46 Bernard Laks, Basilio Calderone and Chiara Celata
the tail of the curve. Some low-frequency realizations are typically produced by
younger speakers, some others are typically produced by older ones. The token fre-
quency value of these evenly distributed lexical environments is low, but they are
numerous. The resulting picture is visibly differentiated across age groups for what
concerns the selection of low-frequency lexical environments featuring liaison.
The analysis of the liaison distribution according to the speakers’ educational
level is conducted in the same way as the analysis by age. Figure 7 illustrates the
data split by educational level (three bottom diagrams) compared to the global dis-
tribution of the liaison data resulting from the summation of the three subgroups.Table 3. Number of occurrences (frequency), percent and cumulative percent for the 25
highest ranked liaison types in the PFC corpus, separately for AGE ≤ 50 and AGE > 50.
Realizations of liaison specified by age
AGE ≤ 50 and AGE > 50
RankLiaison Frequency Cumulative
percentAGE ≤ 50 AGE > 501 on_L_a 1098 8.1 617 481
2 on_L_est 530 12.0 306 224
3 ils_L_ont 434 15.2 250 184
4 en_L_a 338 17.7 190 148
5 on_L_avait 299 19.9 135 164
6 on_L_était 251 21.8 114 137
7 quand_L_on 214 23.4 99 115
8 dans_L_un 175 24.7 109 66
9 dans_L_une 164 25.9 105 59
10 deux_L_ans 158 27.1 101 57
11 est_L_en 158 28.2 95 63
12 les_L_énfants 146 29.3 99 47
13 est_L_un 137 30.3 47 90
14 vous_L_avez 128 31.3 67 61
15 tout_L_à 119 32.1 51 68
16 les_L_autres 114 33.0 75 39
17 trois_L_ans 111 33.8 58 43
18 quand_L_il 109 34.6 42 67
19 vous_L_etes 101 35.4 47 54
20 ils_L_étaient 97 36.1 59 38
21 n_L_ai 97 36.8 49 48
22 est_L_une 93 37.5 36 57
23 nous_L_a 93 38.2 43 50
24 nous_L_avons 88 38.8 35 53
25 ils_L_avaient 86 39.5 32 54