Perceptual normalization for speaking rate. II: Effects of signal discontinuities.

In a series of experiments, we examined how rate normalization in speech perception is influenced by segments that occur after the target. Perception of the syllable-initial target was influenced by the durations of both the adjacent vowel and the segment after the vowel, even when the identity of the talker was changed during the syllable. These results, together with earlier findings of a temporal window that follows a target phoneme within which segment duration influences perception of the target, help to resolve apparently conflicting results that have been reported previously. Overall, the results fit within a theoretical framework in which the rate at which events take place is extracted early in processing, prior to segregating voices, and the use of this information is obligatory in subsequent processing.

Lexical neighborhood effects in phonetic processing.

Previous research on spoken word recognition has demonstrated that identification of a phonetic segment is affected by the lexical status of the item in which the segment occurs. W. F. Ganong (1980) demonstrated that a category boundary shift occurs when the voiced end of 1 voice-onset time continuum is a word but the voiceless end of another series is a word; this is known as the "lexical effect." A series of studies was undertaken to examine how lexical neighborhood; in contrast to lexical status, might influence word perception. Pairs of nonword series were created in which the voiced end of 1 series had a higher frequency-weighted neighborhood density, whereas the reverse was true for the other series. Lexical neighborhood was found to affect word recognition in much the same way as lexical status.

The processing of duration and intensity cues to prominence.

This paper presents results from two experiments designed to show how duration and intensity are processed during speech perception. Duration and intensity are two physical dimensions which are known to interact psychoacoustically in the perception of both length (a term that will be used for perceived duration) and loudness. The first experiment, a selective attention task, shows that length and loudness are processed as a unit [integrally, in the terms of Garner, The Processing of Information and Structure (Erlbaum, Potomac, MD, 1974)], but that the integrality is asymmetric: Extracting length information appears to be easier than extracting loudness information. The results of the first experiment make the prediction that listeners would not use loudness by itself in making prominence judgments, since the extraction of loudness in the presence of duration variation appears to require a (relatively) high processing load. The second experiment, a traditional trading relation experiment in which duration and intensity were varied orthogonally, appears to bear out this prediction. Listeners' responses were predicted from computed measures of length and loudness in a linear multiple regression analysis. Results show a negligible independent contribution of loudness to listeners' responses. Listeners' behavior is best predicted by computed measures of length.

Perceptual normalization for speaking rate: effects of temporal distance.

A series of studies was undertaken to examine how rate normalization in speech perception would be influenced by the similarity, duration, and phonotactics of phonemes that were adjacent or distal from the initial, target phoneme. The duration of the adjacent (following) phoneme always had an effect on perception of the initial target. Neither phonotactics nor acoustic similarity seemed to have any influence on this rate normalization effect. However, effects of the duration of the nonadjacent (distal) phoneme were only found when that phoneme was temporally close to the target. These results suggest that there is a temporal window over which rate normalization occurs. In most cases, only the adjacent phoneme or adjacent two phonemes will fall within this window and thus influence perception of a phoneme distinction.

Auditory coding, cues, and coherence in phonetic perception.

C. T. Best, M. Studdert-Kennedy, S. Manuel, and J. Rubin-Spitz (1989) reported that listeners given speech labels showed categorical-like perception of a series of complex tone analogs to a /la/-/ra/ speech series, whereas nonspeech listeners were unable to classify the stimuli consistently. In 2 experiments, a new training and testing procedure was used with adult listeners given nonspeech instructions. They classified the /la/-/ra/ tone analogs consistently, showed categorical-like perception, and generalized their training to a new, /li/-/ri/ tone analog series. Two sets of auditory attributes were described for coding the /l/-/r/ distinction, and 1 was shown to quantitatively predict listeners' classification of both series. These results are consistent with models of perception in which a rich, abstract auditory code is computed and forms the basis for both speech and nonspeech auditory categories.

Automaticity and the detection of speech.

The development of automatic perceptual responses to speech stimuli was examined. In the first experiment, phoneme-monitoring performance for speech syllables was examined under conditions in which stimulus-to-response mapping and memory load were manipulated. The results indicated that automaticity develops under consistent-mapping conditions. In the second experiment, a dual-task procedure was combined with mapping and selective attention manipulations in order to examine the development of automaticity across single- and multiple-channel conditions. The results indicated that performance under consistently mapped training conditions was interfered with by dividing attention across multiple channels of input. It is concluded that there may be differences in the way that automaticity develops across visual and auditory modalities and that these differences need to be examined more closely.

Integral processing of phonemes: evidence for a phonetic mode of perception.

To investigate the extent and locus of integral processing in speech perception, a speeded classification task was utilized with a set of noise-tone analogs of the fricative-vowel syllables (fae), (integral of ae), (fu), and (integral of u). Unlike the stimuli used in previous studies of selective perception of syllables, these stimuli did not contain consonant-vowel transitions. Subjects were asked to classify on the basis of one of the two syllable components. Some subjects were told that the stimuli were computer generated noise-tone sequences. These subjects processed the noise and tone separably. Irrelevant variation of the noise did not affect reaction times (RTs) for the classification of the tone, and vice versa. Other subjects were instructed to treat the stimuli as speech. For these subjects, irrelevant variation of the fricative increased RTs for the classification of the vowel, and vice versa. A second experiment employed naturally spoken fricative-vowel syllables with the same task. Classification RTs showed a pattern of integrality in that irrelevant variation of either component increased RTs to the other. These results indicate that knowledge of coarticulation (or its acoustic consequences) is a basic element of speech perception. Furthermore, the use of this knowledge in phonetic coding is mandatory, even in situations where the stimuli do not contain coarticulatory information.

When selective adaptation and contrast effects are distinct: a reply to Diehl, Kluender, and Parker.

Two alternative conceptualizations of selective adaptation with speech have recently received attention. The adaptation level theory (AL) outlined by Diehl (1981) and a two-stage model outlined by Sawusch and Jusczyk (1981) can both account for much of the adaptation and paired-comparison data. Recently, Diehl, Kluender, and Parker (1985) proposed that all adaptation and contrast data can be accounted for by AL theory. They reported the results of a study that showed evidence of streaming in selective adaptation and claimed that their results provide a counterdemonstration to recent studies that have argued against the AL approach. In the present article, an outline of how the Diehl et al. results can be accounted for by both the two-stage model and AL theory is presented. In addition, a new set of results comparing adaptation and paired-comparison procedures is presented. These results are precisely as predicted by the two-stage model, but they can not be handled by AL theory.

Dichotic training.

Children (aged 7-10 yrs) who were referred because of learning disabilities and found to have abnormal SSW scores were divided into Experimental and Control Groups. All had essentially normal pure-tone ac HTLs, tympanograms, and W-22 word discrimination scores. Pre-therapy baselines were also obtained on a speech-to-noise (S-in-N) task, a test of phonemic synthesis (PS) and a new procedure with staggered dichotic digits (SDD) patterned after the SSW. The therapy for the 5 Exper. Ss consisted of 15 lessons, given 1 hr each, 2 per week, using staggered dichotic digits [termed Digit Offset Therapy (DOT)]. The "competing" portion of the items ranged from complete overlap to 500-msec separation. The therapy program began with the most widely separated items first, followed by those with greater overlap in the succeeding lessons. Following DOT, the original test battery was readministered to both groups. During the therapy, significant improvement was noted for the Exper. Grp from Sessions 5 through 15 (less than .05 level) compared to the pre-therapy baseline. On the central tests, pre- and post-therapy, the Exper. Grp improved significantly over the Controls (.01 level) on the SDD test. There were also trends toward improvement for the Exper. Grp over the Controls on the SSW and S-in-N tests but neither reached statistical significance. Neither group improved in PS. Explanations for the test findings are discussed.

Adaptation and contrast in the perception of voicing.

The results of experiments using selective adaptation with stop consonants have been interpreted in terms of auditory feature detector fatigue, phonetic feature detector fatigue, and response contrast. In the present studies, both a selective adaptation procedure and a procedure involving paired comparisons between successively presented stimuli were used to sort out these explanations. A fricative-stop-vowel syllable ([spa]) was constructed using an [s], followed by 75 msec of silence, followed by a 10-msec voice onset time [ba]. The perceived phonetic identity of this syllable was [p] even though the spectral structure of the stop vowel within this syllable was identical to a stimulus from the [ba] end of a [ba]-[pha] test series. As adaptors, the [spa] and [ba] endpoint syllables had identical effects. In paired-comparison procedure, the [spa] caused an ambiguous test item to be labeled "B," whereas the [ba] caused the test item to be labeled "P." Results of these experiments indicate that neither response contrast nor phonetic feature detection are involved in selective adaptation effects found for a voicing stop-consonant series. Results are interpreted as supporting the position that selective adaptation effects arise at an early, auditory level of processing that is responsive to the spectral overlap between adaptor and test items.

Some relationships between speech production and perception.

EMG studies of the American English vowel pairs /i-I/ and /e-epsilon/ reveal two different production strategies: some speakers appear to differentiate the members of each pair primarily on the basis to tongue height; for others the basis of differentiation appears to be tongue tension. There was no obvious reflection of these differences in the speech wave-forms or formant patterns of the two groups. To determine if these differences in production might correspond to differences in perception, two vowel identification tests were given to the EMG subjects. Subjects were asked to label the members of a seven-step vowel continuum, /i/ through /I/. In one condition each item had an equal probability of occurrence. The other condition was an anchoring test; the first stimulus, /i/, was heard four times as often as any other stimulus. Compared with the equal-probability test labelling boundary, the boundary in the anchoring test was displaced toward the more frequently occurring stimulus. The magnitude of the shift of the labelling boundary was greater for subjects using a production strategy based on tongue height than for subjects using tongue tension to differentiate these vowels, suggesting that the stimuli represent adjacent categories in the speakers' phonetic space for the former, but not for the latter, group.