Speech Physiology, Speech Perception, and Acoustic Phonetics

Cambridge Studies in Speech Science and Communication

Speech physiology, speech perception, and acoustic phonetics

Philip Lieberman and Sheila E.BIumstein

In this series: The phonetic bases of speaker recognition Francis Nolan Pal terns o f sounds Ian M addieson From tex t to speech: the M ITalk system Jonathan Allen, M. Sharon Hunnicut and Dennis Klatt Neurolinguistics and linguistic aphasiology: an introduction David Caplan

Speech physiology, speech perception, and acoustic phonetics

Philip Lieberman and Sheila E. Blumstein Department o f Cognitive and Linguistic Scicnccs, Brown University

T h e rig h t o f th e I 'n ir v r .u tv o f C a m b rid g e lo p r in t u m l s e lf u ll m a n n e r o f b u n k s •«■ “O a.

E
o c

[kimzj; simplification errors in which a sound or syllable is deleted, e.g. green [gin]; addition errors in which an extra sound or syllable is added, e.g. see —►[stij; and environment errors in which a particular sound error that occurs can be accounted for by the influence o f the surrounding phonetic environment, e.g. roast beef -> [rof bif). Note that the final consonant cluster o f the word roast is assimilated to the final [f] o f b e e f(Blumstein, 1973). It is worth noting that analyses o f the phoneme substitution errors o f aphasics are consistent with the view that sound segments are comprised o f phonetic features. As we discussed in C hapter 8 , many linguists and speech researchers have suggested that sound segments arc comprised o f a bundle of phonetic features. Whjen patients substitute one sound segment for another, the substituted sound is usually different by one phonetic feature from that of the attempted or target sound. For example, patients may make voicing errors, doll -» [tal], place errors, e.g. teams -> [kimz], but they rarely make errors involving both voicing and place errors in the same sound substitution, e.g. doll -> *[kal]. Nearly all aphasics show this same pattern o f errors. Speech production impairments are not the only type o f speech deficit found in aphasic patients. M any aphasics also show speech perception impairments. They often have difficulty in discriminating words or nonsense syllables which contrast by a single phonetic feature, e.g. bait vs. date or [beip] vs. [deip] (Blumstein et al., 1977a; Jauhiancn and Nuutila, 1977; Miceli et a i s 1978; Miceli et a l., 1980; Baker, Blumstein and Goodglass, 1981). The testing procedures are quite simple and straightforward. The patient is presented via a tape recording with pairs of words or nonsense syllables, such as bait date or bait-bait. On each trial, he is asked to press a key marked YES if the two stimuli are the same, and N O if the two stimuli are different. In this way, the patient need only press a button. He does not have to speak which may be very difficult. Results o f such discrimination tests have shown that aphasic patients, regardless o f type o f aphasia, have more difficulty discriminating words or nonsense syllables which contrast by a single phonetic feature than by more than one feature, and they have particular difficulty when that contrast is for place o f articulation. If aphasic patients have difficulty discriminating stimuli contrasting in voicing or place of articulation, perhaps they will not show categorical perception of the sounds o f speech. As we discussed in C hapter 7, adult listeners show categorical perception o f speech in that they are able to discriminate between sounds reliably only when the pair of sounds they are discriminating between lie across a category boundary. Moreover, infants as 218

Som e current topics in speech research young as a few days old also show categorical-like discrimination functions. How would aphasics perform on such tasks? Studies exploring categorical perception in aphasics have focused on two phonetic dimensions - voicing (Basso, Casati and Vignolo, 1977: Blumstein et a l 1977b; G a n d o u r and D a rd a ran an d a , 1982) and place o f articulation (Blumstein et al.s 1984). The stimuli used in these studies were computergenerated. One series varied in VO T to explore the voiced- voiceless dimen­ sion. The other series explored the dimension o f place o f articulation. The stimuli varied in the frequency of the form ant transitions apropriate for the syllables [ba], [da] and [ga], and in the presence or absence o f a burst preceding the transitions. Aphasic subjects were asked to perform two tasks. The first task was to identify the initial consonant by pointing to the appropriate written letter. The second task was to discriminate pairs o f stimuli by pressing a button marked YES if the pair o f stimuli were the same, and NO, if they were different. Three patterns o f results emerged. One group o f patients were able to perform the task like normal subjects and thus showed categorical perception. A second group found both tasks too difficult and could neither reliably identify nor discriminate the sounds. A third group, however, showed an interesting dissociation. They were unable to identify the sounds, but they were able to discriminate them. More importantly, as Figure 9.7 shows, they showed a categorical-like discrimination function. T h at is, they were only able to discriminate those stimuli which lay across a category boundary. More importantly, the shape o f the obtained function and boundary values were similar to those found for subjects who could both identify and discriminate the stimuli. Thus, these aphasic patients showed discrimination boundaries that appear to be similar to those obtained for normal adults and for infants. These results underscore the stability o f the categorical nature o f speech.

Excrcises 1.

A b r illia n t s u r g e o n d e v is e s a p r o c e d u r e f o r s u p r a la r y n g e a l v o c a l t r a c t t r a n s p l a n t s . W h a t w o u ld h a p p e n to th e s p e e c h o f a n o th e r w is e n o r m a l a d u l t if th e s u p r a l a r y n g e a l v o c a l tr a c t o f a n a d u lt c h im p a n z e e w e re e x c h a n g e d w ith h im ? T h e p e r s o n is a n a c t o r w h o w a n ts to lo o k a u t h e n t i c f o r a ro le in a m o v ie a b o u t A u s tr a lo p i lh e c i n e s ( h u m a n - li k e a n im a ls w h o liv ed b e tw e e n 3 a n d I m illio n y e a r s a g o ) . W o u ld th e c h im p a n z e e b e a b le to ta lk if h e h a d th e a c t o r 's h u m a n s u p r a l a r y n g e a l v o c a l tr a c t? W h y ?

2.

W h a t a n a t o m i c a l a n d n e u r o lo g ic a l f a c to r s u n d e r lie th e p r e s e n t f o rm o f h u m a n sp eech?

3.

C h i l d r e n a r e s a id to a c q u ir e th e in t o n a t i o n o f t h e i r n a tiv e la n g u a g e in th e first y e a r o f life. H o w c o u ld y o u te s t th is th e o r y ? In a p a r a g r a p h o r tw 'o p r e s e n t th e p r o c e d u r e s th a t y o u w o u ld u se to te s t th is h y p o th e s is .

219

Som e current topics in speech research

F ig u re 9.7. D is c r im in a tio n o f f h a d n g n ] s tim u li f o r n o r m a l s u b je c ts a n d a p h a sic s. T h e s tr a ig h t lin e s c o r r e s p o n d to th e b u rs t p lu s tr a n s itio n s tim u li a n d th e d o t t e d lin es th e tr a n s itio n o n ly s t i m u l i . T h e v e r tic a l lin e a t s tim u lu s p a ir 7 - 9 in th e to p a n d b o tto m p a n e ls in d ic a te th a t th e c o m p u te d fu n c ti o n s f o r th e [ b d ] d is c r im in a tio n p a irs w a s b a s e d o n a d iffe r e n t n u m b e r o f s u b je c ts ( N ) th a n th e [ d g f d is c r im in a tio n p a irs. F ro m B lu m s te in cl a l., 1984.

4.

In w h a l w a y s d o th e d a t a f ro m th e a c q u is itio n o f s p e e c h a n d p a th o l o g y o f s p e e c h p r o v id e d if f e r e n t b u t c r itic a l in s ig h ts i n to th e n a t u r e o f s p e e c h p r o d u c t i o n a n d s p e e c h p e r c e p tio n ?

5.

W h a t d iffe re n c e s a r e th e r e b e tw e e n B ro ca* s a n d W e r n ic k e 's a p h a s ic s in sp e e c h p r o d u c t io n ? W h a t d o th e s e r e s u lts tell u s a b o u t th e n e u r o lo g ic a l b a s e s o f sp e e c h p r o d u c ti o n ?

220

10

Acoustic correlates of speech sounds

We have discussed the sounds o f speech in terms of articulatory and acoustic data and in terms o f theoretical models o f speech. In this last chapter, we present some o f the acoustic correlates o f various speech sounds o f English. This review will not be a comprehensive study o f the acoustic correlates of the sounds of hum an speech or even o f the sounds o f English, but it should be a useful starting point for more detailed study. One of the challenges o f speech research is to determine what aspects o f the acoustic signal are relevant to the listener for perceiving the sounds o f speech. As is apparent from the earlier discussions on the acoustics of speech, the speech signal is complex with temporal, durational and spectral variations. However, research has shown that listeners can perceive speech with a greatly “ stripped d o w n ” version of the acoustic signal. In other words, only certain aspects o f the acoustic signal seem to be relevant to the listener for perceiving the phonetic dimensions o f speech. These relevant attributes for speech are called acoustic correlates or acoustic cues. Let us review some o f the acoustic cues necessary for the perception o f English speech sounds.

Vowels The frequency positions o f the first three formants are sufficient cues for listeners to identify the vowels o f English. The formant frequency relations that specify the vowels of English arc inherently relational rather than absolute since different-sized supralaryngeal vocal tracts will produce different ab so­ lute form ant frequencies. Figure 10.1 shows the mean values o f F ,, / s , and F5 of the vowels o f American English spoken by adult males. Perhaps the best way to remember the formant frequency patterns is to start with th c qu an ta l vowels [i], [u] and [a]. The vowel [i] has the highest F 2 and F$ o f all English vowels. The convergence o f F2 and F3 results in a well-defined high frequency peak in the spectrum o f the vowel’s transfer function. The vowel [i] also has a low F t . The numbers entered on Figure 10.1 are the means derived from the Peterson and Barney (1952) analysis. The means o f F ,, F2 and F3 o f [i] are 270, 2290, and 3010 Hz respectively. Like [i], the vow7el [u] also has a low F x at 300 Hz. 221

Acoustic correlates o f speech sounds 3

3010

2550 2480

2410

2440

2410

2290 2240

2240

1990 1840 N

1720

X

> > o c [k£n]. In some languages o f the world, vowel nasalization is distinctive phonetically in that it is used to distinguish words o f different meanings, e.g. in French m ot [mo] meaning “ w ord” is distinguished from mon [mo] meaning “ my.” Research results on the perception o f nasal and oral vowels in speakers of English (which docs not have distinctive vowel nasalization) and other languages (which do have distinctive vowel nasalization) have shown that for all of the listeners, the primary acoustic cue for vowel nasalization is a reduction in the spectral prominence o f the first formant. This is accomplished by either broadening the F\ peak (making it wider in bandwidth) or creating an additional spectral peak nearby (Delattre, 1968; Hawkins and Stevens, 1985). The perceptual conse­ quences o f vowel nasalization are higher confusions when people are asked to identify them (Bond, 1976). 223

Acoustic correlates o f speech sounds A

Burst

[ta]

I ka]

Ti me F ig u re 10 .2 . F o r m a n t tr a n s itio n s a n d h u r s ts f o r th e sy lla b le s [ b n d a g a p a ta k a ] .

Stop consonants The stop consonants o f English [p t k b d g] share the same manner of articulation in that they arc produced with the rapid release of a complete closure in the vocal tract. There are several acoustic cues that contribute to the perception o f a stop consonant. One is the release burst, and the other is the rate and duration of the formant transitions, particularly of the first formant (Libcrman et al., 1956; Keating and Blumstein, 1978). Both cues are the acoustic consequence o f the rapid release of the stop closure. The duration of the burst is generally of the order of 5 15 milliseconds and that of the transitions is 20 40 milliseconds. Either cue alone will be sullicient for the listener to perceive a stop consonant. The stop consonants [ p b f d k g] vary in p 1 ’ce o f articulation - [p b] are labials, [t d] are alveolars and [k g] are velars. The acoustic cues to place of articulation in stop consonants include the spectrum o f the burst release and form ant transitions. In the top part of Figure 10.2, bursts and formant frequency patterns are shown for the consonant- vowel syllables [ba], [da] and [ga]. The steady-state formants correspond to the formant frequency values 224

Acoustic correlates o f speech sounds for the vowel [a]. N ote that the second and third formants both rise from a lower frequency relative to the vowel steady-states for [ba], and arc thus described as having rising transitions, and the burst is appended to the second form ant transition. Fo r [da], the second a nd third form ants are fa llin g relative to the vowel steady-state, and the burst excites F3. And for [ga], the second and third formants diverge or “ spread a p a r t” from each other with F2 falling and F 3 rising. The burst is juxtaposed to the F2 and F 3 transitions. The frequency of the burst and the pattern o f form ant transitions give rise to the perception o f different places o f articulation. The actual frequencies at which the bursts occur and the formant transitions begin vary as a function o f vowel context (Cooper et a i , 1952). While the burst and transitions can be visually separated on a spectrogram, as is shown in the schematic drawing of Figure 10.2, it is not clear that the perceptual system separately extracts these cues in the perception o f place of articulation in stop consonants. Rather, they may form a single integrated c u t (Stevens and Blumstein, 1979). There is always a spectral continuity between the burst and form ant transitions. T h a t is, there is a continuity between the frequency spectrum o f the burst and the frequency spectrum at the beginning o f the form ant transitions. W ithout this continuity, the perceptual perform ­ ance for place o f articulation is diminished (D orm an, Studdert-Kennedy and Raphael, 1977; Fischer-Jorgensen, 1972). A m ong the stop consonants o f English, [p t k] are voiceless and [b d g] are voiced. There are a multiple o f acoustic cues distinguishing voiced and voiceless stops. One o f the first that was systematically investigated was F x cutback. If form ant frequency patterns for [ba da ga] as shown in Figure 10.2A are synthesized but with the F y transition portions eliminated, as shown in Figure 10.2B, then the consonants are perceived as voiceless, i.e. [p t k] (Cooper et a l, 1952). Moreover, when the F2 a n d / ^ transitions are “ excited” by noise rather than by voicing during the cutback interval, voicing responses are more consistently heard. The noise excitation in the transitions corre­ sponds to noise generated at the glottis (Fant, 1960, p. 19). Burst am plitude and aspiration am plitude also play a perceptual role, with greater values resulting in an increase in voiceless percepts (Repp, 1979). It is the com bination o f both Fi cutback and aspiration which corresponds to the voice-onset time param eter which we discussed in C hapter 8 . O ther acoustic cues also contribute to the voiced-voiceless distinction in stop consonants, particularly when the stop consonant appears in other phonetic environments such as between two vowels, e.g. [beikig] baking vs. [beigiq] begging, or in syllable-final position, e.g. [beik] bake vs. [beig] beg. These include differences in F x transition, duration o f the closure interval, and duration o f the preceding vowel (cf. Lisker, 1978).

Acoustic correlates o f speech sounds

Nasal consonants The nasal consonants o f English are [m n i] ]. Similar to stop consonants, nasal consonants are produced with a closure in the supralaryngeal oral cavity. However, in contrast to stop consonants, the velum is open. Sound is propagated through the nose during the oral occlusion o f a nasal consonant, and it is propagated through both the nose and mouth as the oral occlusion is released. There are several acoustic consequences o f opening the nasal cavity in the production o f a nasal consonant. A nasal “ m u rm u r” occurs prior to the release o f the closure. This m urm ur has a spectrum dom inated by a low frequency prominence usually around 250 Hz. In addition, the m urm ur has additional resonances above 700 Hz which have lower amplitudes. A nother change in the spectrum is the presence of anti formants or zeros. Zeros selectively absorb acoustic energy, thus reducing the amplit ude components at or near the antiresonant frequency (cf. C hapter 7 for further discussion). The nasal m u rm ur is not the same for nasal consonants varying in place of articulation (Fujimura, 1962). However, studies exploring the acoustic cues for nasal consonants have shown that the presence o f a low frequency bro a d ­ band m u rm ur around 250 Hz preceding the onset o f the formant transitions is sufficient for the perception o f a nasal consonant irrespective o f place of articulation (Cooper et a i , 1952). The onset frequencies and direction o f the formant transitions cue place of articulation in a m anner similar to stop consonants. Thus, if a nasal m urm ur were synthesized preceding the onset o f the form ant transitions replacing the burst shown in Figure 10.2A, then the synthesized syllables will be perceived as [ma], [na] and the non-English syllable [13a] respectively. Although the fine details do differ, it is the case that the onset frequencies and direction o f the form ant transitions are simil ;11 for “ hom organic” nasals and slops, i.e. nasals and stops that share the same place of articulation.

Liquids and glides The sounds [1 r] are called liquids, and the sounds [w y] are called glides. All of them are produced with a partial constriction in the vocal tract. Research on the acoustic cues for the perception o f liquids and glides has shown that the rate, i.e. the duration, o f the form ant transitions provides the essential cue for these speech sounds (O 'C o n n o r et «/., 1957). In particular, if the formant transitions are about 40 milliseconds or longer, listeners will perceive a glide. If the transition durations are shorter, listeners may perceive stop consonants (Liberman et al., 1956). Perception o f glides and liquids are enhanced if the form ant transitions remain at their onset form ant frequencies for some 30 226

Acoustic correlates o f speech sounds milliseconds o r so before the transitions begin moving toward the steady-state formant frequencies o f the vowel. The perception o f the particular liquid o r glide, i.e. [1 r w y], depends on the onset frequencies and direction o f the form ant transitions, particularly o f the sccond and higher formants. For example, if the formant transitions are rising relative to the frequency o f the vowel steady-state, the listener will perceive a [w]. In contrast, if F 2 is falling and F3 is steady, listeners will perceive [1].

Fricatives The acoustic cues for fricative consonants are quite complex and will only be briefly summarized. In English, the fricativc consonants include [f 0 s s v S z z]. Experiments using synthetic speech have shown that the primary acoustic cue to the fricative manner o f articulation, irrespective of place of articulation and voicing, is the presence o f aperiodic noise in the spectrum (Delattre, Liberman and C ooper 1962). The duration o f this noise should be at least 20 milliseconds (Jongman, 1986). In natural speech, however, the duration of the fricative noise is considerably longer, of the order of 100 milliseconds. The onset o f the noise must be fairly gradual. If it is too abrupt, the stimulus will be perceived as an affricate or a stop (Gerstman, 1957; Cutting and Rosner, 1974). The overall amplitude o f the noise and the distribution of spectral peaks contribute to the perception of the different places o f articulation for fricative consonants. F o r example, both [s s] have overall greater amplitude than [f 6] (Strevens, I960; Heinz and Stevens, 1961). Listeners’ perception of fricative place of articulation is affected by varying the overall amplitude of the frication noise relative to that o f the vowel. The same stimulus is perceived as [s] when the amplitude of the high frequency noise is greater than that of the vowel, but it is perceived as [0] when the amplitude o f the noise is lower than that o f the vowel (Stevens, 1985). Moreover, if the major frequency peak o f the noise is lowered from around 5 kHz to aro un d 2.5 kHz, listeners’ perception will change from [s] to [s] (Delattre et al., 1964; Stevens, 1985).

Exercises 1.

R e v ie w th e a c o u s tic c u e s f o r th e p e r c e p tio n o f v o w e ls . W h a t a r c th e y ? C o m p a r e th e ro le o f f o r m a n t tr a n s i ti o n s a n d v o w e ls.

2.

H o w d o th e a c o u s tic c u e s f o r s to p c o n s o n a n t s d iffe r f ro m t h o s e f o r n a s a l c o n s o n a n t s ? D o th e y s h a r e a n y a t t r ib u t e s ?

3.

Is t h e r e o n e a c o u s tic c u e w h ic h c o r r e s p o n d s to o n e p h o n e tic s e g m e n t? D is c u s s .

227

Bibliography

A bbs. J. H. (1986) In variance and variability in spcech prod u ctio n : A distinction between linguistic in ten t an d its n e u ro m o to r im plem entation. In J. S. Perkell and D. H. K latt (eds.). Invariance and variability in speech processes. H illsdale, N ew Jersey, E ribaum A b ram so n , A. an d Lisker, L. (1970) D iscrim inability along the voicing co ntinuum : C ro ss­ language tesls. Proceedings o f the S ix th International Congress o f P honetic Sciences. Prague: A cadem ia A rm stro n g , L. E. and W ard, I. C. (1926) H andbook o f English intonation. Leipzig and Berlin: B. G. T eu b n er A ssm an n , P. F. (1979) T he role o f context in vowel perception. M a ster's thesis, U niversity o f A lb erta, C an ad a A tal, B. S. an d H an au e r, S. L. (1971) Speech analysis and synthesis by linear prediction o f the speech wave. Journal o f the A coustical Society o f A m erica 50, 637-55 A tk in so n , J. R. (1973) A spects o f in to n atio n in speech: Im plications from an experim ental study o f fu n d am en tal frequency. U npublished P h.D . d issertatio n . U niversity o f C onnecticut, S torrs B aker, E.. Blum stein, S. E. an d G o o d g lass, H. (1981) In terac tio n betw een phonological and sem antic facto rs in a u d ito ry co m prehension. N europsychologia 19, 1-16 Basso, A., C asati, G . an d V ignolo, L. A. (1977) Phonem ic identification defects in ap h asia, C ortex 1 3 .8 4 -9 5 Bell, A. M . (1867) Visible speech or self-interpreting physiological letters fo r the writing o f all languages in one alphabet. L ondon: Sim pkin and M arshall Bell, C. G ., F ujisaki, H ., H einz, J. M ., Stevens, K. N. an d H ouse, A. S. (1961) R eduction o f speech sp ectra by analysis-by-synthesis techniques. Journal o f the A coustical Society o f A m erica 33, 1725-36 Bell-Berti, F. (1973) T he velopharyngeal m echanism : An electryom yographic study. U npublished P h.D . thesis. C ity U niversity o f N ew Y ork: also S tatu s R e p o rt on Speech R esearch, S upplem ent, H ask ins L ab o rato ries, 270 C row n Street, N ew H aven, C onn. 06510 B eranek, L. L. (1949) A coustics. N ew Y ork: M cG raw -H ill B ertoncini, J., Bijeljac-Babic, R., Blum stein, S. and M ehler, J. (1987) D iscrim ination in neonates o f very sh o rt CVs, Journal o f the A coustical S ociety o f A m erica 82, 31-7 B ladon, R. an d L indblom B. (1981) M odeling the judgem ent o f vowel q u ality differences. Journal o f the A coustical Society o f A m erica 69, 1414 22 Bloom field, L. (1933) Language. New Y ork: H olt B lum stein, S. E. (1973) A phonological investigation o f aphasic speech. T he H ague: M o u to n B lum stein, S. E. and Stevens. K. N. (1979) A coustic invariance in speech pro d u ctio n : Evidence from m easurem ents o f the spectral characteristics o f slo p co n so n an ts. Journal o f the A coustical So ciety o f A m erica 66, 1001 -17 B lum stein, S. E. and Stevens, K. N. (1980) P erceptual invariance and onset spectra for stop c o n so n a n ts in different vowel en v iro n m en ts. Journal o f the A coustical Society o f A m erica 67, 648 62

228

Bibliography Blum stein, S. E. an d Stevens, K. N . (1981) P honetic features and acoustic invariance in spcech. Cognition 10, 25 32 Blum stein, S. E., Baker, E. and G oodglass, E. (1977a) P honological factors in au d ito ry co m p reh en sio n in ap h asia, N europsychologia 15, 19 36 Blum stein, S. E.. C o o p er, W. E., Z urif, E. and C a ram azz a, A. (1977b) T he perception and p ro d u ctio n o f voice-onset tim e in ap h asia, Neuropsychologia, 15, 371-83 Blum stein, S. E., C o o p e r, W. E., G o o d g lass, H ., S tatlender, S. an d G o ttlieb , J. (1980) P ro d u c tio n deficits in aphasia: A voice-onset tim e analysis. Brain and Language 9, 153—70 Blum stein, S. E., T a rtte r, V. C ., N igro, G . an d S tatlen d er, S. (1984) A coustic cues for the p ercep tio n o f a rticu latio n in ap h asia, Brain and Language 22, 128-49 Bogert, C. M . (1960) T he influence o f so u n d on the beh av io r o f am p h ib ian s an d reptiles, in W. E. L anyon an d W. N. T avoglga (eds.) A nim al sound and com m unication, A m erican Institute o f Biological Sciences, W ashington D C Bolt, R. H ., C o o p e r, F. S., D avid, E. E. Jr, D enes, P. B., P ickett, J. M . and Stevens, K. N. (1973) S peaker identification by speech spectrogram s: Som e fu rth er observations, Journal o f the A coustical S o ciety o f A m erica 54, 531-4 Bond, Z. S. (1976) Identification o f vowels excerpted from neutral nasal contexts, Journal o f the A coustical S o ciety o f A m erica 59, 1229 -32 Bosm a, J. F. (1957) D eglutition: P haryngeal stage, Physiological Review 37, 275-300 B ouhuys, A. (1974) Breathing. N ew Y ork: G ru n e and S tra tto n B ouhuys, A ., P ro c to r, D. F. and M ead, J. (1966) K inetic aspect o f singing. Journal o f A pplied Physiology 21, 483 -96 B radshaw , J. L. an d N ettleto n , N . C. (19 8 1) T he n atu re o f hem ispheric specialization in m an. The Behavioral a nd Brain Sciences 4, 51-92 Broca, P. ( 18 6 1) N ouvelle ob serv atio n d ’aphem ie p ro d u ite p a r une lesion de la m otie posterieure des deuxiem e et troisiem e circonvolutions frontales. Bulletin de la Societe d'A natom ique. Paris 6 (series 2), 398 407 Bryden, P. (1982) Laterality: Functional a sym m etry in the intact brain. New Y ork: A cadem ic Press C ap ran ica, R. R. (1965) The evoked vocal response o f the bullfrog. C am bridge, M ass.: M IT Press C h ap in , C ., T seng, C . Y. an d L ieberm an, P. (1982) S h o rt-term release cues for sto p c o n so n a n t place o f a rticu latio n in child speech. Journal o f the A coustical S ociety o f A m erica 71, 179 86 C hiba, T. an d K ajiy am a, J. (1941) The vowel: Its nature a n d structure. T okyo: T o k y o -K aiseik an P u blishing C o. C hom sky, N . (1957) S yntactic structures. T he H ague: M o u to n C hom sky, N. an d H alle, M . (1968) The sound pattern o f English. N ew Y ork: H a rp e r an d R ow C lark, H. H. an d C lark , E. V. (1977) P sychology and language. N ew Y ork: H a rc o u rt Brace Jovanovich Cole, R., R udnicky, A. I.,Z u e , V. W. and R eddy, D. J. (1980) Speech as pattern s on pap er. In R. A. C ole (ed.). Perception and production o f flu e n t speech. N ew Y ork: L aw rence E rlbaum C ollier, R. (1975) Physiological correlates o f in to n atio n p attern s, Journal o f the A coustical So ciety o f A m erica 58, 249-55 C ooper, W. an d C o o p e r, J. P. (1980) S y n ta x a n d speech. C am bridge, M ass.: H arv ard U niversity Press C o o p er, W. an d S o ren son J. M. (1981) Fundam ental fre q u en cy in sentence production. N ew Y ork: S pringer C ooper, F. S., D elattre, P. C., L iberm an, A. M ., B orst, J. M . and G e rstm an , L. J. (1952) Som e experim ents on th e perception o f synthetic speech so u n d s. Journal o f the A coustical S ociety o f Am erica 24, 597-606 Crelin, E. S. (1969) A n a to m y o f the newborn: An atlas. P hiladelphia: Lea and Fcbiger Crelin, E. S. (1973) Functional anatom y o f the newborn. New H aven: Y ale U niversity Press

229

Bibliography C u llin g , J. E. (1972) Plucks and bow s arc categorically perceived, som etim es, Perception and Psychophysics 31. 462 76 C u ttin g , J. E. (1974) T w o left hem isphere m echanism s in speech perception, Perception and P sychophysics 16. 601 12 C u ttin g . J. E. and Eim as, P. D. (1975) P honetic analysers an d processing o f speech in infants. In ). F. K a v a n a g h a n d J. E. C u ttin g (eds.). The role o f speech in language. C am bridge. M ass.: M IT Press C u ttin g , J. and R osncr, B. (1974) C ategories and b o u n d aries in speech and m usic. Perception and Psychophysics 16, 564 70 DanilofT, R. an d M oll, K. (1968) C o a rticu latio n o f lip ro u n d in g . Journal of Speech and Hearing Research I I , 707-21 D arw in, C. (1859) On the origin o f species. Facsim ile ed. 1964. C am bridge, M ass.: H arvard U niversity Press D eacon, T. W. (1985) C o n n e ctio n s o f the inferior p eriarc u ate area in the brain o f M acaco fascicularis: An experim ental an d co m p arativ e n cu ro an ato m ical investigation o f language circuitry an d its evolution. P h.D . d issertatio n . H arv ard U niversity D ejours, P. (1963) C o n tro l o f respiration by arterial chem oreceptors, in Regulation o f Respiration, Annals of The New York A cadem y o f Sciences 109, 682 95 D elattre, P. C. (1968) D ivergences cn tre la nasalilc vocaliquc et co n so n an tiq u e en fran