Interaction between syllable structure and segmental properties: The case of glide distribution in Swiss French

Abstract

The distribution of glides in French and their alternation with high vowels is a recurrent topic in the literature on French phonology. Having served as testing ground for various theoretical approaches, three questions in particular that on a regular basis revive the discussion are (i) the position of the glides in the syllable structure, (ii) the nature of the process, and (iii) the nature of the glides themselves, see e.g. Schane (1968) for a classic derivational approach, Kaye and Lowenstamm (1984); Klein (1991) for a syllabic approach, Bullock (2002); Durand and Lyche (1999); Hall (2006) for analyses within OT, and most recently Côté (2018), who proposes alternation to be the result of phonetic reduction.

A third aspect that makes the analysis still more complex, is the variation attested within and across French varieties (Côté, 2018; Durand & Lyche, 1999; Kelly, 2015). With the development of speech corpora containing comparable data, e.g. the PFC database (Durand, Laks, & Lyche, 2002, 2009), openly available at www.projet-pfc.net, we are now in a position to better determine the current distribution of glides in French varieties, and thereby build an empirical basis that allows comparison of the proposed analyses.

This paper focuses on Swiss French varieties, which are traditionally perceived as preferring vowel sequences over glide formation, e.g. [nue] > [nwe]. A recent pilot study (Andreassen, 2018), examining glides in four investigation points in Switzerland and one investigation point in Northern France, to a large extent confirmed this perception. It further revealed a certain level of inter-variety variation within Switzerland, in particular for derived forms (2).

In this paper, we take the study one step further, and aim at looking deeper into the factors determining the distribution of glides in Swiss French, e.g. morphological complexity, segmental context, and position in the word. By adding data to the existing dataset, we further aim at revealing more solid tendencies, which in turn will allow a discussion on how to analyse the results.