Kenneth de Jong
Dept. of Linguistics, 322 Memorial Hall, Indiana Univ., Bloomington, IN 47403
Elizabeth C. Zsiga
Georgetown Univ., Washington, DC 20057
Ohala (1975, 1981) has proposed that sound changes are caused by listeners misperceiving coarticulatory effects. This paper examines variable ``r-dropping'' in Brooklyn English, and discusses the coarticulatory exigencies that may encourage speakers toward weakened forms. It is argued that production strategies do play a role in driving sound change apart from creating misperceptions. Recordings were made of speakers of Brooklyn and other dialects as part of the development of a larger multidialect database (Hertz et al., 1994). Analyses of nuclear r's (as in bird, and burl) show that neighboring l's both lower the second formant and raise the third formant, obscuring the r. Analyses of coda r's in ``r-ful'' speakers also show following coronals raise the r's third formant. Analyses of three Brooklyn speakers show one consistently produces coda r's, one never does, and a third does so variably as evident in a bimodal distribution of formant patterns. The variable speaker produced r-less tokens particularly in codas which contained coronals, especially l. This pattern suggests that coarticulatory influences affect the application of variable rules. Thus coarticulation seems to exert pressure within a speaker toward a changed form. [Work supported by the NIDCD.]