Variable Liaison, Diglossia, and the Style Dimension in Spoken French

This article tests the diglossia hypothesis, according to which informal/spoken and formal/written French have diverged to the point of being separate High and Low varieties in Haugen's (1966) terms, using a corpus of data from 96 speakers examined for variable liaison in scripted and unscripted style. While the data do not lend support for a diglossia model, they do not in themselves refute it, because the the hypothesis as it stands is empirically unfalsifiable. A comparison of the speakers investigated here and 'professionnels de la parole publique', i.e. individuals for whom speaking in public is an occupational requirement, suggests nonetheless that the diglossia model offers a poor fit for liaison data, and an alternative four-level model for this complex variable is proposed.