It is commonly observed that some phonological positions are associated with prominence while others are not. Prominent positions tend to be characterized by having more robust phonetic cues or a wide range of phonological contrasts, while non-prominent positions have weaker cues and fewer contrasts (e.g., Beckman 1999, Smith 2000). Prominence, however, has often been used as an explanation (e.g., there are more contrasts in a certain “strong” positions because they are psycholinguistically or phonetically prominent; see Beckman 1999), rather than a phenomenon which must itself be explained. In this paper, we examine what it actually means to be a prominent position and what would cause some positions to be considered prominent while others are not. In particular, we claim that prominent and non-prominent positions are those that are strategic from the perspective of communication.
We take as a starting assumption that language is a type of communication, i.e., the transmission of a message from one point to another, and therefore subject to the same considerations as other systems of communication (cf. Shannon & Weaver 1949). As a general proposition, successful information transmission is dependent on the resolution of uncertainty in the mapping between the signal and the message, and in language, the phonetic signal is manipulated in order to maximize the effectiveness of this transmission. We propose that the most strategic locations for this manipulation to occur are those of either particularly high uncertainty—which are subject to enhancement and become phonologically prominent positions—or particularly low uncertainty—which are subject to reduction and become phonologically non-prominent positions. Thus, word-initial positions and stressed syllables end up as prominent, while their complement, word-final positions and unstressed syllables are non-prominent.
Conceiving of phonological prominence as a by-product of general communicative strategies thus allows us to unify a number of previously disparate observations about phonological phenomena, as will be detailed in this talk. Furthermore, there are well-established mathematical tools for quantifying communication, in the form of information theory and Bayesian probability, which allow for objective description and prediction of phonological processes.