Vowel Information Research Articles

Lateralization effects in reading pointed and unpointed Hebrew.

Hebrew has two forms of spelling, pointed and unpointed. In the pointed spelling, diacritical signs (pointing) are added to consonantal letters to convey vowel information. These are omitted in the unpointed spelling. Since pointing conveys information that is critical for the prelexical derivation of phonology, it was hypothesized that its absence would prove detrimental for left hemisphere (LH) but not for right hemisphere (RH) reading and that, for the former, pointing effects would increase with increasing word length. Three experiments, one involving lexical decision and two involving word pronunciation, yielded little support for these hypotheses; rather, pointing had an overall adverse effect on performance, and this effect tended to be more pronounced for LH reading. In general, however, the results indicated an LH advantage. Since for central vision pointing has been found to aid performance under similar conditions, the results were seen to suggest a distinction between the visual and the phonological effects of pointing: pointing may impair early stages of visual analysis but may aid in the derivation of speech codes.

Lexical access for low- and high-frequency words in Hebrew.

The hypothesis that phonological mediation is involved to a greater extent in the recognition of low- than in the recognition of high-frequency words was examined using Hebrew. Hebrew has two forms of spelling, pointed and unpointed, which differ greatly in the extent of phonological ambiguity, with the unpointed spelling lacking almost all vowel information. A lexical decision task was employed using target words that had only one pronunciation whether pointed or unpointed. Targets were either pointed or unpointed and were preceded by a prime, which for word targets, was either semantically related or unrelated. The results indicated the following: First, the advantage of pointed over unpointed spelling was larger for low-frequency than for high-frequency words, suggesting a stronger phonological mediation for low-frequency words. Second, the size of the pointing effect was independent of word length, suggesting that phonology is obtained on the basis of the printed word as a whole, by looking it up in a phonological lexicon. Third, response latency to nonwords was not affected by the presence or absence of pointing, suggesting that failure to locate the entry corresponding to a letter string in a phonological lexicon results in a NO decision. Fourth, presence of a related prime was not found to compensate for absence of pointing, suggesting that the activation of a word’s representation in the semantic lexicon does not aid access to its corresponding entry in the phonological lexicon.

Identifying vowels in CVC syllables: effects of inserting silence and noise.

Listeners were asked to identify natural vowels in /d_d/ context under various deletion conditions. Deletion intervals, which were centered about the syllable midpoint, ranged from 60% to 90% of the syllable duration and contained either silence or broadband noise. In one condition, the three syllable types were /did/, /ded/, and /dud/; in the other condition, the three syllable types were /did/, /ded/, and/dAdV. Identification performance in the 60% and 70% deletion conditions was not substantially worse than for full syllables. Even the 90% deletion conditions yielded performance well above chance, indicating that significant coarticulatory information for vowels is contained in the first and last 10 or 15 msec of the syllable. For the short-vowel stimuli (/did/, /d∈d/, and /dλV) in the 90% deletion condition, a series of discriminant analyses were performed to assess the relative contribution of several acoustic variables to the statistical separation of the vowel categories. Several different combinations of acoustic variables (including formant frequencies at particular temporal locations and formant frequency differences over time) were sufficient to yield significant separation of the three vowel categories. However, in general, the performance of the discriminant classification program correlated only weakly with the identification performance of the listeners.

Vowel information is integrated across intervening nonlinguistic sounds

When the fricative noise of a fricative‐vowel syllable is replaced by a noise from a different vocalic context, listeners experience delays in identifying both the fricative and the vowel (D. H. Whalen, Perception & Psychophysics, 1984). Listeners (unconsciously) detect a mismatch between the vowel information in the fricative noise and in the vocalic segment. In the current experiment, noises and vowels were again cross‐spliced, but, in addition, the first 60 ms of the vocalic segment either had a nonlinguistic noise added to it or was replaced by that noise. The fricative noise and the majority of the vocalic segment were left intact, and both were quite identifiable. Mismatches of vowel information caused delays for all stimuli, both originals and ones with the noise. Additionally, syllables with a portion replaced by noise took longer to identify than those that had the noise added to them. The results indicate that listeners integrate all relevant information even across a nonlinguistic noise. Similarly, having the signal present along with the noise delayed identifications less than replacing the signal completely. [Work supported by NIH.]

Speech coding in the auditory nerve: II. Processing schemes for vowel-like sounds.

Several processing schemes by which phonetically important information for vowels can be extracted from responses of auditory-nerve fibers are analyzed. The schemes are based on power spectra of period histograms obtained in response to a set of nine two-formant, steady-state, vowel-like stimuli presented at 60 and 75 dB SPL. One class of "local filtering" schemes, which was originally proposed by Young and Sachs [J. Acoust. Soc. Am. 66, 1381-1403 (1979)], consists of analyzing response patterns by filters centered at the characteristic frequencies (CF) of the fibers, so that a tonotopically arranged measure of synchronized response can be obtained. Various schemes in this class differ in the characteristics of the filter. For a wide range of filter bandwidths, formant frequencies correspond approximately to the CFs for which the response measure is maximal. If in addition, the bandwidths of the analyzing filters are made compatible with psychophysical measures of frequency selectivity, low-frequency harmonics of the stimulus fundamental are resolved in the output profile, so that fundamental frequency can also be estimated. In a second class of processing schemes, a dominant response component is defined for each fiber from a 1/6 octave spectral representation of the response pattern, and the formant frequencies are estimated from the most frequent values of the dominant component in the ensemble of auditory-nerve fibers. The local filtering schemes and the dominant component schemes can be related to "place" and "periodicity" models of auditory processing, respectively.

Subcategorical phonetic mismatches slow phonetic judgments.

When an [s] or an [s] fricative noise is combined with vocalic formant transitions appropriate to a different fricative, the resulting consonantal percept is usually that of the noise. To see if the mismatch affects processing time, five experiments were run. Three experiments examined reaction time for identification of [s] and [s], as well as the whole syllable (in one experiment) or only the vowel (in the others). The stimuli contained either appropriate or inappropriate formant transitions, and the vowel information in the noise was either appropriate or not. Subjects were significantly slower in all tasks in identifying stimuli with inappropriate transitions or inappropriate vowel information. Similar results were obtained with stop-vowel syllables in which the release bursts of syllable initials [p] and [k]were transposed in syllables containing the vowels [a] and [u], In the fifth experiment, enough silence was introduced between the initial fricatives and vocalic segment for the vocalic formant transitions to be perceived as a stop (e.g., [stu] from [su]). Mismatched transitions still had an effect on reaction time, as did mismatches of vowel quality. The results indicate that listeners take into account all available cues, even when the phonetic judgment seems to be based on only some of the cues.

Vowel information in postvocalic fricative noises.

When the postvocalic fricative noises of [s] and [f] are excerpted and combined with vocalic segments having inappropriate formant transitions, vowel quality, or both, the fricative percept is determined by the noise. However, there is often a perception of a diphthong in the vowel. This phenomenon was explored for the vowels [a, i, o, u] preceding the fricatives [s] and [∫]. In the first of two experiments, all combinations of the vocalic segments and fricative noises were presented for identification of the vowel. The perception of diphthongs occurred much more often on mismatches of vowel quality than of transition, indicating that there is substantial vowel information in the noise. In the second experiment, just the noises of the syllables were presented, with subjects trying to identify the missing vowel. The high vowels [i] and [u] were reliably identified, while identifications of [a] and [o] were at chance. This result agrees with previous studies of initial fricatives (Yeni-Komshian and Soli, 1981). Fricative noises from [i] and [u] were responsible for the large majority of diphthong percepts in Experiment 1. These results illustrate that fricative noises contain considerable information about preceding high vowels.

Identifying vowels in CVC syllables: Effects of inserting silence and noise

Listeners were asked to identify natural vowels in /d_d/ context under various deletion conditions. Deletion intervals ranged from 60% to 90% of the syllable duration (measured from initial burst to final closure onset) and were centered about the midpoint. The deleted syllable portions were replaced with either silence or broadband noise. In one experiment, the three syllable types used were /did/, /ded/, and /dud/; in the other experiment, the three syllable types were /dΙd/, /dεd/, and /dΛd/. To ensure that identification performance was based solely on spectral information, tokens were selected such that, within an experiment, average syllable duration was approximately equal across types. Identification performance in the 60% and 70% deletion conditions was not substantially worse than for full syllables [cf., W. Strange, J. J. Jenkins, and T. R. Edman, J. Acoust. Soc. Am. Suppl. 1 61, S39 (1977)]. Even the 90% deletion conditions yielded performance well above chance, indicating that significant vowel information is contained in the first and last 10 or 15 ms of the syllable. Results will be discussed in terms of residual acoustic cues.

Perceiving vowels in isolation and in consonantal context.

Recent studies have shown that vowels tend to be identified more accurately in consonantal context than in isolation. This contextual advantage is often explained perceptually, e.g., by assuming that the formant transitions associated with the consonants convey significant vowel information. In two experiments with stylized synthetic speech patterns, we were unable to replicate the contextual advantage. These negative results were probably due to certain unnatural stimulus characteristics. In another experiment we used natural speech stimuli to assess whether nonperceptual factors associated with the identification task contribute to the contextual advantage. Subjects responded to the test items either by (a) circling written CVC syllables, (b) circling written isolated vowels, or (c) vocally mimicking the items (a task that we assume imposes minimal memory load on subjects). Of these response conditions, only the first yielded an advantage for vowels in context, suggesting that the effect depends on two factors: Stimulus-response compatibility and memory load.

Identification of stops and vowels in truncated CV syllables

How is the information specifying place of articulation and vowel color encoded in a stop-vowel syllable? Are place cues located in just the burst, or the first 20 ms of a CV waveform as some investigators have suggested? Alternatively, are place cues encoded simultaneously (i.e., in parallel) with vowel information in the formant transitions? To answer these questions, an identification experiment was conducted using the consonants /b, d, g/ paired with the vowels /i, e, a, o, u/ produced by two male talkers. Using computer techniques, five truncated stops were edited from each original syllable. Cuts were made after the release burst, and then the burst plus 1, 3, 5, or 7 pitch pulses. On different days, naive subjects identified either the consonant or the vowel. Identification functions for /b/ and /d/ reached 95% correct within the first 20 ms. Vowels, except for /i/, were identified near chance for the shortest stimuli. The consonant /g/ needed a longer duration, 40 to 50 ms, for correct place identification, although /gi/ was never identified accurately even at the longest durations. We conclude that sufficient acoustic information for identifying the stops /b/ and /d/ can be found within the first 20 ms of a CV syllable, although information for the /g/ extends over somewhat longer durations. Formant transitions appear to carry information about the vowel although somewhat less information about the consonant, except for /g/ before front vowels. [Supported by NINCDS Grant NS-12179.]

Contextual effects in vowel perception II: evidence for two processing mechanisms.

Recent experiments have indicated that contrast effects can be obtained with vowels by anchoring a test series with one of the endpoint vowels. These contextual effects cannot be attributed to feature detector fatigue or to the induction of an overt response bias. In the present studies, anchored ABX discrimination functions and signal detection analyses of identification data (before and after anchoring) for an [i]-[I] vowel series were used to demonstrate that [i] and [I] anchoring produce contrast effects by affecting different per­ ceptual mechanisms. The effects of [i] anchoring were to increase within-[i] category sensitiv­ ity, while [I] anchoring shifted criterion placements. When vowels were placed in CVC syllables to reduce available auditory memory, there was a significant decrease in the size 'of the [I]-anchor contrast effects. The magnitude of the [i]-anchor effect was unaffected by the reduction in vowel information available in auditory memory. These results suggest that [i] and [I] anchors affect mechanisms at different levels of processing. The [i] anchoring results may reflect normalization processes in speech perception that operate at an early level of perceptual processing, while the [I] anchoring results represent changes in response criterion mediated by auditory memory for vowel information.

Talker‐independent information for vowel identity

A male and a female talker each produced a set of /bVb/ syllables (V = /i,I,e,ε,æ,a,ʌ,ɔ,o,ʋ,u/) at metronome‐controlled rates. The central 60% of the voiced regions of these syllables was edited out, producing single‐talker silent‐center syllables [cf. W. Strange, J. J. Jenkins, and T. R. Edman, J. Acoust. Soc. Am. 61, S39 (1977)]. Comparable “hybrid” silent‐center syllables were created by appending the final 20% of one talker's syllable to the initial 20% of the other talker's production of the same syllable, again with a 60% silent region intervening. Vowel identification errors were relatively low for both the single‐talker (22.4%) and the hybrid (27.0%) silent‐center syllables, though errors in each case exceeded those found for the original (unedited) syllables, though errors in each case exceeded those found for the original (unedited) syllables (9.3%). Hybrids were generally heard as coherent syllables from a single source; the error pattern for the hybrids could not be interpreted as a joint function of errors on the initial pieces (55.4% on average) and the final pieces (72.9%). Results indicate that considerable information for vowels is contained in the dynamic regions early and late in a syllable. Moreover, they suggest that this information does not simply subserve computation of an underlying normalized target. The information may specify properties of production dynamics that are invarient for a vowel for all talkers of a dialect. To the extent that such information is available, normalization in perception is unnecessary. [Research supported by NICHD.]

Some perceptual dependencies between vowel color and pitch

How to acoustic dimensions combine in speech perception? Wood [C. C. Wood, J.E.P.: HP & P, 104, 3–20 (1975)] suggested that information specifying consonants is dependent on the earlier processing of pitch because pitch judgments are unaffected by consonant variation whereas consonant judgments are slower when pitch varies. Is vowel information also dependent on the prior processing of pitch? In contrast to the results obtained with consonants, Kuhl [P. K. Kuhl and J. D. Miller, J. Acoust. Soc. Am. Suppl. 1, 58 (1975) and P. K. Kuhl, J. Acoust. Soc. Am. Suppl. 1, 60 (1976)] reported that infants respond selectively to vowel color when pitch varies but not to pitch when vowel varies. In this study we examined adults' judgments of vowel color and pitch in a reaction time task. Subjects classified stimuli as /i/ or /ɛ/, or as high or low pitch. Consistent with Kuhl's results, vowel judgments were unaffected by pitch variation whereas pitch judgments were slower when the vowel varied. However, this result was obtained only when the pitch difference was small (15 Hz). With a larger pitch difference (75 Hz), no interference between pitch and vowel was observed. The asymetric interference found in the first case, then, was likely due to the greater discriminability of vowel color over pitch. Thus, conclusions about perceptual dependencies in speech perception must be made cautiously as the effects may be due to the discriminability of the specific values on the dimensions under study. [Supported by NIMH.]

Context-conditioned specification of vowel identity

Consonantal environment does not aid vowel identification universally, but preserves and sharpens the information for vowels differentially. In a vowel recognition task, listeners heard one of two test series: /pVp/ syllables or isolated vowels. The test items were spoken by a single male talker and contained a balanced distribution of the vowels /i, ɪ, ε, æ, a, ɔ, ʌ, ᴜ, u/. During each block of 90 test trials, listeners targeted for one of the nine vowel types, checking “Yes” when they recognized an instance and “No” otherwise. For both misses and false alarms, there was a strong interaction between vowel type (close versus open) and context type (/pVp/ vs /V/). The open vowel pairs (/ɛ/æ/ , /ʌ/a/) showed consistently superior recognition in the /pVp/ consonantal environment; others did not. Asymmetries in confusion errors were observed for several vowel pairs in both environments. A dynamic articulatory model of the syllable can provide a parsimonious account for these effects. Listeners may be sensitive to acoustic patterning that specifies vowels as dynamic gestures; vowel recognition will vary in precision depending on how the consonantal envirionment structures the dynamic information in sound. [Research supported by NIH grant HD-01994 to Haskins Laboratories.]

Effects of altering vowel feedback on speech movements

Delayed auditory feedback, masking and movement tracking techniques have been used to investigate the role of auditory feedback in speech production. The results of these experiments are not conclusive and questions still remain as to whether auditory feedback plays a role, when it plays a role and how it plays a role in speech production. In the present study it was hypothesized that if acoustic information concerning the articulatory positions relevant to vowel production is used in speech production, then altering vowel information would disrupt the movements to subsequent sounds in a C1V1C2V2 utterance. High‐speed cineradiography (150 fps) was used to monitor movements of the jaw, lips, and tongue. During their production of C1V1C2V2 utterances subjects received (1) normal auditory feedback; (2) feedback in which V1 was altered to another vowel; (3) high‐level masking noise. Displacements, velocities, and accelerations were calculated for articulatory movements to V1, C2, and V2 in each condition. The...

Discriminability, response bias, and phoneme categories in voice‐onset‐time discrimination

Considerable theoretical importance has been attached to the “phoneme boundary effect,” in which discrimination across a boundary between phoneme categories is superior to discrimination of comparable stimulus differences with a phoneme category. In order to asses the relative contribution of increases in discriminability and changes in response bias to the phoneme boundary effect, a “same‐different” task compared discrimination of 300‐msec synthetic syllables varying in voice‐onset time (VOT) from −50 to 70 msec. Both a marked increase in discriminability and an equally marked shift in response bias toward “different” responses occurred at VOT values near the phoneme boundary. A second experiment investigated the role of phonetic categorization per se in the phoneme boundary effect. Variations in VOT were isolated from syllable context by eliminating the final 180 msec of steady‐state vowel information from each stimulus. An increase in discriminability comparable to that of the first experiment was obtained for isolated VOT values corresponding to the phoneme boundary. Together with the recent data of Miller et al. [J. Acoust. Soc. Am. 55, 390 (1974)], these results suggest that the phoneme boundary effect for VOT does not depend exclusively upon phonetic categorization but may reflect a more general property of auditory temporal sensitivity.

Perception of vowel and syllable duration in VC and CVC syllables

Results of previous experiments [Raphael et al., “Vowel information conveyed by consonant transitions,” J. Acoust. Soc. Am. 57, Suppl. No. 1, S49(A) (1975)] indicate that perceptual cues for initial consonants are, in part, incorporated by listeners into their estimates of vowel/syllable duration when they make voicing judgments of word‐final consonants. In the present experiment the inventory of initial consonants tested was expanded to include those containing voiced transitions of varying durations, both contiguous and noncontiguous with those of the following vowel; aperiodic transitions; and transitionless noise. Results indicate that listeners incorporate more of the duration of initial consonant cues when they are conveyed by formants than when they are not; when the formants are periodic rather than aperiodic; and when the formants are contiguous with those of the following vowel than when they are not. The degree of incorporation was also found to vary directly with the duration of the formant transitions.

Vowel information conveyed by consonant transitions

It is well established that information specifying place and manner of articulation for stop consonants is conveyed by the transitions of the vocalic segment. The purpose of this study was to determine whether these transitions also convey information about vocalic duration. Accordingly, four series of synthetic stimuli were generated: three of the series consisted of CVC syllables in which the initial C transitions were set at durations of 30, 60, and 90 msec; a fourth series consisted of VC syllables. In all four series the duration of the steady-state portion of the vowel was varied from 10 to 190 msec in 10-msec steps. The final consonant was synthesized with transitions appropriate for an unreleased /d/. Listeners were asked to identify the final consonant as voiced or voiceless (/t/ or /d/). For vowel steady states of equal duration, the addition of initial consonant transitions caused a shift in the perceptual boundary between the voiced and voiceless categories. This outcome suggests how a portion of the initial “consonant” transition is used by the listener in computation of “vowel” duration, at least for the purpose of cueing the voicing distinction in final position.

“Same-Different” Reaction Times to Consonants, Vowels, and Syllables

It has been proposed that phonetic segments are encoded into perceptual units of syllable size. The aim of this study is to investigate how consonant and vowel information is transmitted (i.e., serial vs parallel), and to what extent auditory and phonetic levels of analysis may interact in perceptual analysis. Synthetic CV syllables, /ba/, /bæ/, /da/, and /dæ/, were arranged to produce four types of stimulus pairs: (1) consonant same-vowel same (CS-VS, e.g., /ba/-/ba/); (2) consonant same-vowel different (CS-VD, e.g., /ba/-/bæ/); (3) consonant different-vowel same (CD-VS, e.g., /ba/-/da/); and (4) consonant different-vowel different (CD-VD, e.g., /ba/-/dæ/). Same-different reaction times (RTs) were obtained separately for consonants, vowels, and syllables. In the consonant and vowel conditions, RTs for “same” responses were faster when the irrelevant feature was the same (i.e., CS-VS); “different” response RTs were faster when the irrelevant feature was different (CD-VD). In the syllable condition, RTs for different responses were faster when the vowel was different and the consonant was the same (i.e., CS-VD) than when the consonant was different and the vowel was the same (i.e., CD-VS). This suggests that vowel information may be more easily accessible for a comparison than consonant information. These results provide evidence for the parallel transmission of information about stop consonants and vowels in syllables, and an interaction of auditory and phonetic levels of processing in speech perception.

Some Perceptual Mechanisms in Speech Processing

Enough is known about the acoustical cues in speech to enable highly intelligible synthesis from simple algorithms, but, in most cases, the most valid expressions of cues from the point of view of perception and the mechanism of detection, weighting, and decision are not known. Experiments that manipulate several cues and the phonetic context offer some hope of charting information flow through the human perceptual process and providing detailed block diagram models. Several examples are given. Perception of voicing in initial stops depends upon the place of articulation as well as the traditional voicing cues such as VOT—i.e., the boundary VOT value differs with place. Two separate experiments show that this effect is carried not by acoustical place information but by a perceptual decision about place, logically prior to the: voicing decision. But, in another context effect, dependence of consonant place upon the adjacent vowel, the effect appears conditioned by the acoustical vowel information and not by the decision as to linguistic identity. Implications for perceptual theory and speech recognition are pointed out. [Work supported by J.S.R.U.—U.K.]