This article is a reanalysis of cluster simplification in Quebec French (QF) in terms of government-licensing, a condition which requires non-nuclear governing heads to be licensed by a following vowel. It is suggested, contra Côté (1997, 1998), that simplification is triggered by a structural constraint rather than a constraint on sonority. It is shown that in QF, simplification does not apply to word internal clusters such as appartement and vendredi because the following vowel is realized, but applies to forms like table and casque, and converts them into [tab] and [kas] respectively at the surface level due to the lack of a final vowel. However, cluster reduction does not apply to final clusters such as barbe, gorge, and solde in which the first member is a liquid. To account for why simplification applies in one case and not in the other, it is suggested that the two types of final clusters differ with respect to syllabification: liquids are within branching nuclei, whereas the first member of other clusters is within a branching rhyme. The case of word-final cluster simplification attested in Haitian Creole is also examined.

Paradis and LaCharité (1996, 1997) have proposed a model of loanword adaptation, couched within the Theory of Constraints and Repair Strategies (Paradis 1988a,b). One of the mechanisms used in their model, called the Threshold Principle, first advanced by Paradis, Lebel, and LaCharité (1993), poses problems. This principle, whose implementation implies arithmetic counting, goes counter to standard views of generative phonology against counting. In this article, an analysis of deletion contexts found in loanwords which accounts for the data observed on structural grounds only is developed without any appeal to arithmetic counting. Based on the adaptation of French rising diphthongs and nasal vowels in two languages, Fula and Kinyarwanda, it is argued that an analysis based solely on the segmental representations of the foreign forms to adapt and the segmental and syllabic constraints of the borrowing language is sufficient to make correct predictions regarding the adaptation patterns found in these languages.