Electronic Thesis and Dissertation Repository


Doctor of Philosophy




Dr. François Poiré


This thesis examines phonetic and phonological aspects of gliding in Hexagonal French. In particular, we ask: Are glide phenomena as predictable as portrayed in modern descriptions? Do all three glides /j, w, ɥ/ or corresponding high vowels /i, u, y/ behave alike in all potential glide contexts? Given the duality of French glides (vowel and consonant), we use the term vocoid and the archiphoneme convention /I, U, Y/ in our discussion of glide contexts and glide phenomena. Our historical survey shows the glides of French (/j, ɥ, w/) evolve separately and during this period the high front vocoid /I/ occurs early and is involved in greater variety of contexts showing considerable variability. The other two glides emerge later, primarily through diphthongisation, and show less variability.

In a study of glide contexts in the spontaneous speech of native speakers from three regions of France (data from the Phonologie du Français Contemporain project), we examine the distribution of all three high vocoids and their surface realisations. For 3415 tokens identified, we determine if HVV (high vocoid plus vowel) tokens are realised with dieresis, with syneresis, or with the high vocoid deleted. Our findings show glide contexts are consistently distributed at a rate of about 85% lexicalised and 15% derived. The limited variability in lexicalised contexts involves mainly the non-round vocoid /I/ realised with dieresis. Distribution across the three-glide inventory of French shows that lexicalised glide contexts follows a general markedness hierarchy: I ⨠ U ⨠ Y. Tokens involving the front non-round vocoid /I/ are most prevalent followed by the back rounded vocoid /U/ and finally the front rounded /Y/. Derived contexts include word medial tautomorphemic high vowel + vowel /HV+V/ sequences resulting from suffixation or inflection, and also cross-word-boundary /HV+V/ sequences which have very rarely been studied before; we show that cross-word-boundary data largely follow the same phonological constraints as derivational data. In each of these contexts the general markedness hierarchy observed above is changed, giving preference to the front rounded /Y/ over the back rounded /U/ while /I/ remains most prevalent.