Band 57, Heft 2-4

This paper looks at patterns of reduction and elaboration in speech production, taking the phenomenon of plosive-related glottalization in German spontaneous speech, on the basis of the ‘Kiel Corpus’, as its point of departure, and proposes general principles of human speech to explain them. This is followed by an enquiry into the nature of a production-perception link, based on complementary data from perceptual experiments. A hypothesis is put forward as to how listeners cope with the enormous phonetic variability of spoken language and how this ability may be acquired. Finally, the need for a new paradigm of phonetic analysis and phonological systematization is stressed, as a prerequisite to dealing adequately and in an insightful way with the production and perception of spontaneous speech.

Emotional expressivity in singing is examined by comparing neutral and expressive performances of a set of music excerpts as performed by a professional baritone singer. Both the neutral and the expressive versions showed considerable deviations from the nominal description represented by the score. Much of these differences can be accounted for in terms of the application of two basic principles, grouping, i.e. marking of the hierarchical structure, and differentiation, i.e. enhancing the differences between tone categories. The expressive versions differed from the neutral versions with respect to a number of acoustic characteristics. In the expressive versions, the structure and the tone category differences were marked more clearly. Furthermore, the singer emphasized semantically important words in the lyrics in the expressive versions. Comparing the means used by the singer for the purpose of emphasis with those used by a professional actor and voice coach revealed striking similarities.

One aim of our study is to discuss some of the relations of prosodic parameters to the speech production process, e.g. how speech intensity is related to the vocal tract filter and the voice source and the underlying aerodynamics. A specific problem of phonetic interest is the role of subglottal pressure and fundamental frequency as intensity determinants and their covariation in speech. Our speech analysis displays, incorporating perceptually scaled syllable prominence, are suitable for multilevel studies of speech parameters. A new intensity parameter, SPLH, related to sonority is introduced. In combination with the standard sound pressure level it provides information on the voice source spectral slope. In Swedish, high long stressed vowels approach a semi-closed target, and thus a sonority minimum, which suggests a motor component in prominence perception.

The Converter and Distributor (C/D) model is a generative description of articulatory gesture organization for utterances. Its input comprises specifications for syllables by features, a paraphonologically augmented metrical structure, and system parameters for utterance conditions. A syllable-boundary pulse train is computed as a time function representing the skeletal rhythmic structure of the utterance. Control functions for articulators are computed by superimposing consonantal elemental gestures onto the base function, which includes voicing, vocalic, mandibular, and tonal functions associated with the pulse train. A preliminary analysis of microbeam data for experimental dialogues, using jaw opening as an approximate measure of the syllable magnitude, inferred syllable and boundary durations consistently with durational characteristics observed in the acoustic waveform.

The consonantal segments that underlie an utterance are manifested in the acoustic signal by abrupt discontinuities or dislocations in the spectral pattern. There are potentially two such discontinuities for each consonant, corresponding to the formation and release of a constriction in the oral cavity by the lips, the tongue blade, or the tongue body. Acoustic cues for the various consonant features of place, voicing and nasality reside in the signal in quite different forms on the two sides of each acoustic discontinuity. Examples of these diverse cues and their origin in acoustic theory are reviewed, with special attention to place features and features related to the laryngeal state and to nasalization. A listener appears to have the ability to integrate these diverse, brief acoustic cues for the features of consonants, although the mechanism for this integration process is unclear.

The phenomenon of vowel reduction is investigated by modeling ‘gesture reduction’ with the use of the Distinctive Region Model (DRM). First, a definition is proposed for the term gesture, i.e. an acoustically efficient command aimed at deforming, in the time domain, the area function of the vocal tract. Second, tests are reported on the perception of vowel-to-vowel transitions obtained with reduced gestures. These tests show that a dual representation of formant transitions is required to explain the reduction phenomenon: the trajectory in the F₁–F₂ plane and the time course of the formant changes. The results also suggest that time-domain integration of the trajectories constitutes an integral part of the auditory processing of transitions. Perceptual results are also discussed in terms of the acoustic traces of DRM gestures.

The ‘hyperspace effect’ in vowel perception may be taken as evidence that adaptive dispersion is an active perceptual process. However, a previous study tested for adaptive dispersion in isolated vowel stimuli spoken in a voice unfamiliar to the listeners. The experiment reported in this paper addressed both of these potential concerns and found that both consonant context and talker familiarity modulate the hyperspace effect. However, the reductions induced by context and familiarity were slight. Listeners’ preferred perceptual spaces remained hyperarticulated relative to the production vowel space.

Purported units of speech, e.g. phonemes or features, are essentially categories. The assignment of phonemic (or phonetic) identity is a process of categorization: potentially discriminable speech sounds are treated in an equivalent manner. Unfortunately the extensive literature on human categorization has typically focused on simple visual categories that are defined by the presence or absence of discrete features. Speech categories are much more complex. They are often defined by continuous values across a variety of imperfectly valid features. In this paper, several kinds of categories are distinguished and studies using human subjects, animal subjects and computational models are presented that endeavor to describe the structure and development of the sort of complex categories underlying speech perception.

Rarely, animals do what they do because they are aware of the function of the behaviour or its outcome. Instead, they will very often perform behaviour out of context, spontaneously, as play. The impression (strengthened by introspection in the human species) is that they do it because they get some kind of internal reward. Nevertheless, such seemingly meaningless behaviour may have an ultimate function to adjust behavioural programs to the body, to practice, to perfect the execution of the behaviour. If the proximate reason for doing what the animal does may be to attain a pleasurable state, the ultimate, evolutionary reason may still be that increased practice will give some gain in fitness. If one presupposes internal rewarding and punishing systems as intervening factors, it becomes much simpler to explain why birds sing, kittens play or babies babble without any outer reward and out of any functional context, more than needed from a strictly functional view, spontaneously and just for the fun of it.

For more than a century it has been noted that the adult human vocal tract differs from that of other mammals, in that the resting position of the larynx is much lower in humans. While animals habitually breathe with the larynx inserted into the nasal cavity, adult humans are unable to do this. This anatomical difference has been cited as an important factor limiting the vocal potential of nonhuman animals, because the low larynx of humans allows a wider range of vocal tract shapes and thus formant patterns than is available to other species. However, it is not clear that the static anatomy of dead animals provides an accurate guide to the phonetic potential of the living animal’s vocal tract. Here I present X-ray video observations of four mammal species (dogs Canis familiaris, goats Capra hircus, pigs Sus scrofa and cotton-top tamarins Sagunius oedipus ). In all four species, the larynx was lowered from the nasopharynx, and the velum was closed, during loud calls. In dogs this temporary lowering was particularly pronounced. Although preliminary, these results suggest that the nonhuman vocal tract is more flexible than previously supposed, and that static postmortem anatomy provides an incomplete guide to the phonetic potential of nonhuman animals. The implications of these findings for theories of speech evolution are discussed.

A three-dimensional model of the body was used to simulate two different motor tasks: vertical jumping and normal walking on level ground. The pattern of muscle excitations, body motions, and ground-reaction forces for each task were calculated using dynamic optimization theory. For jumping, the performance criterion was to maximize the height reached by the center of mass of the body; for walking, the measure of performance was metabolic energy consumed per meter walked. Quantitative comparisons of the simulation results with experimental data obtained from people indicate that the model reproduces the salient features of maximum-height jumping and normal walking on the level. Analyses of the model solutions will allow detailed explanations to be given about the actions of specific muscles during each of these tasks.

Understanding the potential relationships between perception and production is crucial to explanation of the nature of early speech acquisition. The ‘embodiment’ perspective suggests that mental activity in general cannot be understood outside of the context of body activities. Indeed, universal motor factors seem to be more responsible for the distribution of early production preferences regarding consonant place and manner, and use of the vowel space than the often considerable cross-language differences in input available to the perceptual system. However, there is evidence for a perceptual basis to the establishment of a language-appropriate balance of oral-to-nasal output by the beginning of babbling, illustrating the necessary contribution of ‘extrinsic’ perceptual information to acquisition. In terms of representations, at least one assumption that segmental units underlying either perception or production in early phases of acquisition may be inappropriate. Our work on production has shown that the dominant early organizational structure is a relatively unitary open-close ‘frame’ produced by mandibular oscillation. Consideration of the role of ‘intrinsic’ (self-produced) perceptual information suggests that this frame may be an important basis for perceptual as well as production organization.

The intelligibility of a word in continuous speech depends on the clarity of the word and on linguistic and nonlinguistic contextual information available to the listener. Despite limited knowledge of language and the world, infants in the first 2 years are already beginning to make use of contextual information in processing speech. Adults interacting with infants tend to modify their speech in ways that serve to maximize predictability for the immature listener by highlighting focussed words and using frequent repetition and formulaic utterances. Infant-directed speech is viewed as a form of ‘hyperspeech’ which facilities comprehension, not by modifying phonetic properties of individual words but rather by providing contextual support on perceptual levels accessible to infants even in the earliest stages of language learning.

Although it is generally accepted that phonological development is grounded in phonetic learning, there is less agreement for the proposition supported here, that the first phonological structuring constitutes a developmental discontinuity. Data from the phonetic and lexical learning of Finnish consonant duration are presented to illustrate the role of (1) child selection of adult words for early context-supported production based on phonetic learning and (2) child adaptation of adult words to an idiosyncratic template for later production as part of an incipient system. We argue that the latter, but not the former, reflects the construction of a first phonology.

This paper examines three auditory hypotheses concerning the location of category boundaries among vowel sounds. The first hypothesis claims that category boundaries tend to occur in a region corresponding to a 3-Bark separation between adjacent spectral peaks. According to the second hypothesis, vowel category boundaries are determined by the combined effects of the Bark distances between adjacent spectral peaks but that the weight of each of these effects is inversely related to the individual sizes of the Bark distances. In a series of perceptual experiments, each of these hypotheses was found to account for some category boundaries in American English but not others. The third hypothesis, which has received preliminary support from studies in our laboratory and elsewhere, claims that listeners partition the vowel space of individual talkers along lines corresponding to relatively simple linear functions of formant values when scaled in auditorily motivated units of frequency such as Bark.

The paper argues that the discrete phonetic segments on which language is raised are subjective gestural structures that emerge ontogenetically (and perhaps emerged evolutionarily) from the process of imitating a quasi-continuous acoustic signal with a neuroanatomically segmented and somatotopically organized vocal machinery. Evidence cited for somatotopic organization includes the perceptual salience in the speech signal of information specifying place of articulation, as revealed both by sine wave speech and by the pattern of errors in children’s early words.Copyright © 2000 S. Karger AG, Basel‘Almost every insight gained by modern linguistics, from Grimm’s Law to Jakobson’s distinctive features, depends crucially on the assumption that speech is a sequence of discrete entities.’Morris Halle [1964, p. 325]

A comparison of babbling and early speech, word patterns of languages, and, in one instance, a protolanguage corpus, reveals three basic movement patterns: (1) a ‘Frame’ provided by the cycles of mandibular oscillation underlying the basic mouth close-open alternation of speech; this Frame appears in relatively ‘pure’ form in the tendency for labial consonants to co-occur with central vowels; (2) two other intracyclical consonant-vowel (CV) co-occurrence patterns sharing the alternation: coronal consonants with front vowels and dorsal consonants with back vowels; (3) an intercyclical tendency towards a labial consonant-vowel-coronal consonant (LC) sequence preference for word initiation. The first two patterns were derived from oral movement capabilities which predated speech. The Frame (1) may have evolved from ingestive cyclicities (e.g. chewing). The intracyclical consonant-vowel (CV) co-occurrence patterns involving tongue position constraints common to consonants and vowels (2) may result from the basic biomechanical property of inertia. The third pattern (LC) was a self-organizational result of pressures for interfacing cognition with action – a result which must have numerous analogs in other domains of movement organization.

In this paper fragments of a theory of emergent phonology are presented. Phonological patterns are seen as products of cultural evolution adapted to universal biological constraints on listening, speaking and learning. It is proposed that children develop adult phonology thanks to the interaction of the emergent patterning of phonetic content and the adaptive organization of sound structure. Emergence – here used in the technical sense of qualitatively new development – is illustrated with examples from the study of perception, motor mechanisms and memory encoding. In this framework, there is no split between ‘behavioral phonetics’ and ‘abstract phonology’. Phonology differs qualitatively from phonetics in that it represents a new, more complex and higher level of organization of speech behavior. Accordingly, the phonology that the child ends up with as an adult is neither abstract nor independent of use. It represents an emergent patterning of phonetic content.