Search

Prepausal postvocalic stops in English are reported to occur both with and without audible release bursts, more or less randomly, and this difference is said to be without distinctive function. However, there is evidence that an English final stop, absent its release, may be of reduced intelligibility, particularly as to its place of articulation. Without audible release a final stop’s place is conveyed mainly, perhaps entirely, by frequency shifts in the vowel formants. Among the vowels of English some are diphthongal, a property also signaled by formant shifts. The question arises: Is the intelligibility of an unreleased stop significantly affected by the phonetic nature of the vowel? Perceptual testing of appropriately chosen nonsense monosyllables ending in [p? t? k?] indicates that these stops are generally somewhat less intelligible after diphthongs. However, not all three stops are affected equally, [k?] perception being especially reduced in this context.

This study represents an addition to the literature describing the role of the larynx in the production of voiced and voiceless stop and fricative consonants. Electromyographic recordings from the intrinsic laryngeal musculature and measurements of sub- and supraglottal air pressures were obtained from a speaker of Standard Dutch, who produced nonsense forms preceded by a short carrier Dutch phrase. The forms included intervocalic voiced and voiceless stops and fricatives, as well as certain combinations of these consonants. The data obtained are in general conformity with previous studies of larynx management in consonant voicing, indicating systematic differences for voiced vs voiceless and for stop vs fricative categories. The different EMG patterns suggest that the voicing dimension involves primarily the varying adjustment of static glottal width by means of the adductor-abductor muscles, while the stop-fricative difference involves both this variable and a feature of longitudinal vocal fold tensioning. The evidence is negative so far as providing support for a view that this latter feature plays a significant role in the voicing distinction.

The voiced/voiceless distinction for English utterance‐initial stop consonants is primarily realized as differences in the voice onset time (VOT), which is largely signaled by the time between the stop burst and the onset of voicing. The voicing of stops has also been shown to affect the vowel’s F0 after release, with voiceless stops being associated with higher F0. When the VOT is ambiguous, these F0 ‘‘perturbations’’ have been shown to affect voicing judgments. This is to be expected of what can be considered a redundant feature, that is, that it should carry a distinction in cases where the primary feature is neutralized. However, when the voicing judgments were made as quickly as possible, an inappropriate F0 was found to slow response time even for unambiguous VOTs. This was true both of F0 contours and level F0 differences. These results reinforce the plausibility of tonogenesis, and they add further weight to the claim that listeners make full use of the signal given to them, even when overt labeling would seem to indicate otherwise.

The post-stop-release rise or fall of fundamental frequency (F0) is known to affect voicing judgments of syllables with ambiguous voice onset times (VOTs). In 1986, Silverman claimed that the critical factor was not direction of F0 change but rather its direction relative to the intonational contour. He further claimed that only F0s that start above and fall to the contour have an effect proportional to the size of the frequency change; F0s that rise to the contour by different amounts were claimed to be equivalent. In our first experiment, we examined the effect on voicing judgments of five onset F0s preceding a single, flat contour. Only falling F0s were differentiated in the first set of judgments, but after increased exposure to the syllables, even F0s below the contour differentially affected the voicing judgment. In a second experiment, the contour of the final part of the syllable was flat, rising or falling. F0 contour affected the judgments, as did onset F0s, but the two factors did not interact, indicating that the onset values were not being judged by reference to the contours. However, the contour which was predicted to result in more voiceless judgments also ended at a higher F0 in the vowel, and another effect of voicing is that the F0 is higher throughout the vowel after voiceless stops. In a third experiment, F0 contours were created to contrast contour and mean F0. The effect of the F0 during the vocalic segment appeared to be attributable to the average F0 rather than the contour. In all three experiments, the F0 onset values contributed to the voicing judgment whether they were above or below the putative intonation contour. The contribution of the lower F0s, while significant, was not as great as that of the higher F0s, which argues for a noncategorical contribution of intonation.

We have shown in earlier work that speakers of Spanish differentiate voiced /bdg/ from voiceless /ptk/ in word-initial position by means of voice onset time (VOT), the temporal relation between the onset of glottal pulsing and acoustic features of supraglottal articulation. More recently, we demonstrated the perceptual efficacy of VOT for the three categories of Thai and the two of English that lie along the timing dimension. The two Spanish categories differ from those of English in that Spanish voiced stops are produced with VOT values that lead the consonant release, while voiceless stops show VOT upon release or immediately thereafter. In the present study, native speakers of Latin American Spanish identified stops synthesized with VOT varying in small steps. Their responses showed a fairly good fit between production and perception, differing from English in the expected direction. Some listeners also discriminated the variants in a psychoacoustic test format. The latter results, along with the earlier ones for English and Thai, suggest that discriminability is largely determined by language experience, although some subjects reveal considerable sensitivity to changes in the acoustic signal at some remove from the phonemic boundary.

Preliminary data on four acoustic variables as possible cues to the contrast between voiced and voiceless unaspirated stops in fluent American English speech were reported at the 119th Meeting. They were burst amplitude, duration of the preclosure vowel, duration of the closure interval, and the nature of the closure signal. A single English speaker recorded the sentence pair He's known as Jack the Ribber and He's known as Jack the Ripper, which served as sources of waveform‐edited stimuli that were identified by a jury of ten American English speakers. Additional word identification data have since been collected for three additional sentence pairs recorded by the same single speaker, with the new word targets chosen to include dorsal as well as labial stops after both the short vowel [&smcapi;] and the long vowel [o]. For each acoustic variable two values were tested, one normal for the word as originally produced and the other typical of the minimally contrasting word. The data allow us to rank the four acoustic properties in respect to their roles in perception: burst amplitude < vowel duration  closure duration < closure signal. [Work supported by NIH Grant HD‐01994 to Haskins Labs.]

Durational differences between voiced and voiceless stops have been regularly reported, and phonological value has been attributed to some of them, i.e., VOT, closure duration (CD), and voicing‐gap duration (VGD). Thus English /p,t,k/ and /b,d,g/ can sometimes be separated by all three measures, and these are then acoustico‐articulatory properties of the voicing feature. But VOT also separates subphonemic (allophonic) variants of /p,t,k/ aspirated [ph, th, kh] and unaspirated [p,t,k], and it is customary to regard this difference as phonologically irrelevant. Furthermore, CD cues voicing state only when accompanied by a voicing gap, when a short VGD is effectively (perceptually?) equivalent to one of zero duration, so the voicing gap may be ‘‘dismissed’’ as a consequence of pressure equalization rather than a linguistically motivated laryngeal maneuver. Elsewhere the value of CD is more prosodic (moraic) than segment‐identifying, distinguishing, e.g., Weplaydominoes. from Weplayeddominoes. Moreover, some VOT, CD, and VGD differences have little linguistic or even perception function. Thus /p,t,k/ show durational differences, but these cannot readily be shown to affect stop place perception. [Work supported by NIH Grant No. HD‐01994.]

At the last ASA meeting it was shown that for a set of synthetic vowel–glide–vowel sequences varying in F2 trajectories a range of frequencies can equally well serve as mid segments in patterns identified as English /iwi/ and /uyu/. Many of these mid segments, when lengthed and presented in isolation, resemble the high front rounded vowel [y], so that when short and intervocalic one might expect them to be heard as the glide [inverted h], at least by phonetically trained listeners. However, perceptual data suggest that intended [inverted h] in /i/—/i/ is heard as /w/, while in /u/—/u/ it is /y/. Assuming the syllabifications /i$wi/ and /u$yu/, one might then expect an initial [inverted h] to be heard as /w/ in [inverted hi] and as /y/ in [inverted hu]. Test data indicate otherwise. Unlike /iwi/ and /uyu/, /wi/ and /yu/ show no overlap of F2 values in their initial steady‐state segments—there is instead a range of F2 values for which listeners report both not‐/wi/ and not‐/yu/. Thus for English /w,y/ the steady‐state and transitional intervals have different perceptual weights initially and medially. [Work supported by NIH Grant No. HD‐01994 to Haskins Laboratories.]

English /w/ and /y/ are often identified phonetically with /u/ and /i/: /w/ and /u/ involve dorso‐velar constriction and lip rounding, /y/ and /i/ raising of the tongue toward the palate without rounding. Although the glides are thus more or less equated with the vowels, most linguists keep /w, y/ apart from /u, i/. The justification is that /w, y/ are nonsyllabic ‘‘glides,’’ whereas /u, i/ are not. As a phonetic basis for keeping /w, y/ and /u, i/ apart, this is not very compelling. Aside from the fact that the articulatory postures for /w, y/ may not always be identical with those for /u, i/, the phonological argument advanced here is that /w/ differs from /u/ in being distinctively rounded, but not distinctively dorso‐velar, while /y/ is distinctively palatal, but not distinctively unrounded. In fact, the same articulatory configuration, that of a rounded palatal [inverted aitch], is differently interpreted in the contexts /i–i/ and /u–u/, so that [i(inverted aitch)i] is /iwi/, while [u(inverted aitch)u] is /uyu/. Thus the same vocal tract configuration, reflected in an identical F pattern, is differently interpreted in the two vocalic contexts. It would seem that the listener knows that it was achieved by two quite different articulatory maneuvers. [Work supported by NIH Grant HD‐01994 to Haskins Labs.]

Phonetic studies of distinctive consonantal voicing in English have focused on utterance‐initial and final positions. Much less attention has been given to medial stops following stressed and preceding unstressed vowels, as in pairs like tugging/tucking. The high frequency of occurrence of unaspirated voiceless stops in this context is commonly overlooked because of the usual emphasis on the aspiration of/p t k/in stressed syllables. The voiced stops in this context are likely to have glottal pulsing through much or all of the closure, unlike their utterance‐initial counterparts, which may not show glottal pulsing until the moment of release. For this study, a number of such word pairs were recorded in carrier sentences at a normal rate by three native speakers of American English: Three acoustic features were measured: voice timing, closure duration, and amplitude of release burst. Values of the first two features were clearly different for the two categories. With respect to the third feature, burst amplitude, the two categories differed much less clearly. To generate sets of stimuli for perception testing, the three features were covaried incrementally by waveform editing. Labeling data collected from a jury of native American English listeners will be presented to show the relative efficacy of these features for perception.

A simple answer to the question is this: A stop is aspirated when a linguist/phonetician judges it to be so. This property, often referred to in descriptions of the stop consonants of English and many other languages, has been variously defined in physical terms, aerodynamically, articulatorily, and acoustically. Linguists seem generally to agree that in English aspiration is a predictable, i.e., context determined, and hence a nondistinctive attribute of/p t k/in certain contexts. In some other languages, e.g., Hindi, aspiration is a distinctive feature of certain stop phonemes, so that [t] contrasts with [th]. Hindi‐English bilinguals, however, when speaking English, regularly use Hindi /t/ and not/th/, where a “predictably” aspirated English/t/would be expected. For native English speakers, including linguists, this choice seems erroneous. Consistent with this choice by Hindi speakers, is the general opinion of many Hindi‐speaking linguists that, contrary to the opinion of American linguists, English /p t k/are not aspirated. Labeling tests of edited naturally produced nonsense CV syllables yielded data showing that Indian and American linguists differ considerably in locating an inaspirate‐aspirate boundary along the VOT dimension.