Note on Frequency in Phonetic Change

It is claimed that frequency of individual words may not be sufficient to predict the likelihood of a particular word being affected by a phonetic process: Given two environments that can equally trigger phonetic change, the same lexical item may be more likely to undergo change in one environment than in another, depending on how frequently the environment is associated with that item. An ongoing raising process in colloquial modern Hebrew illustrates the point: Raising of e to i verbinitially, when immediately preceded by a word-final i, is more likely to apply where the conditioning environment (here the pronoun) is obligatorily present, e.g. ani extov 'I will write'>ani ixtov, i evina 'she understood'> i ivina. Although analogy may in part be involved (with u ixtov 'he will write' and n-initial stems like /inpil/ >ipil 'he dropped'), obligatory presence of ani 'I' in 1st per. masc. sg. future and i 'she' in 3rd per. fem. sg. past accounts for e-Raising applying to such forms more often than to