Nasal Air Flow During Normal Production

AMY E. THOMPSON, M.S. THOMAS J. HIXON, Ph.D. Tucson, Arizona 85721

Nasal air flow was measured during the speech of 112 normal subjects (59 females and 53 males) ranging in age from three years to 37 years, six months. Flow was zero during nearly all oral and utterances, suggesting that velopharyngeal closure was air-tight. Flow occurred during all nasal and during adjacent to nasal consonants. Both age and sex effects were demonstrated for flow on vowels preceding nasal consonants. These effects were interpreted as showing that progressively older subjects and female subjects demonstrate earlier anticipatory coarticulation in prepara- tion for forthcoming nasal consonants.

Introduction is not to be expected in normal speakers (Nus- The velopharyngeal mechanism is a me- baum, Foley, and Wells, 1935; Moll, 1962; chanical valve situated between the oral and Van den Berg, 1962). nasal portions of the upper airway. Its control The second question inquires into the pos- of oral-nasal coupling accomplishes two im- sibility of a developmental schedule for gain- portant things in normal speech production: ing control over the velopharyngeal-closure one, it enables the development of oral air mechanism. The impetus for this question is pressure and air flow sufficient to produce the lack of information pertinent to this topic many of the consonants of speech, and two, it and the fact that several motor-speech behav- enables the generation of perceived to iors continue to be refined toward adultlike be free of hypernasal quality. Despite the standards well after normal speech is acquired recognized importance of normal velopharyn- and sometimes as late as 11 to 12 years of age geal function, many of the details of such (Kent, 1976). function remain to be specified. The present The third question of importance to the study is one of a series of studies being con- present investigation is an inquiry as to the ducted in the Department of Speech and possible existence of a sex difference with Hearing Sciences at the University of Arizona respect to control of the velopharyngeal-clo- to provide needed information about velopha- sure mechanism. A paucity of information ryngeal function in normal speech produc- exists in this regard, although that which is tion. available (McKerns and Bzoch, 1970) would Three questions are of central importance lead one to believe that there could be differ- in this part of this series of studies. The first ences in function between the sexes based on concerns the extent to which air-tight velo- differences in velopharyngeal structure. __ pharyngeal closure is characteristic of so- The general design of the present investi- called phonetically-oralized speech elements. gation involved an attempt to address these The impetus for this question is the belief that three questions through the measurement of is commonly held by many individuals that nasal air flow during the speech of a large complete closure of the velopharyngeal port group of males and females whose ages were distributed over a wide range. More specifi- cally, subjects were selected to encompass a The authors are affiliated with the University of Ari- range extended downward to the youngest zona, Tucson, Arizona. Dr. Hixon is a Professor in the age at which standard tasks could easily be Department of Speech and Hearing Sciences. This work was supported by Research Grant NS- performed and extended upward into middle- 09656 from the National Institute of Neurological and aged adulthood. Flow was selected as the Communicative Disorders and Stroke. method for studying velopharyngeal function 412 Thompson and Hixon, Nasar FLow 413 because of the simplicity of its measurement and the resulting analog of the speech signal and because, for the purposes of answering was displayed on the second channel of the the questions mentioned above, it constituted storage oscilloscope. When it was deemed nec- a sufficient tool for the study of normal speak- essary to record data permanently, displays ers (Warren, 1967). on the screen of the oscilloscope were photo- graphed with a Polaroid camera. For certain Method activities, an electronic metronome was used Suspects. One hundred twelve individuals to aid subjects in the pacing of their speech. served as subjects. Ninety-two of these sub- SpEECH SAMPLE. Utterance tasks consisted jects, 49 females and 43 males, ranged in age of three groups of activities: sustained pro- from three to 18 years. Ages were distributed ductions, syllable repetitions, and nonsense relatively evenly so that at least one subject, productions embedded in a carrier phrase. and usually more than one, fell within each Sustained productions included individual four-month interval within this age range. utterances of /i/, /s/, /z/, and /n/ at normal Typically, subjects of both sexes were in- conversational pitch, loudness, and quality. cluded within each interval. The remaining Each utterance was made following an inspi- persons studied were 20 adults, ten of each ration of about twice normal depth and lasted sex, ranging in age from 18 years, two months, for about five seconds. Syllable repetitions to 37 years, six months. Most of these subjects included productions of /ti/, /di/, /si/, /zi/ were less than 25 years of age. All 112 subjects and /ni/, each in trains of seven utterances, were selected to meet the following criteria: made at normal conversational pitch, loud- normal speech and voice commensurate with ness, and quality, with equal stress on each age, hearing reported to be within normal syllable, and at an utterance rate of three per limits, no known structural or neurological second. Nonsense productions included the disorders, and no known respiratory infections vowel-consonant-vowel (VCV) combinations or allergies at the time of testing. All subjects /iti/, /idi/, /isi/, /izi/, and /ini/ in the carrier were monolingual English speakers living in phrase "Say ___ again," produced at con- Tucson, Arizona. versational pitch, loudness, and quality, with EquiprmEnt. Two aspects of speech produc- normal prosody, and with stress placed on the tion were recorded: nasal air flow (hereinafter second vowel in the VCV. Each of the sus- referred to as flow) and the acoustic pressure tained productions and syllable repetition wave (hereinafter referred to as the speech trains were performed three times in succes- signal). Flow was channeled through an in- sion and each of the VCV productions was dividually-custom-formed mask placed over performed in a separate carrier phrase three the nose. Masks were made of Dux-seal, a times in succession. Utterances included in relatively compliant, clay-like material. Flow the study were chosen to encompass a variety from the mask was sensed by a double-coned, of phonetic contexts that could be executed two-square inch, Silverman-type, pneumo- easily by the younger subjects. Design of the tachometer coupled to a differential air-pres- sample was also guided by a desire to include sure transducer. The output analog voltage of a variety of manners of consonant production the latter was amplified, and the resulting (, , and nasal), and to include signal was low-pass filtered through a system voicing contrasts for cognate pairs. A single with a sharp roll-off that markedly attenuated place of consonant and vowel production was the signal above 20 Hz. The conditioned-flow chosen, namely, lingua-alveolar. This place analog was displayed on one channel of a had the advantage of avoiding lip gestures two-channel storage oscilloscope. The flow that would interfere with the seal formed by analog was calibrated through use of a con- the nasal mask. The high vowel /i/ was se- tinuously-variable flow pump whose output lected for study within the front vowel series was directed through a rotameter and the because of its presumed stringent velopharyn- pneumotachometer in series. The speech sig- geal-closure requirements (Moll, 1962), a fac- nal was sensed by a condenser microphone tor we considered important for insuring the positioned approximately six inches below strength of a target behavior were a develop- and to the front of the subject's lips. The mental schedule to be manifested. The utter- output from the microphone was amplified ance rate of three per second was found in a 414 Cleft Palate Journal, October 1979, Vol. 16 No. 4 pilot study to be the maximum rate at which all 333 sustained productions of /n/ involved very young subjects could perform reliably. nasal flow. Criterion scores were computed ProcEpURrE. Each subject was seated up- for each of the subjects as the mean peak flow right and positioned so that the screen of the on the three /n/-productions. The resulting oscilloscope was out of the visual field. A mask scores for the subject group yielded a mean was molded for the nose and was positioned value of 98.9 cc/sec, with a range of 23.3 cc/ on the subject. Each mask formed a small sec to 246.7 cc/sec, and a standard deviation tube, the proximal end of which fitted airtight of 38.1 cc/sec. Perusal of the criterion score against the upper lip and mid-facial region data for /n/ revealed no indication that nasal surrounding the nose. Care was taken to in- flow was related either to the age or to the sex sure that the walls of the mask did not block of the subjects studied. the external nares or contact the nasal alae. SyLLABLE ReEpPETITIONS. The syllables /ti/, Next, the microphone was positioned, and the /di/, /si/, /zi/, and /ni/, each performed in subject was instructed to repeat each utter- trains of seven, comprised the group of sylla- ance produced by an investigator. The model ble repetitions. The middle three repetitions utterances for the subjects met the criteria in each train were selected for analysis. No discussed in the previous section. In the case nasal flow occurred for any of the utterances of syllable repetitions, the investigator mo- of the syllables /ti/, /di/, /si/, and /zi/. That' delled utterance rate to the flashing and click- is, no flow occurred for 2664 sound elements ing of the metronome. Subject utterances (4 different syllables X 2 phonemes each X judged to be inappropriate copies of the 3 repetitions X 111 subjects) involving oral model were not accepted, and additional consonant and vowel elements." By contrast, speech samples were elicited until an appro- of the 666 speech sounds - - 333 /n/'s and 333. priate copy was obtained. Those utterances /1i/'s - - involved in the /ni/ syllable repeti- produced appropriately were stored on the tions, all of the nasal consonants and all but screen of the oscilloscope for analysis. An in- nine of the vowels were accompanied by nasal vestigator recorded the flow magnitudes of flow. The nine speech sound utterances not interest directly from the stored display. In accompanied by flow were performed by cases where the flow analog could not be three subjects who uttered /i/ (three times segmented instantly with regard to phonetic each) without nasal flow. Criterion scores events, the oscilloscopic display was photo- were computed for both /n/ and /i/ for each graphed for later perusal. subject on the three productions of each pho- neme. These computations were derived from Results

We choose to divide the description of the 111 participants. To prevent this subject's data from results in accordance with the three groups of unduly distorting the descriptive statistics for the overall activities performed by the subjects: sustained group, we deemed it reasonable to isolate these data for separate discussion. They are considered in a subsequent productions, syllable repetitions, and non- footnote at the end of the RESULTS section. sense productions embedded in a carrier * The statement "no nasal flow" should be qualified phrase. in the context of these first four syllable repetitions and SustaAmEp Propuctions. The phonemes in the context of the first four nonsense syllables studied /i/, /s/, /z], and /n/ comprised the group of to mean that no air passed from the mouth to the nose by way of the velopharynx. Very small magnitudes of sustained productions investigated. No nasal nasal flow were observed for some subjects' productions flow occurred for any of the utterances of the of the utterances mentioned. These were typically less first three of these phonemes. That is, velo- than & 5 cc/sec, involved very gradual changes in flow, pharyngeal closure presumably was air-tight and often involved swings in flow from positive to nega- tive and vice versa. Such observations are consistent with for the 999 samples (3 phonemes X 3 repeti- those of Lubker and Moll (1965), who demonstrated that tions X 111 subjects') involving sustained their occurrence is related to minor displacements of the oral consonant or vowel elements. By contrast, velum which alter the volume of the nasal cavity, change a its internal pressure slightly, and cause small flows in and out of the nose in the presence of air-tight velopharyngeal ' Data for our subject group proved to be homogene- closure. From the perspective of the measurements of ous. However, one of the 112 subjects presented data that importance to this investigation, these flows were viewed usually were distinctly different from those of the other as artifacts and were specified as zero. Thompson and Hixon, nasar row 415 measurements of flow made at the midpoint from three to 18 years, and subdividing this of each phoneme as determined from the range into three-year intervals, the data accompanying speech signal. Flows measured showed the following proportions of subjects at these points in the utterances proved to to demonstrate nasal flow at the midpoint of correspond closely to the maximum flow dur- the vowel in each interval, respectively: 13.6% ing the /n/ and the minimum flow during at three to six years; 20.0% at six to nine; the /i/. Resulting scores were characterized 42.9% at 9 to 12; 78.6% at 12 to 15; and 71.4% by a subject group mean nasal flow for /n/ of at 15 to 18 years. For the adult group, which 88.8 cc/sec, with a range of 40.0 cc/sec to covered a wide range of unevenly distributed 273.0 ce/sec, and a standard deviation of 34.1 ages, the proportion was 50%. cc/sec. For /i/ the subject group mean was With regard to sex, 32 of the 47 subjects 47.5 cc/sec, with a range of 0.0 cc/sec to 167.0 demonstrating nasal flow at the midpoint of ce/sec, and a standard deviation of 25.3 cc/ the initial /i/ were female. This proportion sec. Perusal of the data for /n/ and /i/ for (i.e., 68.1%) was observed not only overall but the /ni/ utterances revealed no age or sex also in approximately the same magnitude effects. across the five contiguous age subdivisions Nonsense Propuctions EMBEDDED IN A just mentioned and in the adult subject group. CarriEr PuraseE. The syllables /iti/, /idi/, For the final vowel in /ini/, flows measured /isi/, /izi/, and /ini/, embedded in a carrier at the midpoint yielded a subject group mean phrase, comprised the group of nonsense pro- value of 43.3 cc/sec, with a range of 8.0 cc/ ductions investigated. Nasal flow was zero for sec to 120.0 cc/sec, and a standard deviation all of the 3996 individual speech sounds pro- of 23.1 cc/sec. Examination of the criterion duced for the first four of these VCV syllables score data failed to reveal nasal flow effects (4 syllables X 3 phonemes each X 3 repeti- for either age or sex.* tions X 111 subjects). The fifth of the non- Discussion sense syllables studied, /ini/, invariably was accompanied by nasal flow during the nasal The discussion of the results of this investi- consonant segment /n/. At the midpoint of gation will be organized under three subhead- /n/, flow for the subject group (for 333 utter- ings: fully oralized utterances, fully nasalized ances) was of a mean magnitude of 94.2 cc/ utterances and simultaneously oralized and sec, with a range of 30.0 cc/sec to 260.0 cc/ nasalized utterances, and clinical implications sec, and a standard deviation of 38.9 cc/sec. and research needs. Criterion score data for /n/ showed no indi- Furry Orauzep UTrERANCES. The fully cation of age or sex effects. oralized utterances of this investigation Flows measured at the midpoint of the turned out to be the isolated utterances of initial vowel in the /ini/ nonsense productions /i/, /s/, and /z/, the syllable repetitions of resulted in a subject group mean value of 12.2 /ti/, /di/, /si/, and /zi/, and the carrier cc/sec, with a range of 00.0 cc/sec to 80.0 cc/ phrase embedded syllables /iti/, /idi/, /isi/, sec, and a standard deviation of 15.9 cc/sec. and /izi/. Nasal flow was zero and the velo- Flows were zero at the midpoint of the initial pharynx presumably was closed air-tight for vowel in /ini/ for 64 of the 111 subjects ° As mentioned in Footnote 1, one of the subjects studied. Excluding these 64 subjects, recalcu- presented data that were single in kind for the majority lation of the mean, range, and standard de- of utterances sampled. This subject, a female, 13 years, viation for the subject group resulted in values seven months of age, performed in a manner typical of of 28.7 ce/sec, 3.3 cc/sec to 80.0 ce/sec, and the overall subject group on sustained productions and 17.4 cc/sec, respectively. Criterion scores on the utterances of /ni/ and /ini/. However, she demon- strated flow on certain phonemes in non-nasal phonetic the initial /i/ for our 111 subjects revealed contexts for the syllable repetition and nonsense produc- two clear performance effects, one related to tion activities Nasal flow was found to be zero for all subject age and the other related to subject vowel utterances while a clear burst of flow occurred sex. With regard to age, there existed a mod- during the production of almost every /t/, /d/, /s/, and /z/. These bursts were of short duration and ranged in erate trend for an increasing proportion of the magnitude from 20 cc/sec to 80 cc/sec. Flow for the subjects to demonstrate nasal flow with in- voiced element in each cognate pair was roughly twice creasing age. Considering the contiguous ages that for the corresponding voiceless element. 416 Cleft Palate Journal, October 1979, Vol. 16 No. 4 all of these utterances for 111 of the subjects. viction because of the measurement problems The phonemes included in the fully oralized they acknowledge. Hardy (1967) has severely category are routinely categorized by - criticized the instrumentation system used in ticians as oral consonants and vowels. The the study by Van Hattum and Worth (1967) present findings indicate that these speech on a number of bases. Some of the same sounds are fully oralized by both sexes as early criticisms apply equally to the instrumenta- as age three and that the use of air-tight tion of Emanuel and Counihan (1970). velopharyngeal closure for their production is Also relevant to the present discussion are maintained into middle-aged adulthood (37 available cinefluorgraphic data on velopha- years, 6 months, was the upper limit studied ryngeal function during speech conditions of here). the nature studied in the present investiga- These findings on fully oralized utterances tion. Moll (1962) presents data for velopha- are in agreement with those of other research- ryngeal closure on vowels that reveal that, ers who have used pneumotachometers to although the velopharynx usually is closed for study nasal flow during speech samples simi- vowel productions, it is open for a small per- lar to those used in the present investigation centage of utterances, usually in the neigh- (Warren and DuBois, 1964; Lubker and Moll, borhood of 15% for the /i/-vowel contexts he 1965; Machida, 1967; Lubker, Schweiger, and studied. Bzoch (1968), by contrast, found no Morris, 1970; Lubker, 1973). That is, oral velopharyngeal opening in a cinefluoro- consonants and vowels have been found by graphic study of upper airway function. Aside these other authors to be produced with no from his formal investigation, Bzoch also in- nasal flow or with very minimal flow. In those spected some 200 cinefluorographic films of instances where minimal flow occurred, it normal speakers and found only occasional appeared not to be the result of a velopharyn- velopharyngeal opening on oral speech ele- geal leak but of velar elevation in the presence ments. He attributed these occurrences of of an air-tight velopharyngeal seal (Lubker velopharyngeal opening to slow utterance and Moll, 1965; Lubker, 1973). Recall from rates. We are inclined to accept Bzoch's con- footnote 2 that, in the present study, we clusion based on our own further observations treated such occurrences of flow as artifacts of several adult subjects who performed the and specified them as zero. study protocol at slower utterance rates than Researchers who have used warm-wire were studied in the 112 subjects. Two of six flowmeters to study nasal flow during speech such subjects, all of whom demonstrated air- samples similar to ours, have provided data tight velopharyngeal closure for syllable rep- 'that are both supporting and conflicting with etitions at the usual rate of three per second, regard to our findings. For example, Quigley occasionally showed small nasal flows when et al. (1963) and Quigley (1967) found essen- utterance rate was dropped below 1.5 per tially no nasal flow on utterances that are second. Bzoch would appear to have made an routinely classified as oral consonants and important observation when he stated that vowels. By contrast, both Van Hattum and "a slow rate of utterance of syllables is a factor Worth (1967) and Emanuel and Counihan related to the occasional occurrence of velo- (1970) report nasal flows for certain oral con- pharyngeal opening on non-nasal speech ele- sonant and vowel elements. Both pairs of the ments." Clearly, the utterance rate factor latter authors acknowledge the fact that some needs to be specified or controlled in subse- of these flows are probably attributable to the quent investigations. occurrence of velar elevations in the presence Recall that one of the 112 subjects in this of an air-tight velopharyngeal seal. In other study failed to demonstrate air-tight velopha- cases, where the flows are seemingly of too ryngeal closure during some of the consonant large a magnitude to be accounted for solely elements in her productions of the syllable by velar displacements, both pairs of authors utterances of concern here. Her unique nasal express concern that these may reflect leaks flow patterns suggest that she briefly and between the oral and nasal sections of their slightly opened the velopharynx during the flow-measuring apparatus. It is difficult to production of consonant elements but that view either of those reports with much con- she maintained air-tight velopharyngeal clo- Thompson and Hixon, Nnasar rrow 417

sure for adjacent vowel elements. Despite this to be influenced by factors such as phonetic idiosyncratic adjustment pattern, this sub- context and age and sex of the subject, al- ject's speech was normal in all regards. Al- though the nature of such influences are not though air-tight velopharyngeal closure is necessarily congruent from one study to an- clearly typical for the oral consonants and other (Arkebauer, Hixon, and Hardy, 1966; vowels mentioned here, the data on this single Subtelny et al., 1966; Diggs, 1972; Bernthal subject with a velopharyngeal leak and nor- and Beukelman, 1978). Too little is currently mal speech raise the question about the ne- known to make much of the fact that the cessity for air-tight closure. Other authors, present data did not show effects for context, most notably Warren (1967), have docu- age, or sex. mented relatively small nasal flows in subjects The findings on fully nasalized utterances judged to have adequate speech. For example, are in agreement with those of some authors taking a very conservative velopharyngeal or- (Warren and DuBois, 1964; Subtelny et al., ifice size of 10 mm* as representing an "ade- 1966) but are in disagreement with those of quate" degree of velopharyngeal closure, one others (Lubker and Moll, 1965; Lubker, et al., might expect to find nasal flows during voice- 1970). Specifically, in the present investiga- less stop-plosive productions that would reach tion, peak nasal flows were observed at mag- magnitudes in the neighborhood of 100 cc per nitudes approaching about 100 cc per second second. It is important to keep in mind that on the average. Warren and DuBois (1964) the amount of velopharyngeal opening that a and Subtelny et al., (1966) show data of these listener will tolerate before labeling speech as approximate magnitudes for utterances of deviant may depend upon whether the devia- /m/, /n/, and /ng/. By contrast, the studies tion pertains to hypernasal voice quality or to of Lubker and Moll (1965) and Lubker et al. misarticulations - - the first predominantly an (1970), done in the same laboratory, present acoustics problem and the second predomi- data that are typically twice the magnitude nantly an aeromechanics problem. of those presented here and also by other Furry NasauzED UTTERANCES AND SIMUL- investigators. Lubker and Moll (1965), for TANEOUSLY ORALIZED AnD NasaumzEp UTTER- example, show flows of 200 cc per second peak ances. We did not measure oral air flow in magnitude for /n/ in running speech phrases this investigation. Presumably, it was zero comparable to ours, while Lubker et al. (1970) during the sustained productions of /n/, dur- show flows of over 210 cc per second on the ing the major part of the production time of average for the /m/ in an /imi/ utterance all of the /n/'s in the syllable repetitions of context. It is difficult to reconcile flow mag- /ni/, and in the syllable utterances of /ini/ nitudes twice the magnitude of ours without embedded in the carrier phrase (Lubker and invoking an explanation that embodies major Moll, 1965; Subtelny, Worth, and Sakuda, laryngeal adjustments that would result in 1966; Gilbert, 1968). Here, all /n/ produc- substantially higher flows being delivered to tions are categorized as fully nasalized. The the oral cavity. We note that not only are the flow data for such nasalized utterances proved nasal flows reported by Lubker and his asso- to be comparable for the three different pho- ciates higher for the context mentioned above, netic contexts in which /n/ was studied. Fur- but that they are also approximately double thermore, they revealed no apparent age or what we observed under conditions where sex effects for this subject group. It is tempting flows from the nose were believed to be the to generalize these findings as indicative of result of velar adjustments in the presence of fully matured velopharyngeal behavior that air-tight velopharyngeal closure. Perhaps the is acquired by at least age three and is main- nasal-flow-channel calibration for the studies tained through middle-aged adulthood. It of Lubker and Moll (1965) and Lubker et al. must be remembered, however, that several (1970) was off by a constant factor related to factors influence the flow of air from the nose the misreading of the standard scale. We have during speech, including nasal pathway re- no other explanation. sistance, velopharyngeal airway resistance, The simultaneously oralized and nasalized and oral air pressure (Warren, 1967). The last utterances of this investigation were the /i/ of these alone, that is, oral pressure, is reported vowels paired with the /n/ 418 Cleft Palate Journal, October 1979, Vol. 16 No. 4

(that is, /ni/ and /ini/). Oralization is as- erate trend for a greater proportion of the sumed for these utterances on the basis of past subjects to show nasal flow at the mid-point studies of oral air flow (Lubker and Moll, of the initial vowel with increasing age. We 1965; Subtelny et al., 1966; Machida, 1967; interpret these data as evidence that progres- Van Hattum and Worth, 1967; Emanuel and sively older subjects were demonstrating ear- Counihan, 1970). Evidence for the nasalized lier anticipatory coarticulation in preparation component of these utterances is clear from for the forthcoming nasal. This progressively our nasal flow data and from the data of earlier occurrence of flow for older subjects others (Warren and DuBois, 1964; Lubker constitutes a sort of "spread of " and Moll, 1965; Van Hattum and Worth, across the /ini/ syllable that moves in the 1967). direction of the first vowel. Unfortunately, Our findings for nasal flow in the three our measurements at the mid-point of the different contexts for the /i/ vowel revealed vowel do not precisely define the function of that the /i/'s following the nasal consonant the age effect we have mentioned. Obtaining /n/ (that is, in /ni/ and in the second vowel the necessary information in this regard in /ini/) were of closely comparable magni- would require specification of the precise mo- tudes while those measured on the /i/ preced- ment of velopharyngeal opening within each ing the nasal consonant were only about one- vowel utterance. Because data were not re- fourth the magnitude of the other two. The corded. permanently, such determinations occurrence of nasal flow on /i/, a phoneme cannot be made post-hoc. We are inclined to classified as oralized in typical phonetic interpret our observations in terms of motor schemes, is clearly attributable to the coarti- skills acquisition associated with the increased culation of the nasal consonant /n/ with the age of the subjects. That is, with increasing vowels immediately preceding and following. age, up to at least age 12, when behavior In the case of the two contexts in which vowels becomes more like that of adults, it appears follow the nasal, the occurrence of nasal flow that subjects take more and more advantage on the vowel appears to be related to carry- of preparatory gestures for following sounds. over coarticulation in which the velopharynx This may reflect an increased economy of does not close air-tight for the vowel after biomechanics with increasing experience in being open for the nasal. This finding is not using the speech mechanism. Kent (1976), for unexpected and is consistent with the obser- example, has summarized studies that show vations of others who have studied velopha- that several aspects of motor control for ryngeal function in similar contexts (Warren speech do not become fully adult-like until and DuBois, 1964; Lubker and Moll, 1965; approximately this age. We believe the poten- Moll and Daniloff, 1971). Such carryover tial for understanding the developmental _coarticulation is basically a consequence of schedule for motor control of speech may be the biomechanical nature of the speech mech- enhanced by more detailed studies of the anism. spread of the nasalization phenomenon we For the utterance contexts in which /i/ have mentioned. preceeded the nasal consonant (that is, the Recall that a clear sex effect was appar- first vowel in /ini), the occurrence of nasal ent in our flow data on the initial vowel in - flow on the vowel is clearly a result of another /ini/. Specifically, approximately 70% of the form of coarticulation, namely anticipatory. subjects who demonstrated nasal flow at the That is, the speaker begins to open the velo- mid-point of the vowel were females, both pharynx during the initial vowel in /ini/ in overall and in the different age sub-groups. anticipation of the nasal consonant to follow. We are uncertain as to the reasons for this sex This form of nasalization of the vowel in a effect. Possibilities include a potentially nasal context has been observed by research- greater tolerance for the spread of nasalization ers using different observational techniques in females without the perception of hyper- (Warren and DuBois, 1964; Moll and Dani- nasality and biomechanical differences be- loff, 1971). tween the sexes that might favor greater an- With regard to age, our nasal flow data on ticipatory coarticulation by females. We lean the initial /i/ in /ini/ demonstrated a mod- toward the latter as a possible explanation for Thompson and Hixon, Nnasar rrow 419 our findings. McKerns and Bzoch (1970) have mation about the closure demands for normal shown that velopharyngeal function differs speech production. Thus, the anticipated clin- considerably between males and females on a ical observation in normal velopharyngeal number of dimensions. Differences demon- closure on oral sounds would be no flow. strated included the orientation of the velum However, a small amount of flow should not to the pharynx during velopharyngeal clo- be interpreted to mean velopharyngeal in- sure, the relationship of the inferior point of competence. Fortunately, for clinical con- contact of the velum to the palatal plane, the cerns, our results provide the first extensive degree of elevation of the velum, the amount data to illustrate that, at least for the oralized of contact between the velum and the poste- speech samples studied here, the clinician rior pharyngeal wall, the distance from the need not be concerned about subject differ- uvula to the posterior pharyngeal wall, and ences with respect to age (over a wide range), the length of the velum in function for speech. sex, or phonetic context. It seems reasonable to suppose that other Our data on flow for the nasal consonant dynamic velopharyngeal function differences /n/ constitute a substantial body of informa- might exist between the sexes and that these tion relative to the flow associated with fully might be manifested in measures such as the nasalized consonants. They may well repre- nasal-flow recordings made in the present sent a useful standard of comparison for those study. That is, there may be certain mechan- concerned with secondary surgical procedures ical advantages that would permit females and prosthetic management techniques that and not males to engage in an earlier antici- occasionally "over-correct" for velopharyn- patory opening of the velopharyngeal mech- geal incompetence to the extent that nasal anism in preparation for /n/. Far too little is breathing is impaired and hyponasality oc- known about velopharyngeal biomechanics curs. The central tendency and dispersion and aeromechanics to speculate with any de- measures we have provided for nasal flow gree of certainty in this regard. should provide a useful starting point for Cruimicar Impu1cations Anp REsEaRCH those interested in determining "typical" val- NeErEps. The results of this investigation ap- ues for normal subjects. pear to have implications for both clinical With respect to research, there are clear and research endeavors. It is our experience needs to be filled. Knowing that air-tight that many clinicians are of the opinion that velopharyngeal closure must develop some- some degree of velopharyngeal leakage may time between birth and age three, it is ob- routinely accompany the production of what viously important to proceed with studies to are phonetically classified as oralized speech determine the age at which children gain sounds. The results of this investigation of a adult-like control of the velopharynx for or- large number of normal subjects do not sup- alized speech sounds. It should prove profit- port this notion. Rather, it is apparent that, able to study such children with regard to for the types of utterances sampled, air-tight both age and sex. New technological advances velopharyngeal closure is nearly universal. in flow measurement instrumentation, such This finding would appear to have implica- as miniature thermistors, now make it possible tions with regard to the evaluation and the to pursue studies of this type without encum- physical and behavioral management of per- berance to the speech mechanism. Studies sons with velopharyngeal incompetence. That using such procedures are now underway in is, the target for "normal" velopharyngeal the University of Arizona Laboratories. closure is an air-tight seal. It remains to be We acknowledge that the speech utterances determined what precisely the tolerable limits studied in this investigation were limited with are for other than air-tight velopharyngeal regard to sampling the phonemes of American closure. The fact that one of our subjects English. Additional informal study in our demonstrated normal speech in the presence laboratory of a small group of adults suggests of a small amount of nasal flow suggests that to us that our findings of air-tight velopha- air-tight closure may not be a necessary req- ryngeal closure on vowel utterances should uisite for normal speech. More detailed stud- not be generalized to all vowels. For example, ies of subjects of this type may provide infor- some subjects demonstrate air-tight velopha- 420 Cleft Palate Journal, October 1979, Vol. 16 No. 4 ryngeal closure on high vowels but show small and rate of airflow, J. Speech Hear. Res., 10, 650-654, velopharyngeal leaks on low vowels. This 1967. Krent, R., Anatomical and neuromuscular maturation of being the case, subsequent research should the speech mechanism: Evidence from acoustic studies, take into account vowels in addition to the J. Speech Hear. Res., 19, 421-447, 1976. /i/ we studied here. LusBkER, J., Transglottal airflow during stop consonant Finally, a number of aspects of velopharyn- production. J. Acoust. Soc. of Amer., 53, 212-215, 1973. geal function cannot be specified with nasal- LuBkER, J., anp Morr, K., Simultaneous oral-nasal air flow measurements and cinefluorographic observations flow measurements alone. It would seem use- during speech production, Cleft Palate J., 2, 257-272, ful to undertake a comprehensive study of 1965. velopharyngeal function using other forms of LuskErR, J., ScHwEIGER, J., anp MoraRIs, H., Nasal airflow innocuous aeromechanical procedures for characteristics during speech in prosthetically man- aged cleft palate speakers, J. Speech Hear. Res., 13, 326- quantifying such factors as velopharyngeal 338, 1970. orifice area and the relationship between oral MacHipaA, J., Air flow rate and articulatory movement pressure and nasal flow. The technique of during speech, Cleft Palate J., 4, 240-248, 1967. Warren and DuBois (1964) seems well suited McKErns, D., anp BzocH, K., Variation in velopharyn- to such pursuits. geal valving: The factor of sex, Cleft Palate J., 7, 652- 662, 1970. Reprints: Thomas J. Hixon, Ph.D. Morr, K., Velopharyngeal closure on vowels, J. Speech Professor Hear. Res., 5, 30-37, 1962. Department of Speech and Hearing Sciences Morr, K., anp Danxntrorr, R., Investigation of the timing University of Arizona of velar movements during speech, J. Acoust. Soc. of Amer., 50, 678-684, 1971. Tucson, Arizona 85721 NusBAuM, E., FourEy, L., anp WELLs, C., Experimental studies of the firmness of the velar-pharyngeal occlu- sion during the production of English vowels, Speech References Monographs, 2, 71-80, 1935. e ArkEBAUER, H., Hixon, T., anp Harpy, J., Peak intraoral QuicLEy, L., A comparison of air-flow and cephalometric air pressures during speech, /. Speech Hear. Res., 10, techniques for evaluation of normal and cleft palate 196-208, 1967. patients, Amer. J. Orthod., 53, 423-445, 1967. BERNTHAL, J., AND BEUKELMAN, D., Intraoral air pressure Quiciey, L., WEBstTER, R., CorrEy, R., KELLENER, R., during the production of /p/ and /b/ by children, AND Grant, H., Velocity and volume measurements of youths, and adults, 7. Speech Hear. Res., 21, 361-371, nasal and oral air-flow in normal and cleft-palate 1978. speech, utilizing a warm-wire flow-meter and two- Bzocx, K., Variations in velopharyngeal valving: The channel recorder, J. Dent. Res., 42, 1520-1527, 1963. factor of vowel changes. Cleft Palate J., 5, 211-218, SuBTELNY, J., WortH, J., anp M., Intraoral 1968. pressure and rate of flow during speech, J. Speech Hear. Dicos, C., Intraoral air pressure for selected English Res., 9, 498-518, 1966. consonants: A normative study of children, Master of Van pen Berc, J., Modern research in experimental Science Thesis, Purdue University, 1972. phonetics, Folia Phoniatrica, 14, 81-149, 1962. EmanuEL, F., aAnp CoOUNIHAN, D., Some characteristics of Van Hatrunm, R., anp WortH, J., Air flow rates in normal oral and nasal air flow during plosive consonant pro- speakers, Cleft Palate J., 4, 137-147, 1967. duction, Cleft Palate J., 7, 249-260, 1970. Warren, D., Nasal emission of air and velopharyngeal GirBERT, H., Oral airflow during stop consonant produc- function, Cleft Palate J., 4, 148-156, 1967. tion, Master of Science Thesis, University of Wisconsin, Warren, D., anp DuBois, A., A pressure-flow technique 1968. for measuring velopharyngeal orifice area during con- Haroy, J., Techniques of measuring intraoral air pressure tinuous speech, Cleft Palate J., 1, 52-71, 1964.