The PENTA Model: Concepts, Use, and Implications

Abstract

Speech is a communication system for transmitting information from one human being to another. 1 The information transmitted is rich and multifaceted, but it is coded by an articulatory system in such a way that the listener can readily decode it. These facts, which may seem too obvious to be worth restating, are the premise of the articulatory-functional view of speech that forms the basis of the parallel encoding and target approximation (PENTA) model (Xu 2005). PENTA is therefore a theory of how multiple layers of information are effectively conveyed through prosody with a neu-rally controlled biomechanical system. In other words, PENTA is about how prosody works as a communication system, how it can be learned, and how it goes through changes over time-in short, how it operates. The mission of PENTA therefore differs from those of many other theories that focus on directly accounting for observed pro-sodic forms. By focusing on operation as its primary goal, PENTA accounts for prosodic forms only as a by-product, rather than as an end in itself. 11.2 The Conceptual Framework Beyond the basic facts just stated, PENTA makes a number of assumptions that are highly hy po thet i cal. The first is that syllable-synchronized sequential target approximation (the TA part of the model) is the rudimentary mechanism of speech prosody, based on which all of the information coding is done. The second is that prosody conveys multiple layers of information si mul ta neously, through encoding schemes that are in parallel to each other, that is, without a hierarchical structure (the PE part of the model). Third, the phonetics of the encoding schemes are specified parametrically rather than based on symbolic repre sen ta tions. Due to their hy po thet i cal nature, each of these assumptions needs in de pen dent justifications, which we will provide after a brief sketch of the model. Figure 11.1 is a schematic of PENTA in its most general form, representing not only prosody but also other aspects of speech (Xu and Liu 2012). The first block from the left represents communicative functions that are conveyed by speech. The functions are parallel to each other, as illustrated by the nonhierarchical stacking in the schematic. The second block represents encoding schemes that are associated with the communicative functions. The schematization here makes it clear that communicative functions do not control surface acoustics directly but through a set of specific encoding schemes. The encoding schemes can be highly stylized and language-specific or more gradient and universal. The third block represents the articulatory par ameters that are