This thesis investigates the formal properties of phonological representation and computation. The starting point of the approach taken here is that these can and should be investigated independently of the effect that extraphonological factors, most notably phonetics, have on the shape of individual phonologies. In chapter 1, I summarise the conceptual and empirical arguments for a model of autonomous phonology. Then I discuss the differences between substancefree approaches to phonology, including the Concordia school (Hale & Reiss 2000a,b, 2003; Hale et al. 2007; Hale & Reiss 2008), the Toronto school (Dresher et al. 1994; Avery 1996; Dresher 1998; Avery & Rice 1989; Rice & Avery 1991; Piggott 1992; Rice 1993; Dresher 2001, 2003 inter alia), Element Theory (Harris 1990, 1994; Harris & Lindsey 1995; Harris 2005, 2006), the Parallel Structures Model (Moren 2003a,b, 2006) and radically substance-free phonology (Odden 2006, this thesis). The approach followed in this thesis is the most substance-free of the alternatives examined: neither phonological computation, nor phonological primes are innately connected to phonetic (or other extra-phonological) correlates. I discuss different formal aspects of phonological representations, and argue for a model using privative indexical features that can freely enter into feature geometrical dependency relations with one another. Finally, I summarise the most important properties of the architecture of radically substance-free phonology. Chapter 2 deals with integrating substance-free phonology and Optimality Theory (Prince & Smolensky 1993). I formalise featural identity constraints in a way that is compatible with a model using privative features and an unrestricted feature geometry. I also formalise Max and Dep constraints on features, and show how the model presented here can account for ‘fea-
Stored only when both papers are in the corpus — 170 citations exist upstream.
Imported from OpenAlex (CC0) — nothing here is generated.