Chapter 2. French liaison and the lexical repository 35
vs. southern French etc. (e.g., Pagliano & Laks 2005; Durand & Lyche 2008; Mallet
2008; Coquillon et al. 2010; Durand et al. 2011). Some of these factors have been
shown to have an impact on the frequency of realization of liaison in facultative
contexts, as in the case of the age factor (old speakers tend to produce more liai-
sons than younger speakers, Durand et al. 2011), while other factors have been
proved not to influence the use of liaison substantially (as in the case of geographic
variation and gender, Coquillon et al. 2010). Finally there are factors that still
need to be thoroughly investigated; one such example is the educational level of
the speakers (see Durand et al. 2011: 121). Influence of socio-economic status still
deserve precise investigations based on large data bases; see for example the recent
analyses of the influence of parental socio-economic status on children acquisition
of liaison by Chevrot et al. (2011), and Hornsby’s (2012) stratificational analysis
of liaison errors and repairs.
In this paper, we propose a frequency analysis of French liaison and liaison
environments focusing on the specific contexts of liaison realization attested in
the PFC database. Liaison environments are the attested combinations of two
consecutive words generating liaison at their juncture. The frequency analysis is
aimed to uncover the distributional aspects of French liaison in its actual lexical
instantiations, under the general usage-based hypothesis that liaison is more fre-
quently realized in those word groups that have strong internal cohesion and high
frequency of co-occurrence (e.g., Bybee 1998, 2001, 2005).
We will verify this distributional hypothesis over a very large amount of pro-
ductions realized by different groups of French speakers varying in terms of age
and educational level. The findings will show that our initial hypothesis is cor-
rect inasmuch as the distribution of liaison is similar to a power-law distribution
in which a few types are ranked high for productivity and account for approxi-
mately one-half of the total observations. We will see that this generalization holds
true for all types of linking consonants and all groups of speakers (as determined
according to age and educational level), with some oscillations concerning differ-
ent types of infrequent liaison patterns.
The paper is structured as follows. Section 2 presents our working definition
of liaison environment and illustrates aims and procedures of the distributional
analysis of French liaison in PFC. Section 3 presents the main results separately for
the PFC corpus (§3.1), for some individual liaison consonants (§3.2) and for some
subgroups of speakers as defined by age and educational level (§3.3). Section 4
presents the general discussion and concludes.