Space and time in models of speech rhythm

Abstract

How do rhythmic patterns in speech arise? There are many representational frameworks for describing rhythmic patterns, but none of these directly connect representations to articulatory processes, which have physical manifestations in the acoustic signal. Here, a new model of speech rhythm is presented, one in which rhythmic patterns arise from spatial mechanisms that govern the organization of articulatory gestures. The roles of time and space in symbolic representations of the metrical structure are analyzed, and conventional understandings of stress and accent are called into question. One aspect of rhythmic patterns, in particular—the directionality of accentual patterns—is examined closely. A novel dynamical model is developed, which proposes a reinterpretation of directionality and other temporal phenomena in speech.

Motoric Mechanisms for the Emergence of Non-local Phonological Patterns

“…The gestural scores of Articulatory Phonology/Task Dynamics do not impose any form of grouping on the gestures in a score. Indeed, there is no direct representation of syllables or moras in standard gestural scores, and this raises a number of challenges for understanding various typological and developmental phonological patterns (see Tilsen, 2016, 2018a). In order to address these challenges, the Selection-coordination model was developed in a series of publications (Tilsen, 2013a, 2014a,b, 2016, 2018b).…”

“…Because the selection-coordination model has been presented in detail elsewhere, only a brief introduction to the model is provided below. Furthermore, discussion of the full range of phonological patterns which the model addresses is beyond the scope of the current paper, and the reader is referred to other work for more thorough exposition (Tilsen, 2016, 2018a,b). Here we present the model in sufficient detail for the reader to understand how it interacts with intentional planning, and we address the question of when gestures may or may not influence intentional fields.…”

“…The reader should note that the motor population differentiation pattern in Figure 7A exhibits a particular spatial arrangement, such that the initial m-system organization for a word form corresponds to the spatial pattern of differentiation in the motoric population. This spatial correspondence is not necessary for our current aim – modeling long-distance phonological patterns – but it is useful for a more comprehensive model in which the directionality of metrical-accentual patterns can be interpreted (see Tilsen, 2018a). Furthermore, it is important to emphasize that the motor population is finite and thus when a word form requires a greater number of m-system differentiations, the size of each m-system population becomes smaller and m-systems become more susceptible to interference.…”

“…Agreement is never sensitive to stress or other metrical structure, and is never bounded by prosodic domains such as the foot; in contrast, such prosodic domain restrictions are common for spreading patterns, such as vowel harmonies and vowel-consonant harmonies (Hansson, 2001; Rose and Walker, 2011). This difference can be interpreted with the idea that domains such as the prosodic word are associated with the selection of accentual gestures (Tilsen, 2018a, 2019), in conjunction with the idea that selection of accentual gestures can influence the promotion and demotion of articulatory gestures. Accentual gestures specify F0 and/or intensity targets, and are associated with stress (i.e., metrical structure) as well as intonation (pitch accents).…”

“…It is important to note that the model developed here involves the planning and production of word forms by an individual speaker, and the articulatory patterns generated by the model are viewed as seeds of potential sound change on larger spatial and temporal scales. The starting point of the model is the gestural framework of Articulatory Phonology (Browman and Goldstein, 1989) and Task Dynamics (Saltzman and Munhall, 1989); recent extensions to this model in the Selection-coordination framework (Tilsen, 2016, 2018a,b) are also incorporated. We will develop an extension of these models in which there are two distinct ways for non-local patterns to arise; these mechanisms are shown to account for the origins of spreading and agreement harmonies, respectively.…”