as it turns out, Scott and I both have interesting, if somewhat predictable, phonetic foibles: I, like many Oklahomans, tend to relax tense vowels, which is especially notable in the "ai" diphthong in the word "while," which I pronounce similarly to "wall."
Scott's invented usage extends the rule that causes "f" to become voiced ("v") when an "s" is added after it at the end of a word. "life" becomes "lives," "wolf" becomes "wolves," "bath" "bathes", "belief" "believes" and so on. He sometimes pronounces the plural of "death" with a voiced fricative - "deatthhz." (there's no letter for that sound!) And the plural of "graph", "gravz."
Taken by themselves by linguists, these observations demonstrate regular changes that languages tend to undergo, or the application of an accent. But the more interesting thing is that these phonetic changes also depend on usage.
I do not pronounce "while" like "wall" in the sentence, "i've been saying it this way for a good long while." And Scott does not say "gravz" in the sentence, "he graphs tangents every thursday."
Both of these phonetic differences are not typically considered meaning-related. In most linguistic studies, phonetic and semantic changes are seen as occuring independently. But my more "careful" pronunciation of "while" when it's used as a noun (especially at the end of a sentence), and Scott's differentiation between singular verb and plural noun are only phonetically predictable when the meaning and the context of the words is taken into account.