Abstract
A new computer-assisted technique for analyzing large corpuses of natural speech, in this case 50 tape-recorded conversations with upper middle-class Parisians, has revealed a general rule of progressive nasal assimilation, in which the acoustic realization of the /b/ in une chamb (re) de bonne /yn∫ãbdÉ™bcn/ is identical to that of the /m/ in une âme de bonne /ynamdÉ™bcn/, at least as far as nasality is concerned. This phenomenon occurs in the context nasal vowel + stop + word juncture + consonant; (1) almost without exception in cases such as the above, where final -re or -le have been suppressed and where another word beginning with a stop or a fricative follows without pause; (2) quite frequently in cases such as grande ville where there is no suppressed –re or –le but the following word begins with a voiced stop or fricative, and in cases such asprend(re) le Metro where a liquid or a nasal consonant follows suppressed word-final –re or –le; and (3) occasionally in cases such as langue populaire where there are no suppressions but where the second word begins with a voiceless consonant, and in the words done, quand and cinq when their final consonant is exceptionally pronounced before another consonant. These results have been corroborated by instrumental measurements of nasal acoustic output versus closure timing and by the results of a questionnaire dealing with the Frenchmen’s preferences.