I'm a teenager from Chicago with a pretty standard contemporary Midwestern/General American accent (not distinctly Chicago). I'm interested in the phonetic phenomenon of Canadian Raising, in which diphthong phonemes such as those in the words "price" and "mouth" (the keywords of Wells' PRICE and MOUTH lexical sets) can be shortened and begin at a noticeably higher location in the vowel space in certain environments. In my accent, MOUTH vowel raising is not present, but PRICE raising absolutely is. This is the case for a lot of the United States.
The standard rule states that the raising occurs before voiceless consonant sounds (presumably, all of them) and not before vowels, pauses in speech, and voiced consonants (presumably, none of them). However, I and others have noticed that this rule is a bit inaccurate and, perhaps, overly reductive.
I have a distinction between the vowels in price/prize, flight/flied, and tripe/tribe - just as the rule indicates. But I also have the writer/rider distinction, where the following consonant sound is the same in both. This can be explained if we adapt the rule to cover PRICE vowels preceding fortis non-syllabic (consonantal) phonemes, rather than voiceless consonant phones. However, there is a famous exception to this: "spider," which has a raised vowel before /d/, as if the word were "spiter."
Sometimes a little extra note is added to the rule specifying that the following voiceless consonant must be in the same morpheme as the diphthong (sometimes they say it must be in the same syllable). This works for "high tide" (both very open) but not for "high school" (raised). I don't know if this is due to the ambisyllabicity of the /s/ in "high school" or the commonality of the phrase.
Some people have Canadian raising pre-/r/, and I don't know why. Analyzing my own accent, the distribution seems extremely strange: I have raising in "fire," and yesterday I caught myself using a very mid-central starting point in "tired." I think it works for all /t/PRICE/r/ and /f/PRICE/r/patterns. Going through a quick scan of consonants, the pre-/r/ raising doesn't seem to occur for me in any other situation - "wire" and "dire" begin pretty open, and so do "sire," "pyre," "shire," and "hire." Either there is something special about /f/ and /t/, or this is a weird form of lexical diffusion.
I would like to know whether or not my Canadian Raising distribution is common, where different exceptions are common or not, and what the most common distribution is. What is the best rule for the occurrence of the raised allophone of the PRICE vowel in accents like mine, and the MOUTH vowel for others?
I am aware that phonetics does not always follow "rules," but Canadian Raising seems to be a phenomenon which is almost always explained by a catchy rule. The examples I described above contest that simple rule, and I feel (possibly incorrectly) they are more than just random exceptions. There must be something other feature, something else to add to the rule - there always is.