English vowel systems and lexical sets

[Darren]

So often does our discussion veer towards the pronunciation of various English vowels ( Travis' dialect) that I reckon a new thread is in order.

cot [kʰɒʔt]
caught [kʰʊːʔt]

I could definitely distinguish /ɒ/ (COT) from /a/ (STRUT) but only with the reinforcement of the rounding of the former.

Your THOUGHT is very high... I thought GenAus had [oː] for THOUGHT...

(Praat is very easy to use in my experience. In the menu bar you can select Help>Info for a tutorial, or you could use this tutorial I found, or this one ignoring the bits about R.)

Thanks brad; I have attempted an Praat. It was my first so please be nice.



I was recording this very quietly on a laptop's inbuilt microphone and got some very wacky results, so take this with a grain of salt. Or maybe take an entire shaker of salt with a grain of this. I hesitate to draw any conclusions from this but it does suggest that

KIT = /ɪ/
NEAR = /iː/
DRESS = /e̠/
SQUARE = /e̠ː/
LAD = /æ/
BAD = /ɛ̟/ ???
NURSE = /ɵː/
FOOT = /o̟/
N/T/F/V = /oː/ (you're right; but in my defense it is the same height as /o̟/ which apparently is what my |ʊ| is, and my only conception of |ʊ| is the FOOT set)
STRUT = /ɐ̠/
S/B/F = /ɤ̠/ ?????
LOT/CLOTH = /ʌ/ ????????????

I seem to have a severe lack of low vowels other than LAD, and also all of my front vowels have become as front as possible and are bouncing off the left border of the vowel quadrilateral while the rest are just kinda hanging out. I will record some louder recordings when people can't hear me and reconsider this. Also with multiple data points rather than just one recording of each.

[bradrn]

One thing I can suggest is to record several words from each set. Then plotting their distribution could compensate somewhat for poor quality. It’s also just good practice in general, to understand the variation within a phoneme.

(Ideally you’d want to find some consonantal frameworks which each admit as many vowels as possible: the tutorials I linked suggest H__D and B__T, with P__T being another classic one.)

[Travis B.]

So often does our discussion veer towards the pronunciation of various English vowels ( Travis' dialect) that I reckon a new thread is in order.

I don't really get what is so about my dialect ─ lol. Sure, when transcribed it doesn't look like what people have been taught English looks like (i.e. idealized RP or GA which ignores much of the phonological processes present in real English). But I have long suspected that many dialects don't look like that. And from my perspective many EngE and AusE varieties are just as strange as mine may seem to EngE or AusE-speakers.

cot [kʰɒʔt]
caught [kʰʊːʔt]

I could definitely distinguish /ɒ/ (COT) from /a/ (STRUT) but only with the reinforcement of the rounding of the former.

Your THOUGHT is very high... I thought GenAus had [oː] for THOUGHT...

(Praat is very easy to use in my experience. In the menu bar you can select Help>Info for a tutorial, or you could use this tutorial I found, or this one ignoring the bits about R.)

Thanks brad; I have attempted an Praat. It was my first so please be nice.

[snip]

I was recording this very quietly on a laptop's inbuilt microphone and got some very wacky results, so take this with a grain of salt. Or maybe take an entire shaker of salt with a grain of this. I hesitate to draw any conclusions from this but it does suggest that

You should get a good headset with a built-in microphone ─ I find mine gives good results.

KIT = /ɪ/
NEAR = /iː/
DRESS = /e̠/
SQUARE = /e̠ː/
LAD = /æ/
BAD = /ɛ̟/ ???
NURSE = /ɵː/
FOOT = /o̟/
N/T/F/V = /oː/ (you're right; but in my defense it is the same height as /o̟/ which apparently is what my |ʊ| is, and my only conception of |ʊ| is the FOOT set)
STRUT = /ɐ̠/
S/B/F = /ɤ̠/ ?????
LOT/CLOTH = /ʌ/ ????????????

I seem to have a severe lack of low vowels other than LAD, and also all of my front vowels have become as front as possible and are bouncing off the left border of the vowel quadrilateral while the rest are just kinda hanging out. I will record some louder recordings when people can't hear me and reconsider this. Also with multiple data points rather than just one recording of each.

I thought AusE STRUT and START/BATH/FATHER were essentially identical except with regard to quantity...

[Darren]

One thing I can suggest is to record several words from each set. Then plotting their distribution could compensate somewhat for poor quality. It’s also just good practice in general, to understand the variation within a phoneme.

(Ideally you’d want to find some consonantal frameworks which each admit as many vowels as possible: the tutorials I linked suggest H__D and B__T, with P__T being another classic one.)

Yeah, I'm working on a better one now. I did B_T (mostly). This time I'm trying some more environments.

You should get a good headset with a built-in microphone ─ I find mine gives good results.

It seems a bit frivolous to get that just for some praating.

I thought AusE STRUT and START/BATH/FATHER were essentially identical except with regard to quantity...

So did I. They feel identical.

[bradrn]

praating

I believe that should be praten.

[Travis B.]

This is my vowel chart; I charted both h_d (for long vowels) and b_t (or in one case, b_k) words (for short vowels). There were some surprises, I must say, e.g. my /ʊ/.

[bradrn]

This is my vowel chart; I charted both h_d (for long vowels) and b_t (or in one case, b_k) words (for short vowels). There were some surprises, I must say, e.g. my /ʊ/.

1. For the diphthongs, are you plotting the starting or ending point?
2. Different words with the same vowel often seem remarkably far apart (e.g. hood/book, hayed/bait, hud/but). Any ideas why that is?

[Travis B.]

Here is an updated version that includes rounded front vowels:

[Travis B.]

This is my vowel chart; I charted both h_d (for long vowels) and b_t (or in one case, b_k) words (for short vowels). There were some surprises, I must say, e.g. my /ʊ/.

1. For the diphthongs, are you plotting the starting or ending point?

These are monophthongs in all these words.

1. Different words with the same vowel often seem remarkably far apart (e.g. hood/book, hayed/bait, hud/but). Any ideas why that is?

Phonetically they're not the same vowel. They originate in the same vowel phonemes, but they are realized distinctly.

[Travis B.]

In retrospect I probably should have chosen a different following consonant than /t/ because its glottalization affects the vowels' realization (e.g. why toot is not nearly as fronted as nude).

[Travis B.]

Oh, BTW, putting the data from Praat into a .dat file and loading it into gnuplot then exporting it as an image is doubleplusgood.

[Darren]

Here's a new and improved version for my dialect, with diphthongs. Note how the "whole" and "who'll" sets are very different from "hoed" and "who'd" even though they're at best marginally phonemic.



Sumperimposing normal IPA values on this we get:

KIT /i/
NEAR /iː/
FOOT /ö/
NORTH/THOUGHT/FORCE/CURE /oː/
DRESS /e/
SQUARE /eː/
STRUT /a̠/
START/PATH/FATHER /ɑː/
LOT/CLOTH /ɒ/
LAD /æ/
BAD /ɛː/ (this is news to me)
schwa /ə/ but with a lot of variation
NURSE /ɵː/

GOAT /ɔʏ̯/ and GOAL/POLL /ɒo̯/
FLEECE /ə̟i̯/
GOOSE /ɘʉ̟̯/ and GALL/POOL /o̞o̯/
CHOICE /oɪ̯/
PLACE /ɐɪ̯/
PRICE /ɒ̝e̞̯/
MOUTH /ɛɒ̟̯/

This was from a larger and more clearly recorded sample. I'd like to produce something even better but it'll take a while.

[Lērisama]

When I have access to a computer I am allowed to download stuff onto, I'll do one for mine. Travis's is just as incomprehensible as I would have expected, while Darren's seems sane/like mine except for the surprises they noted, and e the frint vowel chain shift thing.

[Travis B.]

I noticed that the allophones vary more significantly with surrounding consonants than I had previously thought. In particular, after /n/ long /uː/ becomes [øː] and long /oʊ/ becomes [ɶː] (!).

Edit: I take that back - it is tote which is way more open than node, not the other way around. Still odd though...

[Travis B.]

I think I know what's going on with tote -- /oʊ/ after a coronal and before a glottalized /t/ at the end of an utterance becomes [ɶ] but if the /t/ has coronal articulation or is not at the end of an utterance is [ø̞].

[Travis B.]

About hud, I think the reason STRUT is getting fronted in it is the following /d/ (I checked and this is also the case for bud), because I don't get such fronting in hug or bug.

[Travis B.]

Moral of the story ─ choose the words you test very carefully, as they may have much unintended influence on the realization of vowels...

[Creyeditor]

[...]


I was recording this very quietly on a laptop's inbuilt microphone and got some very wacky results, so take this with a grain of salt. Or maybe take an entire shaker of salt with a grain of this. I hesitate to draw any conclusions from this but it does suggest that

KIT = /ɪ/
NEAR = /iː/
DRESS = /e̠/
SQUARE = /e̠ː/
LAD = /æ/
BAD = /ɛ̟/ ???
NURSE = /ɵː/
FOOT = /o̟/
N/T/F/V = /oː/ (you're right; but in my defense it is the same height as /o̟/ which apparently is what my |ʊ| is, and my only conception of |ʊ| is the FOOT set)
STRUT = /ɐ̠/
S/B/F = /ɤ̠/ ?????
LOT/CLOTH = /ʌ/ ????????????
[...]

If this was any other language but English (and all your length and rounding transcriptions are correct), I would probably transcribe the sounds as follows:

KIT = /i/
NEAR = /iː/
DRESS = /e/
SQUARE = /eː/
LAD = /æ/
BAD = /ɛ/
NURSE = /ʉː/
FOOT = /u/
N/T/F/V = /uː/
STRUT = /ɑ/
S/B/F = /ɤ/
LOT/CLOTH = /ʌ/

I somehow suspect that your S/B/F vowel is long. If so, it could be /ɤː/ instead. The typologically surprising thing for me is really the lack of a back mid rounded vowel, so something like /ɔ/ or /o/. Otherwise it's pretty nice and symetrical. I would probably expect some kind of backness vowel harmony

[Darren]

If this was any other language but English (and all your length and rounding transcriptions are correct),

Ah, but they aren't! I forgot to transcribe length on BAD and S/B/F.

I somehow suspect that your S/B/F vowel is long. If so, it could be /ɤː/ instead. The typologically surprising thing for me is really the lack of a back mid rounded vowel, so something like /ɔ/ or /o/. Otherwise it's pretty nice and symetrical. I would probably expect some kind of backness vowel harmony

I suspect part of the problem is that while there isn't any lip rounding on any of the back vowels, FOOT, N/T/F/C and LOT/CLOTH all sound rounded. It may be something to do with sulcalisation.

[Lērisama]

I suspect part of the problem is that while there isn't any lip rounding on any of the back vowels, FOOT, N/T/F/C and LOT/CLOTH all sound rounded. It may be something to do with sulcalisation.

At least in my English, only FOOT¹ and the final element of diphthongs are rounded, but the rest of the non low² back vowels are sulcalised. I just write them as rounded for simplicity.

¹And GOOSE, but it starts with FOOT
² I am counting LOT as non low, but CUT as low