Raised vowel in START

I was reading an article about Canadian English last week that made me notice something I never have before. My vowel in 'start' is quite different from that of 'star' or 'card'. It seems to be less raised before fricatives then for stops and affricates.
start [s6r/?]
card [kAr/d]
arch [Ar/tS] or [6r/tS]
farce [fAr/s] or [f6r/s]
I think fricatives and affricates might vary between the two.
What accents have this raising, and in what environments?

<<I was reading an article about Canadian English last week that made me notice something I never have before. My vowel in 'start' is quite different from that of 'star' or 'card'. >>

Oh really? ;-) I asked you about this very phenomenon recently, and you said that you didn't have it ( http://www.antimoon.com/forum/t6435.htm ). It's okay; it's likewise taken me a while to notice some features of my speech.

Anyway, as I mentioned in that thread, I do have raising there. For me, raised /Ar/ occurs in the same environments as raised /aI/ and /aU/: before any voiceless obstruent. I have it in:

start ["stVr\t]
arch ["Vr\Ts]
farse ["fVr\s]
hearth ["hVr\T]
park ["p_hVr\k]
Darth Vader [%dVr\T "veI4@`] ;-)

(And of course my [V] is near-back rather than fully back.)

But I don't have it in:

card ["k_hA@`d]
cars ["k_hA@`z]
barge ["bA@`dZ]

Typo: "farse" should be "farce".

I likewise have raised /ar/ before any voiceless obstruent phoneme, having:

start ["stVR?]
arch ["VRtS]
farce ["fVRs]
hearth ["hVRT]
park ["p_hVRk]
Darth Vader [%dVRT"ve:4R=:]

At the same time, I have backing of /ar/ in all other cases, having:

star ["stA:R]
card ["k_hA:Rd]
cars ["k_hA:Rs]
barge ["bA:RtS]

One note: said backing of /ar/ is likely only backing synchronically, as I suspect that diachronically what really happened is that historical [A] in such positions was never fronted by the NCVS, which would be consistent with the raising of /ar/ before voiceless obstruent phonemes to [VR]. This, of course, implies that both the raising of /ar/ to [VR] and the uvular realization of /r/ IMD predate the NCVS.

<<Oh really? ;-) I asked you about this very phenomenon recently, and you said that you didn't have it ( http://www.antimoon.com/forum/t6435.htm ). It's okay; it's likewise taken me a while to notice some features of my speech.>>

No, I didn't notice it back then, but the article was specifically about Canadian English, so I listened a little more closely and I realized that I do.

Oxford Canadian Dictionary has [Q] for everything:

stressed vowel in
Mississauga, Ottawa, father, caught, cot, star, start, Chicago...

(ChicQgo--> ChicAgo-->Chica:go-->ChicAEgo)

Actually, in the dialect here, "Chicago" is pronounced [SI"k_hQ:go:], with the [Q] corresponding to the historical COUGHT vowel not the historical BOTHER vowel. I suspect that the [Q] that is mentioned in said dictionary is not actually the BOTHER vowel either but rather this lowered COUGHT vowel being transcribed phonetically as per its pronunciation in Chicago. On that note, I have never heard "Chicago" pronounced with [a] or [{] by any native English-speakers.

At the same time, I have heard people pronounce "Chicago" with [A] rather than [Q], but in the speech of individuals from the Upper Midwest this may very well be the shifting of the COUGHT vowel all the way down to [A] rather than being related at all to the FATHER or BOTHER vowels.

I pronounce "Chicago" as [SI"k_hA:g7U], putting it in the historical FATHER class.

<<I pronounce "Chicago" as [SI"k_hA:g7U], putting it in the historical FATHER class.>>

It's my impression that most Americans use /A/ for 'Chicago', outside of the Inland North and Upper Midwest, where /O/ seems to be more common. I say [SI"kQgo], but of course that could reflect /A:/, /Q/, or /O:/ from other dialects, so that doesn't help much.

Here are some rather anomalous cases of Canadian Raising in my dialect:

idle ["@:I4M:]
Midol ["m@:I4M:]
Idaho ["@:I4@:ho:]
Fido ["f@:I4o:]
bridle ["br\@:I4M:]

but:

bible ["ba:IbM:]
bridal ["br\a:I4M:]
sidle ["sa:I4M:]

(I think this only applies to words where the /aIdl/ or /aIdo/ all fall within the same morpheme, and does not apply with words with other voiced plosives in them.)

Also while I have raising in some cases of /aI/ followed by /@r/ in the same morpheme:

tiger ["t_h@:IgR=:]
spider ["sp@:I4R=:]
hibernate ["h@:IbR=~:ne?]
hydrogen ["h@:IdZR=:dZI~:n]
hydroelectric [%h@:IdZr\o:@::"L\EktSr\Ik] (note that this does not have /@r/ yet acts like "hydrogen"!)

I lack it in words like:

visor ["va:IzR=:]
fiber ["fa:IbR=:]
miser ["ma:Iz=:]

The thing is that in some cases like this, it seems like there may be a marginal phonemic distinction between [aI] and [@I], but at the same time, this may be just due to incomplete diffusion of sound changes through the vocabulary which has not yet affected all words possible.

Interesting. I use the unraised allophone [aI] in all the words you mentioned.

Some of the above words are rather well-established with the pronunciations mentioned, such as "tiger" and "spider", which clearly have raised pronunciations throughout at least much of the Upper Midwest, whereas some of them have alternations between normal and careful pronunciations, such as "bridle", "hibernate", and "hydroelectric", which may have unraised pronunciations in careful speech. This is consistent with such being due to the diffusion of sound change throughout the vocabulary, with it being more well-established in more common words than less common words.

Here is another weird case in my dialect:

Siberial [s@:I"bI:Ri:@:]

That should be "Siberia" above.