Advances in Sociophonetics

42 Bernard Laks, Basilio Calderone and Chiara Celata

the 14% of the corpus). The lexical environments represented in each of the three

bottom diagrams are mutually exclusive, inasmuch as each lexical environment is

(by definition) realized by one specific liaison consonant.

One can see from the figure that, although /n/ has more data points in the

diagram compared to /t/ and /z/ is even more frequent than /n/, the global distri-

bution is similar for the three consonants. The head of the curve is occupied by a

relatively limited number of high-frequency types, with discrete frequency values.

The 30 most frequent lexical environments are shown in Table 2 according to

consonant type. Five out of the six most frequent types contain /n/, and only one

contains /z/. There is a strong lexical bias among these high frequency types, since

four of the five /n/ types are realized by the indefinite 3rd person pronoun on. On

the other hand, 12 of the 25 most frequent environments contain a /z/. This indi-

cates that, although /z/ is more frequent than /n/ as a liaison consonant overall, the

liaison environments located very high in the frequency curve are specified by /n/,

not /z/. We must therefore exclude those few high-frequency /n/ environments, in

order to get a more balanced picture across consonant types.

RANK

100 101 102 103

16,678 realizations

100

102

100 101 102 103

6,494 /N/ realizations

100

102

100 101 102 103

2,342 /t/ realizations

100

102

100 101 102 103

7,842 /z/ realizations

100

102

OCCURRENCES

Figure 4 (a–d). Log-log plot of liaison environments (or ‘types’) in the PFC corpus (rank

order by number of occurrences) (a); split by type of the liaison consonant (b–d).