portion of the hard palate. The depth of the anesthetic was evaluated by use of a probe. All subjects were insensitive to touch and pressure in all anesthetized areas. The effectiveness of the anesthesia was further evaluated by asking the talkers to discriminate between geometric forms placed successively in the mouth. When the anesthesia was effective, the talkers were unable to discriminate between a square and an oval. The injection series, then, produced a profound loss of oral sensation.
Equipment A 12-French catheter connected to a pressure transducer was used to sense intraoral air pressure. The output of the transducer was amplified and recorded on one channel of an optical oscillograph. Static calibration of the pressure system was accomplished by a U-tube water manometer where known amounts of pressure were equated with the galvanometer deflection of the oscillograph. The insertion gain of the pressure system was determined by sealing the tip of the pressure transducer in a 2-cc coupler. Discrete frequencies from 10 to 350 Hz, in increments of 10 Hz, were introduced by means of a small transducer sealed in the top of the coupler. A condenser microphone, connected to a sound-level meter, was sealed in the bottom of the coupler. The sound pressure level inside the coupler was adjusted to 100 dB at each frequency, and the response was recorded on the oscillograph and measured in millimeters of H20. The response of the transducer was fiat from 10 to 350 Hz, and this response, as read from the oscillograph, was used as the reference in computing the relative amplitudes, in decibels, for the entire pressure system. The procedure was repeated with the catheter inserted between the coupler and the pressure transducer. The frequency response of the entire pressure system was quite 136 1ournal of Speech and Hearing Research
18 133-147 1975
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
fiat except for a resonance at about 100 Hz. This resonance did not interfere with the pressure measurements, since only low-frequency changes in pressure were of interest in the present study, that is, those below 50 Hz. During the actual collection of the speech samples, two high-quality microphones were placed in front of the talker at a distance of 12 inches. One microphone was used to record the speech waveform on a second channel of the oscillograph for comparison with the pressure tracings. The second microphone was used to record the words and sentences on a high-quality magnetic tape recorder. Procedures
In both talking conditions, the catheter was inserted through the nasal cavity of the talker until the experimenter observed the tip in the oropharynx. Each talker was given the opportunity to practice the words and sentences with the catheter in place. In the control condition, all materials were read in a normal conversational mode, and only utterances appropriate to Standard American English were accepted by the experimenter. For both conditions, the talker sat erect facing the microphones and first read the words in Appendix A. At the completion of this task, the talker read the materials in Appendix B and D as declarative sentences. In anticipation of possible series effects, a different randomized arrangement of the nouns, verbs, and sentences was used for each talker in both the control and anesthetic conditions. To counterbalance for possible sequential effects, two of the talkers read the list of nouns before the list of verbs, while the other two talkers read the lists in the reverse order. After completing the recordings of the words and sentences, each talker estimated the magnitude of the effort used in producing the s y l l a b l e s / p a / a n d /ha/. Each talker practiced producing a syllable at a comfortable, conversational level, after which he monitored this standard production visually on an oscilloscope (connected in parallel with the amplifier). The effort involved in this level of production was assigned a numerical value of 10. The talker then produced the same syllable at a level of effort corresponding to the magnitude of a number supplied by the experimenter; the numerical values of 5, 10, 20, and 30 were used in an incompletely counterbalanced arrangement. For each numerical magnitude, the talker uttered a syllable three times in sequence; the median of these three productions was used in the analysis. After each series of four productions, the talker practiced his (conversational) standard three times, monitoring his productions visually. To control for possible sequential effects, two of the talkers first estimated the effort used to produce the syllable/pa/, then the effort used to produce/bo/; the other two talkers scaled /ba/ before /pal. Peak intraoral air pressure was defined as the maximum pressure developed during the production of the consonant. For voiceless consonants, the maximum was easily observed. However, for voiced consonants, pressure variations due
PROSEK, HOUSE: Intraoral Air Pressure 137
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
to the vibrations of the vocal folds were superimposed on the low-frequency pressure changes. The midpoints of the variations were determined, and the peak was taken as the midpoint showing the greatest displacement. Consonant duration was determined by examining the waveform and pressure traces and selecting a point in time where articulatory changes could be observed. The articulatory changes observed for consonant onset were increases in pressure, offset of voicing (for voiceless consonants), and the reduction of waveform amplitude. The changes observed for consonant termination were a reduction in pressure to a level consistent with the following sound, the onset of voicing (for voiceless consonants), and an increase in waveform amplitude. RESULTS
Phonetic Transcriptions The speech changes described in this section of the paper are based on standard phonetic transcriptions of the speech materials. Due to space limitations, only the general findings of the phonetic analysis will be reported. (For more detailed analysis, see Prosek, 1971. ) The general intelligibility of the talkers was not affected by the loss of oral sensation. No formal intelligibility tests were administered, but under ordinary listening conditions there were no instances in which the speech of any talker could not be understood. This evaluation is based on the reports of four listeners. In particular, there were no gross changes in the intonation patterns used in the sentences, and the talkers did not change the assignment of syllabic stress. While all the talkers were insensitive to touch and pressure in the areas anesthetized and showed poor oral discrimination of geometric forms, the effect of the anesthetic on general intelligibility was not the same for all talkers. The speech of two of the talkers was better than the speech of the other two, when the number of articulatory distortions was the criterion. In general, an anesthetized talker modified his characteristic tongue carriage, talked slower, and changed the articulation of consonants in minor, essentially nonphonemic, ways. The tongue carriage was more posterior in the anesthetic condition than in the control condition. The result of this change was that primary closures and constrictions were made further back in the oral cavity. This change in tongue position, which was inferred independently from the transcribed material, was substantiated, in part at least, by the comments volunteered by the anesthetized talkers that their tongues seemed to be laying in the backs of their mouths. A slower rate of speech was observed for the two talkers whose speech was most affected by the anesthetic. This auditory observation was substantiated by measuring the duration of selected sentences in the control and anesthetic conditions. The durations of these sentences in the control condition were 2.5 to 3.5 seconds; in the anesthetic condition, the durations were 1.0 to 1.5 seconds longer than in the control condition. 138 Journal o[ Speech and Hearing Research
18 133-147 1975
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
The articulatory changes usually involved only one or two phonetic features. The primary articulatory effects were alterations in labial activity and the absence of fine adjustments of the apex and blade of the tongue. These findings agree well with the results reported in detail by Scott and Ringel (1971). For all talkers, the production of vowels in the anesthetic condition was very similar to vowel production in the control condition. When changes did occur in the anesthetic condition, they usually were of the following types: high back vowels were produced without expected lip rounding, r-colored syllabic nuclei were produced without expected retroflexion, and vowels were nasalized excessively. Intraoral Air Pressure
The measurements reported were gathered on consonants produced in word pairs differing in syllabic stress placement (see Appendixes A and B). Measurements of peak intraoral air pressure wer e averaged over subjects and categories such as voicing, manner of production, syllabic stress, context, and place of articulation. The intraoral air pressures measured for the consonants in the control condition are in substantial agreement with those of previous studies (Malecot, 1955; Subtelny et al., 1966; Arkebauer et al., 1967; Lisker, 1970). Consonants produced in the anesthetic condition were characterized by greater intraoral air pressure than consonants produced in the control condition. Everything else being equal, however, the difference in peak pressure between the two conditions did not exceed 1 cm H20. Since Malecot (1966) reported a difference ]imen for intraoral air pressure of 1 cm H20, the difference between the two conditions will not be considered significant. The data for individual subjects sometimes showed a pressure increase or decrease between the two conditions of from 1.5 to 2 cm H~O, but these instances were rare. The general findings are summarized in Table 1. The tabulated values show that, in agreement with all previous investigations, voiceless consonants were produced with greater peak pressures than voiced consonants. In agreement with Arkebauer et al. (1967), voiceless stops were produced with greater pressures than voiceless fricatives, and voiced fricatives were produced with greater pressures than voiced stops. The intraoral air pressure developed for consonants before stressed vowels was not greatly different from that for consonants before unstressed vowels. This finding agrees with the results of Lisker (1970), who reported only small changes in peak pressure as a function of syllabic stress. Malecot (1968), however, had reported that consonants before stressed vowels were produced with greater peak intraoral air pressure than consonants in any other position. In the control condition, consonants in isolated words showed slightly greater peak intraoral air pressures than consonants in the sentence contexts; in the anesthetic condition, however, no difference was evident. This is not significant since the difference in the control condition is only 6 mm HzO. There is a PROSEK, HOUSE:Intraoral Air Pressure 139
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
TAnhE 1. Summary of the mean intraoral air pressures (in centimeters of H20) for various consonants produced in the control and anesthetic conditions grouped into categories, as indicated. The consonants are from the words listed in Appendixes A and B. All entries in the difference column represent greater pressure in the anesthetic condition. Consonants Voicing and manner ptk fs vz bdg Syllabic stress stressed unstressed Context words sentences Place of articulation pbfv t d s z
kg
Control
Anesthetic
Difference
6.4 5.9 3.8 3.1
6.7 5.9 4.8 3.6
0.3 0.0 1.0 0.5
4.9 4.7
5.3 5.1
0.4 0.4
5.1 4.5
5.2 5.9.
0.1 0.7
4.5 4.9 5.0
4.6 5.5 5.9
0.1 0.6
0.9
tendency for consonants produced in the back of the oral cavity to have greater peak pressures than consonants produced anteriorly, but once again, no particular significance can be attached to this finding. In summary, the intraoral air pressure developed for consonants produced in the anesthetic condition did not differ greatly from those in the control condition. In both experimental conditions voiceless stops were produced with greater intraoral air pressure than voiceless fricatives, and voiced fricatives were produced with greater intraoral air pressure than voiced stops. Syllabic stress placement, type of context, and place of articulation were not distinguishable on the basis of intraoral air pressure. Duration The measurements of consonantal duration were obtained from the materials listed in Appendixes A and B. The durations were averaged and compared in the same manner as were the peak intraoral air pressures. The results are summarized in Table 2. The durations in the control condition agree well with the values reported by Malecot (1968, 1970) and with the spectrographic measurements reported by Lisker (1957). In the present and previous investigations, voiceless stops were found to have the longest durations and voiced stops were shortest. The average durations of consonants produced by the anesthetized talkers were 10 to 20 msec longer than those produced without sensory deprivation. The data for individual subjects showed increased durations in the anesthetic condition of up to 45 msec in rare instances.
140 Journal of Speech and Hearing Research
18 133-147 1975
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
TABLE 2. Summary of the average durations (in milliseconds ) for various consonants produced in the control and anesthetic conditions, grouped into categories as indicated. The consonants are from the words listed in Appendixes A and B. All entries in the difference column represent longer duration in the anesthetic condition. Consonants
Voicing and manner ptk fs vz bdg Syllabic stress stressed unstressed Context words sentences Place of articulation pbfv tdsz kg
Control
Anesthetic
Difference
150 143 112 93
159 149 123 115
9 6 11 22
132 116
144 131
12 15
129 119
143 131
14 12
122 123 131
134 133 154
12 10 23
In agreement with the findings of Malecot (1968), and independent of experimental condition, consonants before stressed vowels were longer than consonants before unstressed vowels. Similarly, consonants in isolated words were longer than consonants in sentence contexts. Finally, the durations of consonants produced in the back of the oral cavity were greater than those of consonants produced anteriorly or medially in the mouth. This difference is easily explained, however, since the anterior and medial categories include fricatives and stops, while the posterior productions are stops only. In summary, the durations of consonants produced in the anesthetic condition were greater than consonants produced in the control condition. Syllabic stress and context affected the duration of consonants, making them longer before a stressed vowel and in isolated words. Stop Variants"
The stop variants on which this discussion is based are those listed in Appendix C. In general, the voiceless stop variants were produced with lower peak intraoral air pressures and shorter durations in the control condition than the stops reported in the preceding section. The voiced stop variants, however, were produced with approximately the same intraoral air pressures as the more simple stops, but their durations were shorter. The average intraoral air pressures and durations for various classes of stop variants are summarized in Tables 3 and 4, respectively. Stops produced with a nasal release and a lateral release had peak pressures of 2.7 and 2.6 cm HzO, respectively. In addition, the durations of these stop
PROSEK, HOUSE: Intraoral Air Pressure 141
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
TABLE 3. Summary of the mean intraoral air pressures (in centimeters of H20) developed during the production of stop variants. See Appendix C for the word materials. All entries in the difference column represent greater pressure in the anesthetic condition.
Stop Variants
Control
Voiceless Voiced Lip rounded Dentalized Palatalized Voiceless lenis Intervocalic voiced flaps Nasal closure Lateral closure Nasal release Lateral release Unaspirated release Pseudo-aspirated release
5.6 3.3 5.3 5.6 3.8 6.9 4.0 4.4 3.9 2.7 2.6 6.0 6.4
Anesthetic Difference 6.6 4.3 6.4 6.4 5.6 7.9 4.7 5.6 5.4 3.8 2.7 6.4 7.3
1.0 1.0 1.1 0.8 1.8 1.0 0.7 1.2 1.5 1.1 0.1 0.4 0.9
TABLE 4. Summary of the average durations (in milliseconds) for the stop variants listed in Appendix C. All entries in the difference column represent longer durations in the anesthetic condition.
Stop Variants Voiceless Voiced Lip rounded Dentalized Palatalized Intervocalic voiced flaps Nasal closure Lateral closure Nasal release Lateral release Pseudo-aspirated release
Control 130 86 134 104 149 93 103 86 84 53 155
Anesthetic Difference 144 100 149 123 184 105 110 103 98 69 163
14 14 15 19 35 12 7 17 14 16 8
allophones are shorter than for most of the other variant classes. Thus, the release characteristics of the stops seem to affect their production more than place of articulation and voicing. A comparison of the intraoral air pressures and durations of the stop variants in the control and anesthetic conditions shows the same magnitude of difference as for the simple consonants. An exception is the duration of the palatalized alveolar stops, which is 35 msec greater in the anesthetic condition. Thus, the effects of oral-sensory deprivation are the same for the complex allophones as for the simpler stops.
Scaling of Effort In this task each talker produced three syllables in sequence for each numerical magnitude supplied to him by the experimenter, and the median of these three productions was used in the analysis. For each numerical magnitude, 142 lournal of Speech and Hearing Research
18 133-147 1975
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
the geometric mean of the medians was calculated, and a least-squares statistic was used to calculate the slope of the function relating effort to intraoral air pressure. When the apparent magnitude of the effort associated with speech production is plotted against an appropriate physical magnitude (such as intraoral air pressure) on log-log coordinates, the slope of the resulting function corresponds to the exponent of the power function governing the growth of the psychological magnitude. In Figure I the air pressure measurements have been 5o I I I J ~ I IJl~rr ,o-/~//b~/ / // / ;-- -~ONTROL / / /~///~, 30 9
a
EXPER.,
/o
__
-
I0
9
,?
FZCURE 1. Functions relating the effort used to produce/pa/and/ba/ to the intraoral air pressure in the control and anesthetic conditions.
,
ill ,I
I
I I 5
INTRAORALAIR PRESSURE
I
I
Illtill I0
15
(era H20)
so plotted against the numerical magnitudes of effort to produce average functions for /pa/ and /ba/ in both talking modes. The functions in the figure indicate that the effort associated with speech production is prothetic since the data approximate power functions-straight lines in log-log coordinateswith slopes providing exponents close to 2. The control utterances o f / p a / a n d /ba/ both gave slopes of approximately 1.8, values greater than those of Ringel et al. (1967). When the talkers were anesthetized, however, the slopes f o r / p a / a n d / b a / increased to 2.39 and 2.30, respectively (but the functions were still linear). As the figure indicates, the intraoral air pressures developed f o r / p a / w e r e always lower when anesthesia was administered; this effect was not evident for/ba/. Although the data show a change in the slope between the two conditions, the talkers had no difficulty in scaling the effort they used to produce the syllables in the anesthetic condition. The change in slope may reflect the uniqueness of speaking without oral sensation rather than loss of the ability PROSEK, HOUSE: Intraoral
Air Pressure
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
143
to produce articulate speech. The data present indirect evidence that intraoral air pressure is not a primary feedback cue if it must be monitored through the oral mucosa. DISCUSSION
AND C O N C L U S I O N S
The results of the present study show that, when compared to the consonants produced as control data, the consonants produced in the anesthetic condition gave slightly greater intraoral air pressures and longer durations. These changes are similar to those observed during syllabic stress and emphasis. The usual explanatio~n of the complex changes in fundamental frequency, intensity, and duration that accompany these suprasegmental features of speech is to assume that greater driving pressures (controlled by the thoracic musculature) are responsible. The increase in driving pressure can produce any or all of the observed changes, and, of course, also should produce a concomitant increase in the level of intraoral air pressure. The present data may be interpreted as meaning that, when deprived of oral sensation, the talkers are using more effort in speech production, effort being understood as a generic term describing the increase of subglottal pressure primarily through the use of abdominal and thoracic musculature. The question of whether it is appropriate to assign a significant role to changes of effort that can override sensory deprivation is answered affirmatively, in part at least, by the scaling data collected. These results demonstrate that. when sensory input essentially is absent in the oral cavity, a talker can scale levels of efforts in a fashion similar to his control performance. The changes in performance are small, and, most importantly, do not alter the linear scaling that is characteristic of the normal performance. These considerations make untenable Malecot's (1970) suggestion that changes in intraoral air pressure constitute an important parameter in the feedback control loop for articulation. This interpretation of the data in terms of effort, however, is speculative, and alternative explanations based on measurable parameters should be investigated. The question of whether air pressure monitored at other levels of the speech system constitutes a primary control parameter is not eliminated by these findings. Wyke (1967) has suggested, for example, that the mechano receptors of the larynx are sensitive to increases and decreases in subglottal air pressure, and that laryngeal adjustments are performed on the basis of such changes. In this case, the intraoral air pressure would be the result of the laryngeal control of the breathstream. An alternative to monitoring air pressure directly is to control the aerodynamic aspects of speech by the thoracic and abdominal musculature. Sensory information from these muscle systems might be used to establish the airflow rate appropriate for an utterance. The air pressures developed in the vocal tract would result from this flow rate, as well as from the muscular adjustments of the larynx and supraglottal articulators. 144 Journal of Speech and Hearing Research
18 133-147 1975
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
The durations measured in these data indicate that the timing of peripheral events, that is, the articulation of consonants, has been modified. These changes are compatible with the hypotheses of Perkell (1969) and Henke (1967) that tactile information is important for consonant production. The data may be interpreted as meaning that myotactic information concerning closures and constrictions was not adequately monitored in the anesthetic condition, and, therefore, the talkers produced speech more cautiously during that condition. The slowing down of the speech-producing activities, however, was not sufficient to produce gross changes in the articulation of consonants, and, therefore, the general level of intelligibility of the speech was not changed. Models of speech production using closed-loop feedback mechanisms usually are described as operating on a segment-by-segment basis. That is, their control parameters are associated with the production of elements about phoneme size. The segments that function in a given language, however, are not characterized by physical parameters, nor by articulatory descriptions that are invariant; in fact, perceptual differences among allophones of a given phoneme also are commonplace. This variability, or lack of invariance, imposes a heavy burden on closedloop control models, since they must account for the progressive and regressive assimilations that occur in various contexts; that is, they must look backward and forward in time before concluding that a segment has been produced appropriately. If the major contextual constraints on connected speech segments were highly determined by physiological factors, then, in spite of the complexity of the behavior, intraoral air pressure might qualify as a logical candidate for the control of articulation. Many contextual constraints, however, are not explainable in terms of physiology, but rather seem to be determined by rules of the language. While the prevalence of such rule-determined influences on articulation is not well quantified, the phonetic literature contains enough examples to suggest that rule-determined constraints may be more common than clearly identifiable physiological constraints. These considerations lead to the idea that, since most of the variability of segments is accounted for by the language rules, there is no reason to believe, a priori, that the loss of oral sensation will interfere seriously with speech production. Since the disturbances to speech in the anesthetized condition were minor, the generation of speech must have been primarily under the control of a pressure-sensing system other than a closed-loop one. This point of view is compatible with the hypothesis of MacNeilage (1970) that an open-loop component provides phonological information to the speechcontrol mechanism. Such phonological information would include the language rules governing contextual variations. Although the data of the present study do not resolve questions concerning the exact nature of oral-sensory feedback, or of the sensors involved in transmitting peripheral information, they argue strongly that an open-loop component is the primary controller of articulation.
PaOSF,K, HORSE: Intraoral
Air Pressure
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
145
ACKNOWLEDGMENT This research was completed while the authors were at Purdue University and is based on a doctoral dissertation submitted by the senior author to the graduate faculty of that institution. The research was supported by the Purdue Research Foundation. Requests for reprints should be sent to Robert A. Prosek, Army Audiology and Speech Center, Walter Reed Army Medical Center, Washington, D.C. 20012. REFERENCES Aar~Bxt~a, H. J., HixoN, T. J., and HaxaDY,J. C., Peak intraoral air pressures during speech. 1. Speech Hearing Res., 10, 196-208 ( 1967 ). BaOWN, W. S., and McGr.ONE, R. E., Constancy of intraoral air pressure. Folia phoniat., 21, 332-339 (1969). BROWN, W. S., McGtX)NE, R. E., TAaLOW, A., and SmPP, T., Intraoral air pressures associated with specific phonetic positions. Phonetica, 22, 202-212 (1970). HENCe-E,W. L., Preliminaries to speech synthesis based on an articulatory model. Conference on Speech Communications and Processing. Paper C5. Cambridge, Mass. (1967). LISKEn, L., Closure duration and the intervocalic voiced-voiceless distinction in English. Language, 33, 42-49 ( 1957 ). LIsKEa, L., Supraglottal air pressure in the production of English stops. Lang. Speech, 13,
215-230 (1970). MACNEmxCE, P. F., Motor control of serial ordering of speech. Psychol. Rev., 77, 182-196 (1970). MXLECOT, A., An experimental study of force of articulation. Stud. ling., 9, 35-44 (1955). MALECOr, A., The effectiveness of intraoral air-pressure-pulse parameters in distinguishing between stop cognates. Phonetica, 14, 65-81 (1966). Ma.r.F.COT, A., The force of articulation of American stops and fricatives as a function of position. Phonetica, 18, 95-102 (1968). MAta~COT, A., The Ienis-fortis opposition: Its physiological parameters. J. acoust. Soc Amer., 47, 1588-1592 (1970). NETSr.LL, R., Subglottal and intraoral air pressure during the intervocalic contrast of / t / and/d/. Phonetica, 20, 68-73 (1969). P~m~r,n, J. S., Physiology of Speech Production: Results and Implications of a Quantitative Cineradiographic Study. Cambridge, Mass.: MIT Press (1969). PaosEx, R. An evaluation of the role of oral sensation in consonant production. Doctoral dissertation, Purdue Univ. ( 1971 ). RINCEL, R. L., HOUSE, A. S., and MONTGOM~aY,A. A., Scaling arficulatory behavior: Intraoral air pressure. J. acoust. Soc. Amer., 42, 1209 (A) (1967). SCOTT, C. M., and RmCEL, R. L., Articulation without oral sensory control. 1. Speech Hearing Res., 14, 804-818 (1971). S~Ba~X.NY, J. D., WOrtTH, J. H., and SXKVDX,M., Intraoral pressure and rate of flow during speech. J. Speech Hearing Res., 9, 498-518 (1966). WYKE, B., Recent advances in the neurology of phonation: Reflex mechanisms in the larynx. Brit. J. Dis. Commun., 2, 2-14 (1967). Received June 9, 1974. Accepted October 14, 1974. APPENDIX
A
List of the words used in the experiment. The phoneme under study is indicated to the left of each word. Each word may be produced as a verb by placing the stress on the second syllable, or as a noun by placing the stress on the first syllable. p import b rebel t protest d produce k recall g progress f perfect v survey s insult z present 146 ]ournal of Speech and Hearing Research
18 133-147 1975
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
APPENDIX
B
The 20 words (noun forms a n d verb forms ) are shown embedded in sentence contexts. Children sometimes rebel with an insult. They wanted to present him with the perfect gift. A true rebel can protest just about anything. That country must import most of its produce from us. He tried to recall ff their survey was making any progress at all. Auditory recall and vision may be related. He wanted to survey his grandfather's land. The import of many protest movements is hard to assess. At present the theatrical world is on strike. He didn't progress very much in perfecting his invention. To insult a woman is very rude. The magician tried to produce a rabbit from his hat. APPENDIX
C
Words containing stop variants. The words are arranged according to articulatory characteristics, as indicated. The number in parentheses refers to the sentence in Appendix D in which the word appears.
Lip Rounded twice ( 7 ) uiet ( 13 ) une ( 6 ) goose ( 3 )
~
Dentalized
Intervocalic Voiced Flaps
Nasal Release
wrapping ( 5 ) writing ( 13 ) raking ( 5 )
keep most ( 6 ) might not ( 10 ) grab more ( 1 ) could not ( 7 )
Pseudo-Aspirated Release press ( 12 ) trade (2) crest (6)
Lateral Closure
Palatalized
Nasal Closure
Lateral Release
fat share ( 1 ) good years (4)
impassable ( 4 ) intense (11) ink ( 12 ) embodied ( 11 ) indeed (2) anger ( 11 )
title ( 7 ) fiddle (12)
at that ( 8 ) had there (10)
Voiceless Lenis spot ( 13 ) stopped (6) skilled ( 11 )
APPENDIX
melting ( 6 ) building (2)
Unaspirated Release of Normally Aspirated Stops wrap some ( 3 ) hot sun ( 6 ) pack some ( 9 ) D
List of sentences containing the stop variants examined in the present study. He tried to grab more than iust a fat share of the loot. The Chicago Trade Building was indeed an impressive structure. He wanted to wrap some of the goose in newspaper to take home with him. Even in good years that road is impassable. He did not like raking leaves or wrapping garbage. He stopped at the crest of the dune and tried to keep most of his gear from melting in the hot sun. Although he was trying twice as hard to win the title he could not succeed. At that time they couldn't be kept apart. They attempted to pack some valuables before being evacuated. Had there not been a cutback in spending, there might not have been a high unemployment rate. His account of the fight embodied all of the intense anger felt by the two skilled champions. He had to fiddle with the ink in order to keep the press running. The spot she found for writing letters was very quiet.
PROSEK, HOUSE: Intraoral Air Pressure 147
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
Walter Reed Army Medical Center, Washington, D. C. ARTHUR S. HOUSE
Institute for Defense Analyses, Princeton, New Jersey The effects of oral-sensory deprivation on the production of consonants was studied using narrow phonetic transcriptions and measurements of intraoral air pressure and duration. The speech materials were 20 bisyllabic words produced both in isolation and in sentences, and sentences that included words with 34 stop variants. These utterances were produced by four normal talkers and by the same talkers when deprived of oral sensation. The state of oral-sensory deprivation was induced by a series of mandibular, infraorbital, and palatal injections of 2~ xylocaine. The talkers also scaled levels of effort used to produce the syllables /pa/ a n d / h a / , both with and without the anesthetic. In general, after the administration of the anesthetic, the characteristic tongue carriage of the talkers was shifted posteriorly, the rate of speech was slower, and there were minor imprecisions in articulation consisting primarily of alterations in lip and tongue activity. In addition, consonants were produced with slightly greater intraoral air pressures and longer durations. The talkers had no difficulty in scaling levels of effort in either the normal or the anesthetic conditions, and maintained a linear relationship between effort and intraoral air pressure in both conditions. The results suggest that the talkers used more effort in producing speech in the anesthetic condition and are untenable with the idea that intraoral air pressure constitutes an important feedback parameter in controlling articulation. The speech produced by the talkers while anesthetized (that is, while without sensation in the mouth) was reasonably precise and must be postulated to have been under the control of a pressure-sensing system other than a closed feedback loop. Recent interest in oral sensation and perception has led to the development of models of speech production that specify certain oral-sensory mechanisms that m a y operate to control articulation. MacNeilage (1970), for example, has hypothesized that the motor system for speech includes two major control components; an open-loop component provides phonological information, and a closed-loop component provides moment-to-moment information from the peripheral speech mechanism. MacNeilage has suggested that the closed-loop information is supplied by the gamma-efferent system that provides information on instantaneous muscle length and the rate of change of muscle length. This information is used to achieve control of articulator movement that is dependent on the state of the muscle. Other authors, however, have proposed alternative oral-sensory mechanisms 133
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
that may provide the closed-loop information. Perkell (1969) has suggested that, during consonant production, intraoral air pressure is sensed in the oral cavity and monitored with myotactic feedback. In particular, he has proposed three modes of intraoral air pressure increase that provide information relative to three consonant classes: high intraoral air pressures are associated with voiceless fortis consonants; intermediate pressures are associated with voiced lenis consonants; and low intraoral pressures are associated with sonorants. A number of studies of the aerodynamic properties of consonant production have provided data that are compatible with Perkell's model. These studies have attempted to answer two questions about intraoral air pressure during consonant production. First, can consonants be distinguished on the basis of differences in intraoral air pressure? Second, are talkers aware of changes in intraoral air pressure? Several investigations have provided data that answer the first question (Subtelny, Worth, and Sakuda, 1966; Arkebauer, Hixon, and Hardy, 1967; Brown and McGlone, 1969; Netsell, 1969; Brown et al., 1970; Malecot, 1955, 1966, 1968, 1970). These studies show that voiceless fortis consonants are produced with higher peak intraoral air pressures than voiced lenis consonants; for most subjects, the intraoral air pressure developed during the repetition of words and sentences is reliably produced with little variation; and consonants occurring before a stressed vowel have greater peak intraoral air pressures than the same consonants occurring after a stressed vowel, or before and after an unstressed vowel. In fact, Malecot (1970) has suggested that intraoral air pressure is the feedback cue used by talkers to control the production of fortis and lenis consonants, since the differences in their intraoral air pressures are reliable and occur in any context. On the other hand, Lisker (1970) has claimed that consonants are not adequately distinguished on the basis of intraoral air pressure. Lisker's data may be said to demonstrate that intraoral air pressure cannot be the sole feedback cue for consonant production. There is little experimental evidence to answer the second question concerning sensitivity to changes in intraoral air pressure. Malecot (1966) reported that the difference limen for intraoral air pressure is 1 cm H20, or less, and that the difference in peak intraoral air pressure between voiceless fortis consonants and voiced lenis consonants was greater than 2 cm H20. MaIecot has interpreted these findings as evidence that intraoral air pressure can be monitored during speech production. Another bit of evidence has been supplied by Ringel, House, and Montgomery (1967), who reported that the effort used to produce initial consonants /p/ and /b/ was related to the intraoral air pressure by a power function with a slope of approximately one. This result demonstrated that when a talker is asked to change his effort, the intraoral air pressure also changes. It is possible that one of the cues used by the talkers to change their effort was intraoral air pressure. In summary, there is some evidence that intraoral air pressure may be
134 Journal o[ Speech and Hearing Research
18 133-147 1975
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
monitored directly to control consonant production. Differences in peak intraoral air pressure and in consonant duration have been reported between voiceless fortis and voiced lenis consonants in all contexts in which the opposition occurs. Normal talkers are sensitive to intraoral air pressure differences that are less than the differences observed between fortis and lenis consonants in speech. Furthermore, changes in intraoral air pressure are related linearly to changes in effort. The purpose of the present study was to determine the changes in intraoral air pressure and consonant duration that occur when talkers are deprived of oral sensation. Comparisons between the normal (that is, control) and the sensory-deprived (that is, anesthetized) conditions were made for consonants produced in isolated bisyllabic words and in sentences. In addition, the subjects estimated the magnitude of the effort used to produce the consonants/p/ and /b/. This task provided information on the perceptual changes in effort brought about by anesthesia. Finally, all utterances were transcribed to describe the articulatory changes brought about by eliminating sensation in the oral structures. METHOD Speech Material
Twenty bisyllabic words were used in the experiment. Half the words were verbs in which the second syllable received the stress, and half were nouns that differed from the verbs only in stress pattern. The words provided six stops and four fricatives as the initial phonemes of the second syllables. The list of words and the phonemes studied in each word are shown in Appendix A. To examine the effects of oral-sensory deprivation on the production of consonants in utterances that are longer than two syllables, a list of sentences, containing the words described above, was prepared. Subtelny et al. (1966), Malecot (1970), and Brown et al. (1970) have reported that differences in intraoral air pressure between voiceless fortis and voiced lenis consonants are observable in sentences, but the absolute values may not be as great as those observed in isolated words. The sentences used in the present study are listed in Appendix B. In addition to the sentences in Appendix B, a list of 13 sentences was prepared to provide a wide variety of allophonic variants of the stop consonants under study. These variants represent articulations that differ from simple stops in their closure, release, and voicing characteristics, but which occur frequently in connected speech. These sentences have been included to provide data on the effects of oral-sensory deprivation on more complicated articulations. The types of stop variants examined and the word in which the variant appears are delineated in Appendix C; the number in parentheses following each word refers to the sentence number in which the word appears. The words are shown embedded in sentence contexts in Appendix D.
PrtosEK, HousE: Intraoral Air Pressure 135
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
Experimental Conditions and Subjects The speech materials were produced in a control and an experimental mode by four young adult talkers. Each talker had normal speech and hearing, normal oral structures, and no history of sensory or motor disturbances. In the control condition (normal speech), the talkers were able to use all their usual sources of oral-sensory information. In the experimental condition, the oral structures were anesthetized to eliminate sensation. In both conditions the talkers were able to hear themselves normally. The state of oral-sensory deprivation (the anesthetic condition) was achieved by a series of nerve block injections of 2~ xylocaine. The injection series consisted of (1) bilateral mandibular injections that eliminated sensation in the lower lip and teeth, the anterior two-thirds of the tongue, the skin of the cheek, and the mucous membrane of the mouth and lower gingiva; (2) bilateral infraorbital injections eliminating sensation in the upper lip and teeth; (3) bilateral posterior palatine iniections that eliminated sensation in the mucous membrane of the velum and uvula; and (4) medial nasopalatine injections that eliminated sensation in the mucous membrane of the alveolar ridge and the anterior portion of the hard palate. The depth of the anesthetic was evaluated by use of a probe. All subjects were insensitive to touch and pressure in all anesthetized areas. The effectiveness of the anesthesia was further evaluated by asking the talkers to discriminate between geometric forms placed successively in the mouth. When the anesthesia was effective, the talkers were unable to discriminate between a square and an oval. The injection series, then, produced a profound loss of oral sensation.
Equipment A 12-French catheter connected to a pressure transducer was used to sense intraoral air pressure. The output of the transducer was amplified and recorded on one channel of an optical oscillograph. Static calibration of the pressure system was accomplished by a U-tube water manometer where known amounts of pressure were equated with the galvanometer deflection of the oscillograph. The insertion gain of the pressure system was determined by sealing the tip of the pressure transducer in a 2-cc coupler. Discrete frequencies from 10 to 350 Hz, in increments of 10 Hz, were introduced by means of a small transducer sealed in the top of the coupler. A condenser microphone, connected to a sound-level meter, was sealed in the bottom of the coupler. The sound pressure level inside the coupler was adjusted to 100 dB at each frequency, and the response was recorded on the oscillograph and measured in millimeters of H20. The response of the transducer was fiat from 10 to 350 Hz, and this response, as read from the oscillograph, was used as the reference in computing the relative amplitudes, in decibels, for the entire pressure system. The procedure was repeated with the catheter inserted between the coupler and the pressure transducer. The frequency response of the entire pressure system was quite 136 1ournal of Speech and Hearing Research
18 133-147 1975
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
fiat except for a resonance at about 100 Hz. This resonance did not interfere with the pressure measurements, since only low-frequency changes in pressure were of interest in the present study, that is, those below 50 Hz. During the actual collection of the speech samples, two high-quality microphones were placed in front of the talker at a distance of 12 inches. One microphone was used to record the speech waveform on a second channel of the oscillograph for comparison with the pressure tracings. The second microphone was used to record the words and sentences on a high-quality magnetic tape recorder. Procedures
In both talking conditions, the catheter was inserted through the nasal cavity of the talker until the experimenter observed the tip in the oropharynx. Each talker was given the opportunity to practice the words and sentences with the catheter in place. In the control condition, all materials were read in a normal conversational mode, and only utterances appropriate to Standard American English were accepted by the experimenter. For both conditions, the talker sat erect facing the microphones and first read the words in Appendix A. At the completion of this task, the talker read the materials in Appendix B and D as declarative sentences. In anticipation of possible series effects, a different randomized arrangement of the nouns, verbs, and sentences was used for each talker in both the control and anesthetic conditions. To counterbalance for possible sequential effects, two of the talkers read the list of nouns before the list of verbs, while the other two talkers read the lists in the reverse order. After completing the recordings of the words and sentences, each talker estimated the magnitude of the effort used in producing the s y l l a b l e s / p a / a n d /ha/. Each talker practiced producing a syllable at a comfortable, conversational level, after which he monitored this standard production visually on an oscilloscope (connected in parallel with the amplifier). The effort involved in this level of production was assigned a numerical value of 10. The talker then produced the same syllable at a level of effort corresponding to the magnitude of a number supplied by the experimenter; the numerical values of 5, 10, 20, and 30 were used in an incompletely counterbalanced arrangement. For each numerical magnitude, the talker uttered a syllable three times in sequence; the median of these three productions was used in the analysis. After each series of four productions, the talker practiced his (conversational) standard three times, monitoring his productions visually. To control for possible sequential effects, two of the talkers first estimated the effort used to produce the syllable/pa/, then the effort used to produce/bo/; the other two talkers scaled /ba/ before /pal. Peak intraoral air pressure was defined as the maximum pressure developed during the production of the consonant. For voiceless consonants, the maximum was easily observed. However, for voiced consonants, pressure variations due
PROSEK, HOUSE: Intraoral Air Pressure 137
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
to the vibrations of the vocal folds were superimposed on the low-frequency pressure changes. The midpoints of the variations were determined, and the peak was taken as the midpoint showing the greatest displacement. Consonant duration was determined by examining the waveform and pressure traces and selecting a point in time where articulatory changes could be observed. The articulatory changes observed for consonant onset were increases in pressure, offset of voicing (for voiceless consonants), and the reduction of waveform amplitude. The changes observed for consonant termination were a reduction in pressure to a level consistent with the following sound, the onset of voicing (for voiceless consonants), and an increase in waveform amplitude. RESULTS
Phonetic Transcriptions The speech changes described in this section of the paper are based on standard phonetic transcriptions of the speech materials. Due to space limitations, only the general findings of the phonetic analysis will be reported. (For more detailed analysis, see Prosek, 1971. ) The general intelligibility of the talkers was not affected by the loss of oral sensation. No formal intelligibility tests were administered, but under ordinary listening conditions there were no instances in which the speech of any talker could not be understood. This evaluation is based on the reports of four listeners. In particular, there were no gross changes in the intonation patterns used in the sentences, and the talkers did not change the assignment of syllabic stress. While all the talkers were insensitive to touch and pressure in the areas anesthetized and showed poor oral discrimination of geometric forms, the effect of the anesthetic on general intelligibility was not the same for all talkers. The speech of two of the talkers was better than the speech of the other two, when the number of articulatory distortions was the criterion. In general, an anesthetized talker modified his characteristic tongue carriage, talked slower, and changed the articulation of consonants in minor, essentially nonphonemic, ways. The tongue carriage was more posterior in the anesthetic condition than in the control condition. The result of this change was that primary closures and constrictions were made further back in the oral cavity. This change in tongue position, which was inferred independently from the transcribed material, was substantiated, in part at least, by the comments volunteered by the anesthetized talkers that their tongues seemed to be laying in the backs of their mouths. A slower rate of speech was observed for the two talkers whose speech was most affected by the anesthetic. This auditory observation was substantiated by measuring the duration of selected sentences in the control and anesthetic conditions. The durations of these sentences in the control condition were 2.5 to 3.5 seconds; in the anesthetic condition, the durations were 1.0 to 1.5 seconds longer than in the control condition. 138 Journal o[ Speech and Hearing Research
18 133-147 1975
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
The articulatory changes usually involved only one or two phonetic features. The primary articulatory effects were alterations in labial activity and the absence of fine adjustments of the apex and blade of the tongue. These findings agree well with the results reported in detail by Scott and Ringel (1971). For all talkers, the production of vowels in the anesthetic condition was very similar to vowel production in the control condition. When changes did occur in the anesthetic condition, they usually were of the following types: high back vowels were produced without expected lip rounding, r-colored syllabic nuclei were produced without expected retroflexion, and vowels were nasalized excessively. Intraoral Air Pressure
The measurements reported were gathered on consonants produced in word pairs differing in syllabic stress placement (see Appendixes A and B). Measurements of peak intraoral air pressure wer e averaged over subjects and categories such as voicing, manner of production, syllabic stress, context, and place of articulation. The intraoral air pressures measured for the consonants in the control condition are in substantial agreement with those of previous studies (Malecot, 1955; Subtelny et al., 1966; Arkebauer et al., 1967; Lisker, 1970). Consonants produced in the anesthetic condition were characterized by greater intraoral air pressure than consonants produced in the control condition. Everything else being equal, however, the difference in peak pressure between the two conditions did not exceed 1 cm H20. Since Malecot (1966) reported a difference ]imen for intraoral air pressure of 1 cm H20, the difference between the two conditions will not be considered significant. The data for individual subjects sometimes showed a pressure increase or decrease between the two conditions of from 1.5 to 2 cm H~O, but these instances were rare. The general findings are summarized in Table 1. The tabulated values show that, in agreement with all previous investigations, voiceless consonants were produced with greater peak pressures than voiced consonants. In agreement with Arkebauer et al. (1967), voiceless stops were produced with greater pressures than voiceless fricatives, and voiced fricatives were produced with greater pressures than voiced stops. The intraoral air pressure developed for consonants before stressed vowels was not greatly different from that for consonants before unstressed vowels. This finding agrees with the results of Lisker (1970), who reported only small changes in peak pressure as a function of syllabic stress. Malecot (1968), however, had reported that consonants before stressed vowels were produced with greater peak intraoral air pressure than consonants in any other position. In the control condition, consonants in isolated words showed slightly greater peak intraoral air pressures than consonants in the sentence contexts; in the anesthetic condition, however, no difference was evident. This is not significant since the difference in the control condition is only 6 mm HzO. There is a PROSEK, HOUSE:Intraoral Air Pressure 139
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
TAnhE 1. Summary of the mean intraoral air pressures (in centimeters of H20) for various consonants produced in the control and anesthetic conditions grouped into categories, as indicated. The consonants are from the words listed in Appendixes A and B. All entries in the difference column represent greater pressure in the anesthetic condition. Consonants Voicing and manner ptk fs vz bdg Syllabic stress stressed unstressed Context words sentences Place of articulation pbfv t d s z
kg
Control
Anesthetic
Difference
6.4 5.9 3.8 3.1
6.7 5.9 4.8 3.6
0.3 0.0 1.0 0.5
4.9 4.7
5.3 5.1
0.4 0.4
5.1 4.5
5.2 5.9.
0.1 0.7
4.5 4.9 5.0
4.6 5.5 5.9
0.1 0.6
0.9
tendency for consonants produced in the back of the oral cavity to have greater peak pressures than consonants produced anteriorly, but once again, no particular significance can be attached to this finding. In summary, the intraoral air pressure developed for consonants produced in the anesthetic condition did not differ greatly from those in the control condition. In both experimental conditions voiceless stops were produced with greater intraoral air pressure than voiceless fricatives, and voiced fricatives were produced with greater intraoral air pressure than voiced stops. Syllabic stress placement, type of context, and place of articulation were not distinguishable on the basis of intraoral air pressure. Duration The measurements of consonantal duration were obtained from the materials listed in Appendixes A and B. The durations were averaged and compared in the same manner as were the peak intraoral air pressures. The results are summarized in Table 2. The durations in the control condition agree well with the values reported by Malecot (1968, 1970) and with the spectrographic measurements reported by Lisker (1957). In the present and previous investigations, voiceless stops were found to have the longest durations and voiced stops were shortest. The average durations of consonants produced by the anesthetized talkers were 10 to 20 msec longer than those produced without sensory deprivation. The data for individual subjects showed increased durations in the anesthetic condition of up to 45 msec in rare instances.
140 Journal of Speech and Hearing Research
18 133-147 1975
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
TABLE 2. Summary of the average durations (in milliseconds ) for various consonants produced in the control and anesthetic conditions, grouped into categories as indicated. The consonants are from the words listed in Appendixes A and B. All entries in the difference column represent longer duration in the anesthetic condition. Consonants
Voicing and manner ptk fs vz bdg Syllabic stress stressed unstressed Context words sentences Place of articulation pbfv tdsz kg
Control
Anesthetic
Difference
150 143 112 93
159 149 123 115
9 6 11 22
132 116
144 131
12 15
129 119
143 131
14 12
122 123 131
134 133 154
12 10 23
In agreement with the findings of Malecot (1968), and independent of experimental condition, consonants before stressed vowels were longer than consonants before unstressed vowels. Similarly, consonants in isolated words were longer than consonants in sentence contexts. Finally, the durations of consonants produced in the back of the oral cavity were greater than those of consonants produced anteriorly or medially in the mouth. This difference is easily explained, however, since the anterior and medial categories include fricatives and stops, while the posterior productions are stops only. In summary, the durations of consonants produced in the anesthetic condition were greater than consonants produced in the control condition. Syllabic stress and context affected the duration of consonants, making them longer before a stressed vowel and in isolated words. Stop Variants"
The stop variants on which this discussion is based are those listed in Appendix C. In general, the voiceless stop variants were produced with lower peak intraoral air pressures and shorter durations in the control condition than the stops reported in the preceding section. The voiced stop variants, however, were produced with approximately the same intraoral air pressures as the more simple stops, but their durations were shorter. The average intraoral air pressures and durations for various classes of stop variants are summarized in Tables 3 and 4, respectively. Stops produced with a nasal release and a lateral release had peak pressures of 2.7 and 2.6 cm HzO, respectively. In addition, the durations of these stop
PROSEK, HOUSE: Intraoral Air Pressure 141
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
TABLE 3. Summary of the mean intraoral air pressures (in centimeters of H20) developed during the production of stop variants. See Appendix C for the word materials. All entries in the difference column represent greater pressure in the anesthetic condition.
Stop Variants
Control
Voiceless Voiced Lip rounded Dentalized Palatalized Voiceless lenis Intervocalic voiced flaps Nasal closure Lateral closure Nasal release Lateral release Unaspirated release Pseudo-aspirated release
5.6 3.3 5.3 5.6 3.8 6.9 4.0 4.4 3.9 2.7 2.6 6.0 6.4
Anesthetic Difference 6.6 4.3 6.4 6.4 5.6 7.9 4.7 5.6 5.4 3.8 2.7 6.4 7.3
1.0 1.0 1.1 0.8 1.8 1.0 0.7 1.2 1.5 1.1 0.1 0.4 0.9
TABLE 4. Summary of the average durations (in milliseconds) for the stop variants listed in Appendix C. All entries in the difference column represent longer durations in the anesthetic condition.
Stop Variants Voiceless Voiced Lip rounded Dentalized Palatalized Intervocalic voiced flaps Nasal closure Lateral closure Nasal release Lateral release Pseudo-aspirated release
Control 130 86 134 104 149 93 103 86 84 53 155
Anesthetic Difference 144 100 149 123 184 105 110 103 98 69 163
14 14 15 19 35 12 7 17 14 16 8
allophones are shorter than for most of the other variant classes. Thus, the release characteristics of the stops seem to affect their production more than place of articulation and voicing. A comparison of the intraoral air pressures and durations of the stop variants in the control and anesthetic conditions shows the same magnitude of difference as for the simple consonants. An exception is the duration of the palatalized alveolar stops, which is 35 msec greater in the anesthetic condition. Thus, the effects of oral-sensory deprivation are the same for the complex allophones as for the simpler stops.
Scaling of Effort In this task each talker produced three syllables in sequence for each numerical magnitude supplied to him by the experimenter, and the median of these three productions was used in the analysis. For each numerical magnitude, 142 lournal of Speech and Hearing Research
18 133-147 1975
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
the geometric mean of the medians was calculated, and a least-squares statistic was used to calculate the slope of the function relating effort to intraoral air pressure. When the apparent magnitude of the effort associated with speech production is plotted against an appropriate physical magnitude (such as intraoral air pressure) on log-log coordinates, the slope of the resulting function corresponds to the exponent of the power function governing the growth of the psychological magnitude. In Figure I the air pressure measurements have been 5o I I I J ~ I IJl~rr ,o-/~//b~/ / // / ;-- -~ONTROL / / /~///~, 30 9
a
EXPER.,
/o
__
-
I0
9
,?
FZCURE 1. Functions relating the effort used to produce/pa/and/ba/ to the intraoral air pressure in the control and anesthetic conditions.
,
ill ,I
I
I I 5
INTRAORALAIR PRESSURE
I
I
Illtill I0
15
(era H20)
so plotted against the numerical magnitudes of effort to produce average functions for /pa/ and /ba/ in both talking modes. The functions in the figure indicate that the effort associated with speech production is prothetic since the data approximate power functions-straight lines in log-log coordinateswith slopes providing exponents close to 2. The control utterances o f / p a / a n d /ba/ both gave slopes of approximately 1.8, values greater than those of Ringel et al. (1967). When the talkers were anesthetized, however, the slopes f o r / p a / a n d / b a / increased to 2.39 and 2.30, respectively (but the functions were still linear). As the figure indicates, the intraoral air pressures developed f o r / p a / w e r e always lower when anesthesia was administered; this effect was not evident for/ba/. Although the data show a change in the slope between the two conditions, the talkers had no difficulty in scaling the effort they used to produce the syllables in the anesthetic condition. The change in slope may reflect the uniqueness of speaking without oral sensation rather than loss of the ability PROSEK, HOUSE: Intraoral
Air Pressure
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
143
to produce articulate speech. The data present indirect evidence that intraoral air pressure is not a primary feedback cue if it must be monitored through the oral mucosa. DISCUSSION
AND C O N C L U S I O N S
The results of the present study show that, when compared to the consonants produced as control data, the consonants produced in the anesthetic condition gave slightly greater intraoral air pressures and longer durations. These changes are similar to those observed during syllabic stress and emphasis. The usual explanatio~n of the complex changes in fundamental frequency, intensity, and duration that accompany these suprasegmental features of speech is to assume that greater driving pressures (controlled by the thoracic musculature) are responsible. The increase in driving pressure can produce any or all of the observed changes, and, of course, also should produce a concomitant increase in the level of intraoral air pressure. The present data may be interpreted as meaning that, when deprived of oral sensation, the talkers are using more effort in speech production, effort being understood as a generic term describing the increase of subglottal pressure primarily through the use of abdominal and thoracic musculature. The question of whether it is appropriate to assign a significant role to changes of effort that can override sensory deprivation is answered affirmatively, in part at least, by the scaling data collected. These results demonstrate that. when sensory input essentially is absent in the oral cavity, a talker can scale levels of efforts in a fashion similar to his control performance. The changes in performance are small, and, most importantly, do not alter the linear scaling that is characteristic of the normal performance. These considerations make untenable Malecot's (1970) suggestion that changes in intraoral air pressure constitute an important parameter in the feedback control loop for articulation. This interpretation of the data in terms of effort, however, is speculative, and alternative explanations based on measurable parameters should be investigated. The question of whether air pressure monitored at other levels of the speech system constitutes a primary control parameter is not eliminated by these findings. Wyke (1967) has suggested, for example, that the mechano receptors of the larynx are sensitive to increases and decreases in subglottal air pressure, and that laryngeal adjustments are performed on the basis of such changes. In this case, the intraoral air pressure would be the result of the laryngeal control of the breathstream. An alternative to monitoring air pressure directly is to control the aerodynamic aspects of speech by the thoracic and abdominal musculature. Sensory information from these muscle systems might be used to establish the airflow rate appropriate for an utterance. The air pressures developed in the vocal tract would result from this flow rate, as well as from the muscular adjustments of the larynx and supraglottal articulators. 144 Journal of Speech and Hearing Research
18 133-147 1975
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
The durations measured in these data indicate that the timing of peripheral events, that is, the articulation of consonants, has been modified. These changes are compatible with the hypotheses of Perkell (1969) and Henke (1967) that tactile information is important for consonant production. The data may be interpreted as meaning that myotactic information concerning closures and constrictions was not adequately monitored in the anesthetic condition, and, therefore, the talkers produced speech more cautiously during that condition. The slowing down of the speech-producing activities, however, was not sufficient to produce gross changes in the articulation of consonants, and, therefore, the general level of intelligibility of the speech was not changed. Models of speech production using closed-loop feedback mechanisms usually are described as operating on a segment-by-segment basis. That is, their control parameters are associated with the production of elements about phoneme size. The segments that function in a given language, however, are not characterized by physical parameters, nor by articulatory descriptions that are invariant; in fact, perceptual differences among allophones of a given phoneme also are commonplace. This variability, or lack of invariance, imposes a heavy burden on closedloop control models, since they must account for the progressive and regressive assimilations that occur in various contexts; that is, they must look backward and forward in time before concluding that a segment has been produced appropriately. If the major contextual constraints on connected speech segments were highly determined by physiological factors, then, in spite of the complexity of the behavior, intraoral air pressure might qualify as a logical candidate for the control of articulation. Many contextual constraints, however, are not explainable in terms of physiology, but rather seem to be determined by rules of the language. While the prevalence of such rule-determined influences on articulation is not well quantified, the phonetic literature contains enough examples to suggest that rule-determined constraints may be more common than clearly identifiable physiological constraints. These considerations lead to the idea that, since most of the variability of segments is accounted for by the language rules, there is no reason to believe, a priori, that the loss of oral sensation will interfere seriously with speech production. Since the disturbances to speech in the anesthetized condition were minor, the generation of speech must have been primarily under the control of a pressure-sensing system other than a closed-loop one. This point of view is compatible with the hypothesis of MacNeilage (1970) that an open-loop component provides phonological information to the speechcontrol mechanism. Such phonological information would include the language rules governing contextual variations. Although the data of the present study do not resolve questions concerning the exact nature of oral-sensory feedback, or of the sensors involved in transmitting peripheral information, they argue strongly that an open-loop component is the primary controller of articulation.
PaOSF,K, HORSE: Intraoral
Air Pressure
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
145
ACKNOWLEDGMENT This research was completed while the authors were at Purdue University and is based on a doctoral dissertation submitted by the senior author to the graduate faculty of that institution. The research was supported by the Purdue Research Foundation. Requests for reprints should be sent to Robert A. Prosek, Army Audiology and Speech Center, Walter Reed Army Medical Center, Washington, D.C. 20012. REFERENCES Aar~Bxt~a, H. J., HixoN, T. J., and HaxaDY,J. C., Peak intraoral air pressures during speech. 1. Speech Hearing Res., 10, 196-208 ( 1967 ). BaOWN, W. S., and McGr.ONE, R. E., Constancy of intraoral air pressure. Folia phoniat., 21, 332-339 (1969). BROWN, W. S., McGtX)NE, R. E., TAaLOW, A., and SmPP, T., Intraoral air pressures associated with specific phonetic positions. Phonetica, 22, 202-212 (1970). HENCe-E,W. L., Preliminaries to speech synthesis based on an articulatory model. Conference on Speech Communications and Processing. Paper C5. Cambridge, Mass. (1967). LISKEn, L., Closure duration and the intervocalic voiced-voiceless distinction in English. Language, 33, 42-49 ( 1957 ). LIsKEa, L., Supraglottal air pressure in the production of English stops. Lang. Speech, 13,
215-230 (1970). MACNEmxCE, P. F., Motor control of serial ordering of speech. Psychol. Rev., 77, 182-196 (1970). MXLECOT, A., An experimental study of force of articulation. Stud. ling., 9, 35-44 (1955). MALECOr, A., The effectiveness of intraoral air-pressure-pulse parameters in distinguishing between stop cognates. Phonetica, 14, 65-81 (1966). Ma.r.F.COT, A., The force of articulation of American stops and fricatives as a function of position. Phonetica, 18, 95-102 (1968). MAta~COT, A., The Ienis-fortis opposition: Its physiological parameters. J. acoust. Soc Amer., 47, 1588-1592 (1970). NETSr.LL, R., Subglottal and intraoral air pressure during the intervocalic contrast of / t / and/d/. Phonetica, 20, 68-73 (1969). P~m~r,n, J. S., Physiology of Speech Production: Results and Implications of a Quantitative Cineradiographic Study. Cambridge, Mass.: MIT Press (1969). PaosEx, R. An evaluation of the role of oral sensation in consonant production. Doctoral dissertation, Purdue Univ. ( 1971 ). RINCEL, R. L., HOUSE, A. S., and MONTGOM~aY,A. A., Scaling arficulatory behavior: Intraoral air pressure. J. acoust. Soc. Amer., 42, 1209 (A) (1967). SCOTT, C. M., and RmCEL, R. L., Articulation without oral sensory control. 1. Speech Hearing Res., 14, 804-818 (1971). S~Ba~X.NY, J. D., WOrtTH, J. H., and SXKVDX,M., Intraoral pressure and rate of flow during speech. J. Speech Hearing Res., 9, 498-518 (1966). WYKE, B., Recent advances in the neurology of phonation: Reflex mechanisms in the larynx. Brit. J. Dis. Commun., 2, 2-14 (1967). Received June 9, 1974. Accepted October 14, 1974. APPENDIX
A
List of the words used in the experiment. The phoneme under study is indicated to the left of each word. Each word may be produced as a verb by placing the stress on the second syllable, or as a noun by placing the stress on the first syllable. p import b rebel t protest d produce k recall g progress f perfect v survey s insult z present 146 ]ournal of Speech and Hearing Research
18 133-147 1975
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
APPENDIX
B
The 20 words (noun forms a n d verb forms ) are shown embedded in sentence contexts. Children sometimes rebel with an insult. They wanted to present him with the perfect gift. A true rebel can protest just about anything. That country must import most of its produce from us. He tried to recall ff their survey was making any progress at all. Auditory recall and vision may be related. He wanted to survey his grandfather's land. The import of many protest movements is hard to assess. At present the theatrical world is on strike. He didn't progress very much in perfecting his invention. To insult a woman is very rude. The magician tried to produce a rabbit from his hat. APPENDIX
C
Words containing stop variants. The words are arranged according to articulatory characteristics, as indicated. The number in parentheses refers to the sentence in Appendix D in which the word appears.
Lip Rounded twice ( 7 ) uiet ( 13 ) une ( 6 ) goose ( 3 )
~
Dentalized
Intervocalic Voiced Flaps
Nasal Release
wrapping ( 5 ) writing ( 13 ) raking ( 5 )
keep most ( 6 ) might not ( 10 ) grab more ( 1 ) could not ( 7 )
Pseudo-Aspirated Release press ( 12 ) trade (2) crest (6)
Lateral Closure
Palatalized
Nasal Closure
Lateral Release
fat share ( 1 ) good years (4)
impassable ( 4 ) intense (11) ink ( 12 ) embodied ( 11 ) indeed (2) anger ( 11 )
title ( 7 ) fiddle (12)
at that ( 8 ) had there (10)
Voiceless Lenis spot ( 13 ) stopped (6) skilled ( 11 )
APPENDIX
melting ( 6 ) building (2)
Unaspirated Release of Normally Aspirated Stops wrap some ( 3 ) hot sun ( 6 ) pack some ( 9 ) D
List of sentences containing the stop variants examined in the present study. He tried to grab more than iust a fat share of the loot. The Chicago Trade Building was indeed an impressive structure. He wanted to wrap some of the goose in newspaper to take home with him. Even in good years that road is impassable. He did not like raking leaves or wrapping garbage. He stopped at the crest of the dune and tried to keep most of his gear from melting in the hot sun. Although he was trying twice as hard to win the title he could not succeed. At that time they couldn't be kept apart. They attempted to pack some valuables before being evacuated. Had there not been a cutback in spending, there might not have been a high unemployment rate. His account of the fight embodied all of the intense anger felt by the two skilled champions. He had to fiddle with the ink in order to keep the press running. The spot she found for writing letters was very quiet.
PROSEK, HOUSE: Intraoral Air Pressure 147
Downloaded From: https://jslhr.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
