Defining Phonological Rules over Lexical Neighbourhoods: Evidence from Canadian Raising

Recent years have seen a heightened interest in incorporating a wide range of nonphonological factors, including phonetic, morphological, lexical, syntactic, social, and usage factors, into explaining phonological phenomena (see, for example, Beddor et al., 1986; Browman & Goldstein, 1989; Keating, 1990; Ohala, 1990, 1993; Flemming, 2001; Hume & Johnson, 2001; Pierrehumbert et al., 2001; Gordon, 2002; Beckman & Pierrehumbert, 2003; Bod et al., 2003; Bybee, 2003; Guy, 2003; Padgett, 2003; Hall et al., 2004, among others). This interest seems to stem from the idea that “phonology” involves the study of linguistic sound systems and whatever factors affect these systems, regardless of more traditional notions of modular grammar. One aspect of this trend has been an interest in lexically based accounts of grammar in general and phonology in particular (e.g., Lahiri & Marslen-Wilson, 1991; MacDonald et al., 1994; Goldinger, 1997; Bybee, 2000; Pallier et al., 2001; Beckman & Pierrehumbert, 2003; Hay et al., 2003). This paper provides further evidence for the idea that the understanding of phonological patterns can be usefully informed by looking at phonological neighbourhoods in the lexicon; that is, by considering word similarity. The focus of investigation is “Canadian raising,” a cover term for the distribution of the vowels [ai] and [i] in certain dialects of English, including many spoken in Canada. It will be shown that the traditional phonological accounts of Canadian raising as a categorical process, governed by the voicing of the following segment and stress patterns, are not in fact sufficient to account for the actual patterns of Canadian raising for at least some speakers. In addition to these factors, phonological neighbourhoods also play a role in shaping the distribution of [ai] and [i]. Section 2 gives a brief overview of Canadian raising and its role in the phonological literature, and outlines an elicitation study carried out to look at the modern pattern of Canadian raising in a particular Ontario community. Section 3 then describes the results of this study, and section 4 discusses how the results can be modelled making reference to phonological neighbourhoods. Finally, section 5 presents conclusions.