I had a bit of a realisation this morning about English, and it's all thanks to Corsican. Corsican tends to weaken certain vowels when they're unstressed. So marking the stressed vowel with bold type, "accende" (the infinitive to light or enflame) becomes "accindite" (present, 2nd person plural). And this happens with almost all Es. Almost. Note the unstressed E at the end of both accende and accindite. But these unstressed Es only seem to occur where they have a specific grammatical purpose -- as far as I can tell any other E becomes I.
Now in certain parts these vowel "mutations" don't occur, but the majority dialects tend to do it. The odd thing, then, is that vowel mutation happens even though speakers of the language are evidently capable of saying the "forbidden" sound. Why not say it if you can?
Well, somewhere along the line, I started thinking about English, and in particular the prefixes pre- and re-.
There's two pronunciations for each: one with schwa and one with /i/. The schwa occurs wherever the syllable has no stress, normally adjacent to the primary stressed syllable -- eg "report", "reply" -- and the /ri/ pronunciation when it has secondary stress. So that's an "ee" sound, like a Corsican "I". It never seems to have an "eh" sound, like a Corsican "E".
And no matter how much I try to, I can't think of a single word in English with the "eh" of "pedal" and "petal" anywhere but in the position of primary stress.
Unless your American, in which case it occurs everywhere.
And that's what I'd never noticed before -- I always thought of the US "reh"-produce as though it was something specific to the re- prefix, but it's a bit more fundamental than that, isn't it...?