116 Zsuzsanna Fagyal
Schwa insertion also applied variably. As one would expect it in Northern
varieties of French in France, AF and EF speakers inserted schwas phrase-
medially to prevent three consonants to cluster in a single onset, e.g., in notre
prof ‘our teacher’ in example (4), but only AF speakers inserted schwas
between two single consonants, such as between the words sur ‘certain’ and
que ‘that’ in example (3). Not altogether foreign to French, such patterns of
schwa insertion are typical in Southern French varieties. Finally, glottalized
onsets that prevent the linking of adjacent vowels (enchaînement vocalique)
within the same accentual phrase are also possible in Northern varieties of
French, but reserved exclusively to emphatic contexts, referred to as ‘empha-
sis by expressive juncture’ (emphase par joncture expressive) (Léon 1993:
143–144). The fact that glottal stops occurred several times in the speech of
AF speakers in the reading of a text, a relatively neutral speech performance
elicited in a relatively formal context, points to possible contact features
retained from the heritage language.
- Summary of ¿ ndings
This study revealed uniformity and differences in the segmental components
of rhythm type distinctions in the readings of two groups of male adolescents
recorded in a French working-class suburb of Paris. The hypothesis that read-
ings of native French heritage speakers of Arabic (AF speakers) would show
inÀ uence from the heritage language was con¿ rmed, but contrary to expecta-
tions it was not borne out by statistically signi¿ cant central and dispersion
measures of variation. It was detectable only in greater allophonic variation,
¿ ne-grained acoustic characteristics of segments that constitute the essential
building blocks of syllables in these speakers’ speech.
Central and dispersion measures revealed uniformity between bilingual
heritage speakers (AF group) and monolingual speakers of French (EF group):
both groups patterned closer to each other and syllable-timed languages than
to any of the stress-timed languages on the rhythm type continuum established
in previous studies. Thus, the hypothesis that AF speakers’ readings would
pattern closer to stress-timed languages had to be rejected: their rhythm were
just as characteristically syllable-timed as EF speakers’.
Signi¿ cant differences were found in the distribution of vocalic and
consonantal interval durations, and regression analyses revealed that the
contribution of ¿ ne-grained rhythm type predictors was outweighed by social-
demographic and performance-related factors. One year of average grade
(approximate age) difference and a one point average difference in students’