Search

During the course of a normal conversation talkers make frequent and extensive changes in speaking rate which affects the acoustic realization of both consonants and vowels. Previous research has shown that the articulation rate of a syllable influences both the production and phonetic categorization of its segmental components. However, little certainty exists regarding which portions of the syllable carry information about overall articulation rate. The present study investigated this question by editing the burst and aspiration phases from natural /bi/ and /pi/ syllables produced at a fast rate of speech, and cross-splicing them onto the same tokens produced at slow rates of speech (and vice versa). These tokens were presented to subjects for speeded classification of the initial consonant. The results indicate that subjects were sensitive to the mismatch in articulation rate between the initial burst and aspiration of the stop consonant and the remainder of the syllable created by the cross-splicing technique. Further investigation using /bi-pi/ continua created in a similar manner indicates that consonant-internal rate information is used during phonetic categorization of the stimuli. These results are discussed in the context of current issues of rate normalization and speech perception. [Work supported in part by Research and Training Center Grant No. P60 DC-01409 from the National Institute on Deafness and other Communication Disorders.] [ABSTRACT FROM AUTHOR]

When CV syllables have longer vowels following a stop /b/ or glide /w/, the duration of transition necessary to perceive a stimulus as /w/ is longer as well. Perceived continuity/duration of following vowels was manipulated by exploiting the phenomenon of auditory induction. All syllables were 315 ms in duration, and transition duration was varied from 15-65 ms for four series of syllables. After the first 100 ms of each syllable (transition + vowel), broadband noise either replaced or was added to the second 100 ms of syllable. Noise was either the same intensity as or 12 dB greater than the intensity of the vowel. Results from a signal detection task indicated that listeners could reliably report when the signal was deleted and replaced by the less intense noise; however, they could not reliably report whether the vowel was deleted when the noise was more intense (+12 dB). Listeners were asked to label these four series of stimuli as beginning with /b/ or /w/, and there was no effect of either the presence or absence of the vowel or of noise intensity. Even when subjects's phenomenal experience was one of intact continuous syllables (+12-dB noise with vowel present or absent, or equal intensity noise with vowel present), subjects' labeling boundaries were virtually identical to those for syllables perceived as interrupted (equal intensity noise with vowel absent) or for syllables of only 100-ms total duration. Theoretical implications and results from follow-up experiments will be reported. [Work supported by NIDCD Grant No. DC-00719 and NSF Grant No. DBS-9258482.] [ABSTRACT FROM AUTHOR]

A number of studies have indicated that a dependency relation exists for the processing of place and manner features of articulation. Other studies have demonstrated that visual information provided by a talker's face can influence the perceived place of articulation of well-specified auditory tokens presented in synchrony with the visual information. Thus, when /bi/ is presented auditorially and /gi/ visually, a person perceives the talker as saying /di/. These findings raise the possibility that visual place information might also influence the processing of VOT. Members of an auditory continuum ranging from /ibi/ to /ipi/ were paired with a video display of a talker saying /igi/. The auditory tokens were heard as ranging from /ibi/ to /ipi/, while the audio/visual tokens were perceived as ranging from /idi/ to /iti/. We found that the VOT boundary for the audio/visual tokens was located at a significantly longer value than the VOT boundary for the auditory continuum presented without the visual information. These results indicate that, under certain circumstances, the processing of VOT information takes into account both auditory and visual place information. [Research supported by NIH.] [ABSTRACT FROM AUTHOR]

It is by now well-established that the listener makes very precise adjustments for speaking rate when perceiving the phonetic structure of speech. The concern of the present investigation was the nature of the rate information to which the listener adjusts. When a speaker changes rate, he or she alters the timing of articulatory gestures. One consequence is a change in the acoustic form of the speech signal-for example, a change in its overall duration. Such changes provide auditory information about speaking rate. Additionally, however, some of these articulatory gestures are visible on the face of the talker and, therefore, provide visual information about the talker's rate of speech. We asked whether the effective rate information for phonetic perception is limited to the auditory domain or, alternatively, whether visual information about rate from the face also influences the perception of speech. To investigate this question, we paired acoustic versions of /bi/ and /pi/ produced by a talker, at a medium rate of speech, with a video display of the same talker producing these syllables at fast and slow rates of speech. The phenomenal experience that resulted from any given audio-visual pairing was that of a unified percept of a single syllable, produced at a single rate of speech. We found that the variation in the speaking rate of the syllable on the video display (with the rate of the acoustic syllable kept constant) systematically influenced not only the judged rate of the syllable, but also the identification of the syllable as /bi/ or /pi/. These results suggest that at least under some circumstances, auditory and visual information about articulatory timing jointly influence phonetic perception. [Research supported by NIH.] [ABSTRACT FROM AUTHOR]

It is now well established that speaking rate exerts a systematic influence on a variety of temporal properties that convey segmental information, and that the listener appropriately adjusts for speaking rate when using these properties to make phonetic decisions. The concern of the present investigation is the nature of the rate information to which the listener adjusts, whether physical rate or perceived rate. Our experiments focused on the distinction between stop consonant and semivowel in the syllables /ba/ and /wa/. We have reported previously that when listeners use transition duration to identify syllables as /ba/ or /wa/, they do so in relation to the overall duration of the syllables: as the syllables become longer and are perceived as having been produced at a slower rate of speech, a longer transition is required to convert /ba/ to /wa/. In the experiments we now report, we investigated whether the adjustment for rate is based on the actual physical rate of the syllables (i.e., syllable duration) or on their perceived rate. We used a contrast procedure to alter the perceived rate of the syllables while keeping physical rate constant, and then assessed the effect of this manipulation on the identification of the syllables as /ba/ or /wa/. We found that changes in perceived rate induced by the contrast procedure had no reliable effect on syllable identification. However, in a control study, changes in physical rate that produced equivalent changes in perceived rate did reliably alter identification of the syllables. Phonetic identification is thus influenced by the physical rate of the speech, not its perceived rate. [ABSTRACT FROM AUTHOR]

Two factors contributing to the identification of a word are the prosody of the sentence in which the word occurs and the semantic relation between the word and its context sentence. To investigate further these two factors, we created a series of target words that ranged from 'bath' to 'path' by systematically varying word-initial voice-onset-time, placed these words in a number of different sentence frames, and asked listeners to identify each word as 'bath' or 'path.' Both the speaking rate of the sentence and the semantic congruity between the target word and the sentence affected the way in which acoustically ambiguous words were identified. However, the effects were quite different in nature. Whereas a change in speaking rate influenced word identification even when the listener was not required by task demands to attend to the context sentence, a change in semantic congruity altered identification only when attention to the context sentence was required. This suggests that in the course of identifying a word the processing system automatically takes into account the prosodic structure of the utterance, but not its semantic content. The implication of these findings for models of word recognition will be discussed. [ABSTRACT FROM AUTHOR]

Investigating speech rhythm by examining an utterance's syllabic-beat organization seems natural, but is problematic. An utterance must first be divided into syllables, then each syllable's beat must be located through mathematical modeling or the collection of subjects' judgments. This study approaches the problem in a more straightforward way by examining the organization of singularities, specifically acoustic envelope amplitude and amplitude rate-of-change maxima, in the acoustic speech wave. These acoustic features are the basis for some acoustic models of syllabic beat location, and so are plausible candidates for more directly measurable correlates of speech rhythm. So far, qualitative analyses suggest that these acoustic features are organized in a quasiperiodic manner which is neither strictly periodicity nor random and that is similar to the organization found in the behaviors of other physical and biological systems. Quantitative analyses will be discussed that further specify the organization of the data. [Work supported by National Multipurpose Research and Training Center Grant DC-01409 from the National Institute on Deafness and Other Communication Disorders.] [ABSTRACT FROM AUTHOR]

Seeing a speaking face can influence observers' auditory perception of syllables [McGurk and McDonald, Nature 264, 746-748 (1976)]. This effect decreases when the speaker's face is inverted [e.g., Green, 3014 (1994)]. Face recognition is also inhibited with inverted faces [e.g., Rock, Sci. Am. 230, 78-85 (1974)] suggesting a similar underlying process. To further explore the link between face and audiovisual speech perception, a speech experiment was designed to replicate another face perception effect. In this effect, an inverted face and an inverted face containing upright lips are perceived as looking normal, but an upright face with inverted lips looks grotesque [Thompson, Perception 9, 438-484 (1980)]. An audiovisual speech experiment tested four presentation conditions: Upright face-upright mouth, upright face-inverted mouth, inverted face-inverted mouth, inverted face-upright mouth. Various discrepant audiovisual syllables were tested in each condition. Visual influences occurred in all but the upright face-inverted mouth condition for some of the syllable combinations thereby mimicking the face perception effect. However, other syllable combinations revealed visual influences in all four conditions. Results are interpreted in terms of articulatory dynamics and the vertical symmetry of the visual stimuli. [ABSTRACT FROM AUTHOR]

One of the obstacles to investigating speech rhythm has been the difficulty in locating the syllabic beat. This study attempts to address that difficulty by using perceptual centers (p-centers) as an index of speech rhythm. P-centers of syllables extracted from natural utterances were determined both by listeners using a method of adjustment procedure and by an acoustic p-center model. The phonetic structure (syllabic onset) and stress patterns of the syllables in the utterances were varied, and the effects of these manipulations on the utterances' rhythms were investigated. As expected, the p-centers of the syllables varied systematically with their own phonetic structure. Preliminary findings indicate that the p-center of the syllables also changes along with the p-center of the previous or subsequent syllable so as to maintain a relatively constant interval between the two p-centers. The study also examines the effects on rhythm of altering stress patterns, by determining whether unstressed syllables might influence p-centers. The results will be discussed in terms of dynamic constraints which might affect speech production. [Work supported in part by National Multipurpose Research and Training Center Grant No. P60 DC-01409 from the National Institute on Deafness and Other Communication Disorders.] [ABSTRACT FROM AUTHOR]

The McGurk effect is a situation in which the perception of the speech signal is influenced by watching a talker's face articulate a conflicting speech utterance. For example, observers typically report the syllable /da/ when the auditory syllable /ba/ is presented in synchrony with a videotape of the talker saying the syllable /ga/. It is well-known that inverted faces are much more difficult to recognize than faces presented in a normal, upright orientation. The current study investigated whether inverting the talker's face would also influence the McGurk effect. Two groups of subjects were presented with a videotape consisting of conflicting auditory and visual syllables designed to produce the typical McGurk effect. The first group viewed the test tape on a video monitor set in its normal upright position. This group showed a typical, robust McGurk effect. The second group viewed the videotape on the monitor set in an inverted position. This group showed a significantly weaker McGurk effect. Thus inverting the face and the corresponding articulatory movements impacts on the integration of phonetic information from the auditory and visual modalities. [Work supported by NIH.] [ABSTRACT FROM AUTHOR]

Listeners perceptually adjust for the acoustic variability brought about by differences in talker (voice) and rate information. To account for this adjustment, researchers have posited the existence of normalization processes. The purpose of the experiments reported here was twofold. First, the nature of the relationship between phonetic coding processes and talker and rate normalization processes was examined using monosyllables varying in three dimensions [phonetic (/bi/, /pi/); talker (male, female); rate (fast, slow)] in a speeded classification procedure [W. R. Garner, The Processing of Information and Structure (Erlbaum, Potomac, MD, 1974)]. Subjects classified the syllables along each of the three stimulus dimensions under two testing conditions: when the syllables varied only in the relevant/attended target dimension (single dimension) and when the syllables varied along the relevant and an irrelevant/unattended target dimension (orthogonal). Results indicated that decisions about phonetic identity were dependent upon both talker and rate information: Irrelevant variation in either dimension interfered with (increased reaction time to) phonetic classification. While rate classification was also affected by phonetic variation (integral), talker classification was not (separable). A final experiment, using the same stimuli and reaction time procedure, explored the relationship between the two normalization processes. Results indicated an integral relationship: Variation in talker information interfered with rate classification and vice versa. [ABSTRACT FROM AUTHOR]

In the 'McGurk' effect, observers typically report the illusory syllable /da/ when they hear the auditory syllable /ba/ presented in synchrony with a video display of a talker saying /ga/. In such experiments, there is usually congruence between the two modalities in that the same talker produces both the auditory and the visual signals. In the experiments reported here, the effect of reducing the congruence between the two modalities on the magnitude of the McGurk effect was examined. This was accomplished by dubbing a male talker's voice onto a video tape containing a female talker's face, and a female talker's voice onto a video tape containing a male talker's voice. These 'cross-dubbed' video tapes were compared to normal video tapes in which the male talker's voice was dubbed onto a male talker's face, and the female talker's voice was dubbed onto a female talker's face. The results show that even though there was clear incompatibility in the talker characteristics between the auditory and visual signals for the cross-dubbed stimuli, there was little difference in the magnitude of the effect compared to the normal stimuli. These results indicate that the mechanism for integrating speech information from the two modalities is not sensitive to certain incompatibilities, even when they are perceptually apparent. [Work supported by NIH.] [ABSTRACT FROM AUTHOR]

Experiments comparing isolated word recognition by human listners with automatic speech recognition systems are valuable because error analyses may lead to improvements in speech recognition technology. Isolated word recognition in adult human listeners has been compared with recognition performance by two commercially available speech-recognition systems. The test stimuli were drawn from the Lincoln Laboratory Stressed-Speech database. The database consists of 6930 stimuli (two iterations of each of 35 words spoken by nine different people in 11 different speaking styles). The vocabulary contains confusable words (i.e., go, hello, oh, no, and zero); the speaking styles include a wide range of naturally occurring variations (i.e., normal, slow, fast, soft, loud, angry). Analyses show that the acoustic characteristics of individual words vary considerably across talkers, and across styles within talkers. Performance of human listeners and the two machine-based recognition systems was tested in a single-talker, multistyle condition, and in a multitalker, multistyle condition. All tests were conducted under two listening conditions: normal, and in the presence of masking noise. The data to be presented are the error patterns of human listeners, versus the machine-recognition systems, exhibited across talkers, across speaking styles, and across training conditions (multitalker, multistyle training versus single talker, single style training). [Work supported by Boeing Aerospace and Electronics.] [ABSTRACT FROM AUTHOR]

Studies have demonstrated that speaking rate provides an important context for the perception of certain acoustic properties of speech. For example, syllable duration, which varies as a function of speaking rate, has been shown to influence the perception of voice-onset-time (VOT) for syllable-initial stop consonants. The purpose of the present experiments was to examine the influence of syllable duration when the initial portion of the syllable was produced by one talker, and the remainder of the syllable was produced by a different talker. A short duration and a long duration /bi-pi/ continuum was synthesized with pitch and formant values appropriate to a female talker. When presented to listeners for identification, these stimuli demonstrated the typical effect of syllable duration on the voicing boundary: a shorter VOT boundary for the short stimuli relative to the long. An /i/ vowel, synthesized with pitch and formant values appropriate to a male talker, was added to the end of each of the short tokens producing a new 'hybrid' continuum. Although the overall syllable durations of the hybrid stimuli equaled the original long stimuli, they produced a VOT boundary similar to the short stimuli. In a follow-up experiment, two new /i/ vowels were synthesized. One had a pitch appropriate to a female talker with formant values appropriate to a male talker, while the second had a pitch appropriate to a male talker and formants appropriate to a female talker. These vowels were used to create two new hybrid continua. The results of this second experiment indicated that continuity of pitch but not of formant structure appears to be the important factor in the integration of speaking rate within a syllable. [Work supported by NIH.] [ABSTRACT FROM AUTHOR]

In the McGurk effect, observers typically report the illusory syllable /da/ when they hear the auditory syllable /ba/ presented in synchrony with a video display of a talker saying /ga/. While the effect itself has been well established, there is still little research on the conditions under which the effect occurs. In the experiment reported here, the number of illusory /d/ responses to the auditory /b/-visual /g/ combination is examined in three vowel environments: /a/, /i/, and /u/. The results of this study indicate that the magnitude of the illusion is not the same across different vowel environments. It appears to be strongest for the /i/ vowel, moderate for /a/, and almost nonexistent for /u/. The results thus show that vowel environment is an important factor in determining the magnitude of the McGurk effect, which needs to be considered in accounts of auditory-visual integration during speech perception. [Work supported by NIH.] [ABSTRACT FROM AUTHOR]

Previous studies using speeded classification paradigms have been used to test whether the information that underlies the perception of place and voicing is processed in an integral as opposed to a separate fashion. Results of auditory only experiments show that reaction times (RTs) for place classification in a condition in which stimuli vary along both place and voicing (the orthogonal condition) are longer than RTs in a control condition in which stimuli vary only in place. These data are taken in support of the idea that the information underlying place and voicing axe integrally processed. This study was concerned with whether or not place classification would demonstrate the same pattern of results when the place feature was determined by both auditory and visual (the face of the talker) information. To do this, the auditory tokens /ibi/ and /ipi/ were paired with a video display of a talker saying /ibi/ or /igi/, and presented to subjects for speeded classification of place or voicing. Because of the 'McGurk effect,' these stimuli were perceived as /ibi, ipi, idi, or iti/. The results showed that RTs for classifying phonetic features in the orthogonal condition were significantly longer than in the control condition. These results were obtained for both place and voicing judgments. Thus auditory-visual results appear to mirror those obtained in previous studies using only auditory stimuli. [Work supported by NIH.] [ABSTRACT FROM AUTHOR]

The 'McGurk effect' is a phenomenon in which an illusory syllable is perceived when discrepant auditory and visual information are combined. For example, an illusory /da/ is perceived when an auditory syllable /ba/ is paired with a video display of a talker saying /ga/. In a series of studies reported here, it was found that the level of the auditory signal substantially affects the illusion. Moreover, this effect of auditory level works in a direction that is counterintuitive. One might have postulated that as the level of the auditory signal is increased, making the auditory signal more pronounced, the illusion would decrease. These results indicate just the opposite. As the level of the auditory signal goes up, rather than it outweighing the visual signal and thereby decreasing the illusion, the effect is to increase the number of illusory responses. Paired were the same auditory syllable /ba/ at three levels, soft (45 dB SPL), moderate (58 dB SPL), and loud (66 dB SPL), with a video display of /ga/. Significant increases in illusory /da/ responses occurred as the auditory signal increased from soft to moderate and again from moderate to loud. The work thus shows that auditory level is an important determinant in the perception of the illusion. The paradox of the situation is that as the auditory signal increases in strength, the resulting percept becomes less veridical rather than more so. [Work supported by NIH.] [ABSTRACT FROM AUTHOR]