n. The process by which a phonological change begins by applying only to certain words and then spreads gradually to other phonologically similar words. In some cases, lexical diffusion stops at some point, leaving all remaining words permanently unaffected; in other cases, the process eventually goes to completion by affecting all remaining words. A famous example is the raising and tensing of /æ/ in Philadelphia.
Before final /d/, speakers consistently pronounce a tense vowel in mad, bad and glad but a lax one in sad, had and dad, suggesting that the diffusion has stopped operating only part way through the relevant class of words. In other contexts diffusion appears to be still occurring: only a small minority of speakers have a tense vowel in manner and flannel, whereas a large majority have one in planet. Implicitly or explicitly defended by the nineteenth-century Romance dialectologists and by a number of twentieth-century structuralists, lexical diffusion only became widely accepted after the publication of overwhelming evidence in the 1970s; it is now known to be a major pathway of phonological change, and its existence constitutes a direct refutation of the Neogrammarian Hypothesis. See Sommerstein (1977:251–252) for a historical summary, Chen and Wang (1975) for the classic presentation, and Labov (1994) for extensive discussion. Wang (1969).
This is the complete article, containing 214 words
(approx. 1 page at 300 words per page).
n. The process by which a phonological change begins by applying only to certain words and then spreads gradually to other phonologically similar words. In some cases, lexical diffusion stops at some point, leaving all remaining words permanently unaffected; in other cases, the process eventually goes to completion by affecting all remaining words. A famous example is the raising and tensing of /æ/ in Philadelphia. 
