|
|
Phonemic vowel length
This came up on another thread, and I thought maybe it deserved one for itself. Lazar brought up the difference between English tense vowels and the corresponding vowels in Spanish, and I though I would expand on that a bit. I don't speak Spanish, but I speak (Quebec) French. There is an allophonic tense-lax distinction in Quebec French close vowels, but no length difference between the pairs. There is also an allophonic length distinction that applies to the tense vowels, so there is are three realisations of /i/ [i, I, i:]. I find that my English /i/ more closely resembles my French [i:] than [i]. I've recorded a sample, so that you can evaluate. Here's what I'm saying:
petit, petite, tige [p@tsi] [p@tsIt] [ti:Z]
see, sit, seed, Sid, seat (I won't transcribe these, so you can decide whether they're short or long)
My impression is that /i/ is long in all cases, while /I/ is short before /t/, but long before /d/, so I suppose the length difference is neutralized before voiced stops. I'll have to check, but I've read that voiceless fricatives tend to lengthen more than voiced ones (though not in Travis's dialect).
So where is the recording? :-)
I don't think English generally has phonemic vowel length. For the difference to be phonemic, you really need a minimal pair, where two existing words differ in pronunciation ONLY by vowel length.
I hear (from Wikipedia: http://en.wikipedia.org/wiki/Vowel_length#Phonemic_vowel_length_2) that Australian English does have phonemic vowel length, with minimal pairs such as "ferry" and "fairy", but those two are homonyms for me.
- Kef
>>This came up on another thread, and I thought maybe it deserved one for itself. Lazar brought up the difference between English tense vowels and the corresponding vowels in Spanish, and I though I would expand on that a bit. I don't speak Spanish, but I speak (Quebec) French. There is an allophonic tense-lax distinction in Quebec French close vowels, but no length difference between the pairs. There is also an allophonic length distinction that applies to the tense vowels, so there is are three realisations of /i/ [i, I, i:]. I find that my English /i/ more closely resembles my French [i:] than [i]. I've recorded a sample, so that you can evaluate. Here's what I'm saying:
petit, petite, tige [p@tsi] [p@tsIt] [ti:Z]
see, sit, seed, Sid, seat (I won't transcribe these, so you can decide whether they're short or long)
My impression is that /i/ is long in all cases, while /I/ is short before /t/, but long before /d/, so I suppose the length difference is neutralized before voiced stops. I'll have to check, but I've read that voiceless fricatives tend to lengthen more than voiced ones (though not in Travis's dialect).<<
You seem to have a more asymmetric vowel length allophony system than that present in the dialect here, where vowel length allophony is basically solely determined by consonants and vowels in general, including diphthongs, are all equally affected by it aside from long tense vowels aside from long /o/ having a slight tendency to diphthongize which they do not share with short tense vowels. The way you describe it, though, it seems to be a hybrid system between a system with historical English vowel length (like most English English dialects, where vowel length allophony is rather marginal) and a system with pure vowel length allophony alone (like many other North American English dialects).
I wonder, though, how "early" vowel length allophony applies in your dialect - in my dialect at least, it seems to be one of the first things that occurs between underlying forms and surface forms, such that it is often quite detached from the actual consonant phonemes which trigger it. This allows it to effectively stand in the place of elided or neutralized consonants in a quite consistent manner and allows other phonological processes to effect length itself, allowing overlong vowels. Of course, it makes it likely to be phonemicized as well... as in "with" [wIT] versus "width" [wI:T] and "breath" [br\ET] versus "breadth" [br\E:T] (I really do not know what to make of these, as these seem practically like phonemic vowel length to me, and yet overall my dialect seems to consistently treat vowel length as allophonic, albeit at a very deep level within its phonolgy).
<<Of course, it makes it likely to be phonemicized as well... as in "with" [wIT] versus "width" [wI:T] and "breath" [br\ET] versus "breadth" [br\E:T] (I really do not know what to make of these, as these seem practically like phonemic vowel length to me, and yet overall my dialect seems to consistently treat vowel length as allophonic, albeit at a very deep level within its phonolgy).>>
Interesting. "with" and "width" (and likewise "breath" and "breadth") are exact homophones for me as [wIT] and [brET]. There's no length difference for me. Where exactly does the length come from in your accent?
I have:
with ["wIT]
width ["wItT]
breath ["br\ET]
breadth ["br\EtT]
I'm Californian and I have:
with [wID]
width [witT]
breath [br\ET]
breadth [br\EtT]
I posted another sample. I realized that citation forms might not be entirely accurate, so I included some sentences using the sample words. I also included samples of the /e:, E/ and /u:, U/ pairs. My /o:/ doesn't really have a lax pair to contrast with because of the father-bother and cot-caught mergers. I think my /Q/ is probably long, but I'll have to look into it more.
This is how I transcribe the distinctions:
see [si\:]
sit [sI?]
seed [si:d]
Sid [sI:\d]
seat [si:\?]
My tense vowels have half-length by default and are lengthened by voiced stops following; my lax vowels are short by default and lengthened by voiced stops to half-long. That's how it seems to me.
This is the text of my new sample:
petit, petite, tige (French)
see, seat, seed; sit, Sid,
Sid had a seed. Every day, he would sit on his seat in front of it to see what could be seen.
Roue, route, rouge (French)
flu, flute, food; foot, good
The man played the flute to earn his food. He had the flu, so it didn’t sound very good. He had to travel on foot
way, wait, wade; wet, wed
The way across the river was deep so I had to wade. My feet got wet, but I could not wait, for that night I was to be wed.
http://media.putfile.com/accent-sample-6---vowel-length
I almost forgot:
with [wIT]
width [wI?tT]
breath [br\ET]
breadth [br\E?tT]
I was running out the door when I wrote that. Here's what it should have said:
see [si:]
sit [sI?]
seed [si:d]
Sid [sI:\d]
seat [si:\?]
My tense vowels are long by default and are shortened by following voiceless stops to half-long; my lax vowels are short by default and lengthened by voiced stops to half-long.
??<<Of course, it makes it likely to be phonemicized as well... as in "with" [wIT] versus "width" [wI:T] and "breath" [br\ET] versus "breadth" [br\E:T] (I really do not know what to make of these, as these seem practically like phonemic vowel length to me, and yet overall my dialect seems to consistently treat vowel length as allophonic, albeit at a very deep level within its phonolgy).>>
Interesting. "with" and "width" (and likewise "breath" and "breadth") are exact homophones for me as [wIT] and [brET]. There's no length difference for me. Where exactly does the length come from in your accent?>>
It's frozen allophonic vowel length. Historically these words would have been:
with /wIT/ -> [wIT]
width /wIdT/ -> [wI:tT]
breath /brET/ -> [br\ET]
breadth /brEdT/ -> [br\E:tT]
However, the [tT] sequences have become just [T], with presumably an intermediate step of [T:], which still shows up in spelling pronunciations (as at least I myself find trying to actually say [tT] very akward, and hence I use [T:] if I try to spelling-pronounce these words at all). However, that matter is that the vowel length did not change to match such but remained as if the underlying /d/ were still present.
Furthermore, excluding spelling pronunciations (which do not show up in normal speech here to begin with), pronunciations with [T:] and *especially* [tT] are practically absent from all registers in my dialect, meaning that it is hard to justify an analysis that posits a "silent" /d/ which could be used to explain the long vowels present here. Consequently, it seems as if vowel length has been phonemicized here - even though this is completely at odds with the nature of the phonology of my dialect as a whole, where vowel length can be consistently traced back to being allophonically generated by underlying forms even if the underlying forms causing it are often obscured at the surface level.
I've been thinking about vowel length in my dialect, and I don't think it's a survival of Middle English vowel length as in RP, but a modern development. First, the low back mergers and pre-/r/ mergers in my dialect seem unlikely in a phonemic-length accent. /A:/ and /O:/ seem unlikely to merge with /Q/, as do /e:/, /E/, and /{/ into /E_r/ before /r/. Second, my /3`/ seems to pattern with the short vowels, despite being long in RP, and third, /a/ and /Q/ seem to be long for me. It seems like I have a lengthening of peripheral vowels rather than a survival of length in tense vowels. This could be related to the diphthongisation of tense vowels that occurs in some accents of the area. The Ottawa Valley, for example, has diphthongal realisations of FACE and GOAT whereas monophthongs or narrower diphthongs are the norm in the city itself.
>>I've been thinking about vowel length in my dialect, and I don't think it's a survival of Middle English vowel length as in RP, but a modern development. First, the low back mergers and pre-/r/ mergers in my dialect seem unlikely in a phonemic-length accent. /A:/ and /O:/ seem unlikely to merge with /Q/, as do /e:/, /E/, and /{/ into /E_r/ before /r/. Second, my /3`/ seems to pattern with the short vowels, despite being long in RP, and third, /a/ and /Q/ seem to be long for me. It seems like I have a lengthening of peripheral vowels rather than a survival of length in tense vowels. This could be related to the diphthongisation of tense vowels that occurs in some accents of the area. The Ottawa Valley, for example, has diphthongal realisations of FACE and GOAT whereas monophthongs or narrower diphthongs are the norm in the city itself.<<
Do these dialects have diphthongal realizations of /i/ and /u/ (even quite narrow ones)? Or are these realizations limited to just /e/ and /o/ (which is the general pattern in English English and North American English dialects in which monophthongization of /e/ and /o/ has not taken place, even though very slight diphthongization of /i/ and /u/ may occur in some dialects)? It would be more likely that such would be related to diphthongal realizations if such realizations applied to /e/, /o/, /i/, and /u/ - and even then, that still raises the question of /A/, as if this were related to diphthongal realizations alone then such should be realized as [A] not [A:].
A fully diphthongal realisation of tense vowels is limited to FACE and GOAT in this area, but FLEECE sometimes has a slight off-glide in open syllables, and GOOSE has a diphthongal variant in certain situations. Some speakers seem to have a diphthongal /u/ in all situations, but I don't think that's the norm. The long value of /A/ and /a/ is still problematic though. The only thing I can think of is that the lengthening of the higher tense vowels in compensation for the loss of the off-glide made the lower peripheral vowels lengthen as well; a sort of chain-shift, I suppose. Or perhaps long /a/ is related to ae-tensing, with the lengthening being applied in non-tensing environments to make it uniform in length. (I'm grasping at straws here). That still doesn't explain long /A/.
The other possibility of course is that it is a survival of Middle English vowel length, but that for some reason /a/ has been lengthened and /3`/ shortened. Maybe it was to balance the vowel space so that all the peripheral vowels were long and all the centralized vowels were short.
|