112 Zsuzsanna Fagyal
allow a clear separation of the speakers in two groups. About 89% (N=1293)
of syllables in the sample had at least one onset consonant, but only AF speak-
ers realized these consonants, in about 2.6% of cases, as glottal stops or pala-
talized affricates. Deletion of onset consonants occurred only in AF speakers’
speech (0.15%). Similarly, schwa insertion and vowel devoicing in CVC syl-
lables only characterized AF speakers, while there were nearly identical pro-
portions of coda deletion in the two groups. Devoicing of syllable nuclei in
CV syllables, on the other hand, was twice as frequent in the EF group (67%)
than in the AF group (33%).
While deletion and, perhaps to some extent devoicing, could be consid-
ered fast speech processes, i.e., resulting from the fact that AF speakers, over-
all, articulated somewhat faster than EF speakers, it is unlikely that schwa
insertion, glottalization, and the affrication of palatalized stop consonants
could be attributed to rate differences. Clearly, something else other than
articulatory rate must lie behind AF speakers’ tendency to dissolve consonant
clusters with epenthetic vowels, and to prevent vowel coalescence (enchaîne-
ment vocalic) through the insertion of glottal stops.
Figure 4.6 shows intra-speaker variations in the frequency of onset,
nucleus, and coda types. Ample variations within the AF group (the ¿ rst ¿ ve
speakers from the top), contrasting sharply with a greater uniformity of allo-
phonic realizations in the EF group (the last ¿ ve speakers), can be observed.
Speakers in both groups resor ted to coda consonant deletion (1) and devoicing
of syllable nuclei in CV syllables (2):
Figure 4.5 Types of onset, nucleus, and coda in the readings of French speakers of
North African and European descent (N of total syllables =1455).