Search

The types of analyses required within a framework that uses dynamically-defined articulatory gestures as the primitive units of phonetic description are outlined. The overall approach is then exemplified using investigations into an overlap hypothesis of reduced vowels, specifically the hypothesis that the difference between the bisyllable “beret” and the monosyllable “bray” can be attributed to a difference in the overlap of the labial closure and tongue rhotic gestures. This overlap hypothesis is supported by a perceptual test of simulations of the utterances as well as by preliminary articulatory analyses of X-ray microbeam data, and leads to some possible overlap typologies for reduced syllables.

A quantitative characterization of articulatory movements, using the parameter values of a linear second‐order dynamical system, was developed in order to compare classes of movements, in particular, classes defined by linguistic factors such as syllable position, stress, and vowel quality. Movements of the lower lip in utterances such as [’bibEbib] and [babE’bab] were partitioned into sections (‘‘windows’’) in two ways: at successive displacement peaks and valleys, and at the right edge of plateau regions around such extreme values. The linguistic factors affected natural frequency in similar ways regardless of whether damping ratio was permitted to vary or held fixed at one of several different values. Damping ratio was generally unaffected by the linguistic factors. For the most part, the type of partition or window did not affect the patterns of the results, with the exception of the closing gesture out of the reduced syllable.

A preliminary model of speech production based on the concept of a gestural score, that is, an organization of dynamically defined articulatory gestures, has been developed. These gestures, which also serve as linguistic primitives, are coordinated movements of the lips and jaw or tongue and jaw. The movements are generated using a set of underlying dynamic parameters that are specified for each gesture in the score, namely, stiffness, rest position, and damping. (We assume a critically damped linear second‐order system with unit mass.) [Work supported by NSF.]

We have argued that dynamically defined articulatory gestures are the appropriate units to serve as the atoms of phonological representation. Gestures are a natural unit, not only because they involve task-oriented movements of the articulators, but because they arguably emerge as prelinguistic discrete units of action in infants. The use of gestures, rather than constellations of gestures as in Root nodes, as basic units of description makes it possible to characterise a variety of language patterns in which gestural organisation varies. Such patterns range from the misorderings of disordered speech through phonological rules involving gestural overlap and deletion to historical changes in which the overlap of gestures provides a crucial explanatory element.Gestures can participate in language patterns involving overlap because they are spatiotemporal in nature and therefore have internal duration. In addition, gestures differ from current theories of feature geometry by including the constriction degree as an inherent part of the gesture. Since the gestural constrictions occur in the vocal tract, which can be charactensed in terms of tube geometry, all the levels of the vocal tract will be constricted, leading to a constriction degree hierarchy. The values of the constriction degree at each higher level node in the hierarchy can be predicted on the basis of the percolation principles and tube geometry. In this way, the use of gestures as atoms can be reconciled with the use of Constriction degree at various levels in the vocal tract (or feature geometry) hierarchy.The phonological notation developed for the gestural approach might usefully be incorporated, in whole or in part, into other phonologies. Five components of the notation were discussed, all derived from the basic premise that gestures are the primitive phonological unit, organised into gestural scores. These components include (1) constriction degree as a subordinate of the articulator node and (2) stiffness (duration) as a subordinate of the articulator node. That is, both CD and duration are inherent to the gesture. The gestures are arranged in gestural scores using (3) articulatory tiers, with (4) the relevant geometry (articulatory, tube or feature) indicated to the left of the score and (5) structural information above the score, if desired. Association lines can also be used to indicate how the gestures are combined into phonological units. Thus, gestures can serve both as characterisations of articulatory movement data and as the atoms of phonological representation.

We propose an approach to phonological representation based on describing an utterance as an organised pattern of overlapping articulatory gestures. Because movement is inherent in our definition of gestures, these gestural ‘constellations’ can account for both spatial and temporal properties of speech in a relatively simple way. At the same time, taken as phonological representations, such gestural analyses offer many of the same advantages provided by recent nonlinear phonological theories, and we give examples of how gestural analyses simplify the description of such ‘complex segments’ as /s/–stop clusters and prenasalised stops. Thus, gestural structures can be seen as providing a principled link between phonological and physical description.

We are testing a model of articulatory coordination and control over time using an articulatory synthesizer (ASY) that converts time‐varying specifications of articulator positions into speed [Rubin, Baer, and Mermelstein J. Acoust. Soc. Am. 70, 321–328 (1981)]. These articulator specifications are now generated from an underlying dynamical model that defines a particular linguistic event. Coordination among individual articulators emerges from this model, as does the time course of the event. No point‐by‐point temporal control is required. Synthesis will allow perceptual evaluation of different underlying dynamical models. [Work supported by NIH and ONR.]

In the present computational model of speech production, an utterance is represented as an organization of primitive linguistic units, gestures, into a larger structure, a gestural score. Each distinct gesture is linked to a particular subset of vocal tract variables (e.g., lip aperture and protrusion) and model articulators (e.g., lips and jaw), and is associated with a set of time‐invariant dynamic parameters (e.g., lip aperture target, stiffness, and damping coefficients). The values of the dynamic parameters and their activation intervals are computed as part of the gestural score for a given utterance using a linguistic gestural model that includes a gesture‐based dictionary of English syllables and a flexible rule interpreter for manipulating dynamic parameters and inter‐gestural phasing. The gestural score serves as input to our task‐dynamic model of sensorimotor coordination. In this model, the evolving configuration of the model articulators results from the gesturally and posturally specific way...

Mistakes in the perceptual process can indicate some of the mechanisms involved in perception. Naturally occurring misperceptions, or “slips of the ear,” are an important source of errors. In this study, over 100 misperceptions were collected by the author and others. The errors are categorized in terms of phonemic similarity and location with respect to possible unit boundaries, such as word and syllable boundaries. Word boundary misperceptions in particular frequently occur—for example, “frigidity” heard for “feature detectors” spoken.

In the November 1979 ASA Meeting, a general purpose set of programs for speech synthesis, called Lingua, was described. Lingua has been used to synthesize speech from demisyllables. In this paper the details of the rule synthesis from demisyllables are described, with particular emphasis on the duration rules. A model is used to adjust the duration of various portions of the demisyllables differentially. The resulting synthesis will be demonstrated.

Speech production involves the formation and release of constrictions at different points in the vocal tract. Progress is reported on a task‐dynamic computational model of autonomous coordination among the components of articulatory synergies (e.g., lips and jaw) that participate in motions along functionally defined tract variables (e.g., lip aperture and protrusion). For each speech gesture, a time‐invariant dynamical system (damped, second‐order) is specified at the tract‐variable level, and is transformed into a gesturally and posturally specific dynamical system for the synergy components. Articulatory movement patterns emerge as implicit consequences of the gesture‐specific tract‐variable control structures and the ongoing postural state of the articulators. Explicit trajectory planning is not required. Significantly, the modeled coordinative processes are exactly the same during simulations of unperturbed, mechanically perturbed, and coproduced speech gestures. All simulations are implemented using...

LINGUA, a general purpose set of programs used to synthesize speech from demisyllables, presented a paper describing its structure and operation in the April, 1980, ASA meeting. At that time, the prosodic rules were quite rudimentary. Since then, a number of refinements have been introduced. For example, pitch and duration are no longer controlled in lock‐step, but may vary independently of each other; thus stress may be correlated with high pitch phrase‐initially, and with increased duration phrase‐medially, as suggested by Nakatani and Aston (1980); stress may also be correlated with low pitch. A variety of pauses have also been introduced in order to improve naturalness; the pauses vary both in duration, and in their effect on the intonation contour. The speech output will be demonstrated.

Lingua is a program used for synthesizing speech from stored LPC‐encoded demisyllables using rules to modify the duration, as well as other aspects, of the stored speech [C. P. Browman, ICASSP 80 Proceedings, 561–564 (1980)]. The original duration rules simply compressed nonphrase‐final demisyllables a certain amount, depending on their stress level. The present duration rules, while still quite simple, are based on a different approach in which a variety of factors independently interact to produce the final durations of the subportions of the demisyllables. The factors include binary representation of stress (stress/unstress), binary representation of within‐phrase position (final/nonfinal), and a variable factor based on the number of syllables in a set unit. The most recent output from Lingua, using the new rules, will be demonstrated.