Chapter 2. French liaison and the lexical repository 45
Figure 5 illustrates the data split by age groups compared to the global distri-
bution of the liaison data. The two bottom diagrams refer to the liaison environ-
ments realized by the ≤ 50 and > 50 year-old speakers, respectively, while the top
of the figure refers to the total amount of liaison environments resulting from
the sum of the two age subgroups. Differently from the analysis by consonant,
the lexical environments represented in each of the two bottom diagrams are not
mutually exclusive by definition, since each lexical environment can be (and often
is) repeated by the speakers of both groups.
The diagrams showed that the global distribution of the data across the curve
was approximately the same for the two age groups and no significant idiosyncra-
sies could be found: as repeatedly observed, the head of the curve is occupied by
the same small number of high-frequency types, with discrete frequency values.
Table 3 shows that the identity and distribution of those high-frequency lexical
types tended to be consistent across the two age groups. On the contrary, as far
as frequency decreases, the number of lexical environments per frequency value
increases. The curve ends up in a flat distribution where many liaison types feature
a frequency value equal to 1. However, the tail zone shows that there are differ-
ences according to the age of the speakers and some types featuring a frequency
value equal to or lower than 10 are evenly distributed across the two classes. This
is particularly evident in looking at Figure 6, which presents a magnification of
RANK102104100 101 102 10313,531 realizations100102104100 101 102 1037,367 realizations AGE ≤ 50100102104100 101 102 1036,164 realizations AGE > 50100OCCURRENCESFigure 5 (a–c). Log-log plot of liaison environments (or ‘types’) in the PFC corpus (rank
order by number of occurrences) (a); split by AGE ≤ 50 and AGE > 50 (b–c).