Onsets: Suprasegmental and prosodic behaviour (review)

In this book, Topintzi argues—contrary to what is usually assumed in moraic theory—that syllable onsets, like nuclei and codas, can bear weight. Evidence for this claim is drawn from a wide variety of phenomena, in particular, stress, compensatory lengthening, word minimality, and geminates. While the claim that onsets can be weightful itself is not new (e.g. Hajek & Goedemans 2003), Onsets is the first comprehensive study on this topic because it unifies a number of very diverse arguments that are not usually treated together.

Ch. 1 provides the reader with the necessary theoretical background, while Chs. 2-5 consider the topics relating to onset weightfulness listed above. Ch. 6 explores several brief case studies, and Ch. 7 discusses various alternative approaches.

T begins Ch. 1 by discussing the literature on syllable weight and the various models that have been proposed to capture it. The first part of the chapter considers stress, compensatory lengthening, word minimality, and geminates, arguing that this type of material can and should be captured with the moraic model (e.g. Hyman 1985, Hayes 1989) as opposed to an approach with skeletal slots (e.g. Clements & Keyser 1983). Although the moraic model is singled out as the most successful theory, that approach is argued to be empirically insufficient because it does not countenance moraic onsets. T holds that the traditional view that onsets are weightless is based on the claim that languages with weightful onsets are rare and that moraic theory should therefore stipulate that onsets are universally weightless (7). In essence, T argues that there are—from the universal perspective—the two types of onsets in 1.

1. 1.

   1. a. Nonmoraic onsets
      

   2. b. Moraic onsets
      

T contends that most languages treat onsets as in 1a, although there are a number of languages with moraic onsets, as in 1b. It is argued at length in later chapters that moraic onsets as in 1b hold not only for word-initial onsets, but also for word-medial onsets. There are two types of moraic onsets discussed in Ch. 1 (and throughout the remaining chapters): distinctive and coerced (cf. Morén 2001). The former refers to phonemic weight distinctions; for example, in a language that contrasts a singleton [C] and a geminate [C:], the latter must be underlying, that is, /C:/. By contrast, coerced weight is weight acquired in the output due to a requirement such as word minimality or weight by position, where an input moraless /C/ turns into a moraic [C_μ]. An optimality-theoretic (OT) treatment of the way in which coerced moraic onsets are derived is presented in Ch. 1.

Ch. 2 explores the connection between onsets and stress. T argues that onset-sensitive stress depends on two dimensions: (i) the quality of the onsets (QO), and (ii) the presence of the onset (PO). For QO languages, only certain types of onsets count as heavy, whereas others are light. For example, in Karo (a Tupi language spoken in Brazil) stress treats only onsets with a voiceless sound as moraic but not onsets with voiced sounds. Only the QO dimension is argued to be attributed to weight, while the PO dimension relates to alignment considerations that require stressed syllables to be onsetful, but not moraic. A four-way typology is posited: (i) languages with QO only, (ii) languages with PO only, (iii) languages with both QO and PO, and (iv) languages with neither QO nor PO. A clear example of a language falling into category (i) is Karo, which T analyzes in Ch. 2 in detail. Languages discussed in support of (ii) are Aranda (an Australian language), Banawá (an Arawan language spoken in Brazil), and Dutch, and for (iii) Pirahã (a language isolate spoken in Brazil). Each of these languages displays a pattern of coerced onset weight that is captured with a ranking of OT constraints. T notes that this constraint hierarchy predicts a pattern in which all onsets are moraic (38). While there are apparently no languages of this sort with respect to...