Six-vowel system for English

[KathTheDragon]

(in the other thread)

This whole venture is very focussed on very specific dialects. It breaks down in much of England, for instance, where positing underlying /r/ for what have become long monophthongs is unjustified.

... though, playing devil's advocate a bit, you could argue that that underlying /r/ surfaces whenever those vowels are followed by another vowel.

You could but this requires contrastive syllabification at some level to account for e.g. starry /ˈstɑːri/ vs tarry /ˈtari/, since positing a geminate /r/ (and only geminate /r/) is unjustified. Either way it's strikingly ad-hoc.

[Ryusenshi]

SSBE as analysed by Geoff Lindsey is solidly 7-vowel based, crossing /ɪ ɛ a ɔ ɵ ə ʌ/ with [-j -w -ː(ɹ)]. It wouldn't stretch it too much to consider commA /ə/ and STRUT /ʌ/ as one vowel, since they each appear only in unstressed and stressed syllables respectively. Boom, 6 vowel English.

This is quite similar to what I was saying, except with IPA symbols instead of plain Latin letters (plus schwa), and a "chroneme" /ː/ instead of /h/.

I'm honestly rather suspicious of combining the STRUT and commA vowels. I don't perceive a specific similarity between them, and when I secondarily stress commA I get NURSE.

As usual, it depends on the accent. Many accents have something close to [ə] for STRUT. Conversely, several accents (Australian, Cockney, old U-RP) have a very open [ɐ] for a word-final schwa.

[Travis B.]

This kind of thing fails spectacularly for me personally, since not only do I have a limited set of phonemic diphthongs, namely /ae əe ɑɔ ɔɪ/, I also do not preserve historical vowel length. Consequently "/e o i u/" are equal in length to "/ej ow ij uw/", which are not diphthongs (except for "/ow/" optionally finally and in hiatus). Also, TRAP does not map well to "/a/", since my TRAP is much closer to [e] than to [a]; rather, LOT/PALM is a much better fit for "/a/" (and START would be /ar/). That raises the question though of what to call TRAP. Also, treating STRUT as a stressed COMMA does not work out well due to my STRUT being a back vowel; rather, in many positions, KIT fits the role of a stressed COMMA due to the most common allophone of KIT, except morpheme-finally, being identical to KIT.

[Richard W]

SSBE as analysed by Geoff Lindsey is solidly 7-vowel based, crossing /ɪ ɛ a ɔ ɵ ə ʌ/ with [-j -w -ː(ɹ)]. It wouldn't stretch it too much to consider commA /ə/ and STRUT /ʌ/ as one vowel, since they each appear only in unstressed and stressed syllables respectively. Boom, 6 vowel English.

I think this scheme has a slight hiccup.

[quinterbeck]

SSBE as analysed by Geoff Lindsey is solidly 7-vowel based, crossing /ɪ ɛ a ɔ ɵ ə ʌ/ with [-j -w -ː(ɹ)]. It wouldn't stretch it too much to consider commA /ə/ and STRUT /ʌ/ as one vowel, since they each appear only in unstressed and stressed syllables respectively. Boom, 6 vowel English.

I think this scheme has a slight hiccup.

Oh, good point (and nicely put).

[Travis B.]

SSBE as analysed by Geoff Lindsey is solidly 7-vowel based, crossing /ɪ ɛ a ɔ ɵ ə ʌ/ with [-j -w -ː(ɹ)]. It wouldn't stretch it too much to consider commA /ə/ and STRUT /ʌ/ as one vowel, since they each appear only in unstressed and stressed syllables respectively. Boom, 6 vowel English.

I think this scheme has a slight hiccup.

Oh, good point (and nicely put).

Here any "unstressed" syllables that receive STRUT are actually secondary-stressed. Same with KIT and DRESS.

[Nortaneous]

You could but this requires contrastive syllabification at some level to account for e.g. starry /ˈstɑːri/ vs tarry /ˈtari/, since positing a geminate /r/ (and only geminate /r/) is unjustified. Either way it's strikingly ad-hoc.

Doesn't require contrastive syllabification, only requires relevance of morpheme boundaries.

SSBE as analysed by Geoff Lindsey is solidly 7-vowel based, crossing /ɪ ɛ a ɔ ɵ ə ʌ/ with [-j -w -ː(ɹ)]. It wouldn't stretch it too much to consider commA /ə/ and STRUT /ʌ/ as one vowel, since they each appear only in unstressed and stressed syllables respectively. Boom, 6 vowel English.

This is where I got the idea for an eight-vowel analysis of Mid-Atlantic AmE - his analysis is structurally the same as mine except for NURSE (/əː/ vs. /ɨ˞ /) and the SSBE THOUGHT-NORTH merger. (/æ a e ʌ o i ɨ˞ u/ + /aj ej oj ij æw ʌw uw ar er or ir/) Unfortunately, the eight vowels can't be reduced to six without either major changes or ignoring the poor-pour contrast. Phonotactically it'd be nice to be able to call THOUGHT /ɑw/, eliminating /o/ entirely, and while calling CHOICE /uj/ is basically reasonable, the need for maintaining poor-pour makes it difficult to deal with the rhotic series.

I guess something like START = /ær/, NORTH = /ar/, TOUR = /ur/ plus backing could work, but it's not ideal.

/ɨ˞ / can be reduced to /ʌr/ with only a little loss of phonotactic parsimony.

There are probably a lot of differences in the unstressed vowels, but that isn't relevant here.

[Travis B.]

I remember Nort commenting a while back on how this kind of thing shows how lects with relatively similar surface forms can have radically different underlying phonologies (vis-a-vis his lect versus my lect).

[Nortaneous]

I remember Nort commenting a while back on how this kind of thing shows how lects with relatively similar surface forms can have radically different underlying phonologies (vis-a-vis his lect versus my lect).

Right, the main advantages of the eight-vowel analysis are some phonetic subtleties and a handful of minor sound changes - if those don't obtain, it's possible for the traditional three-diphthong analysis to be better, and that practically doubles the size of the vowel system.

[Travis B.]

I remember Nort commenting a while back on how this kind of thing shows how lects with relatively similar surface forms can have radically different underlying phonologies (vis-a-vis his lect versus my lect).

Right, the main advantages of the eight-vowel analysis are some phonetic subtleties and a handful of minor sound changes - if those don't obtain, it's possible for the traditional three-diphthong analysis to be better, and that practically doubles the size of the vowel system.

Well, the three-diphthong analysis basically applies to lects where the tense vowels are all monophthongs. The traditional analysis is a five-diphthong one, as it has /eɪ/ and /oʊ/ (/əʊ/ in traditional RP analyses) as diphthongs.

[anteallach]

I think this scheme has a slight hiccup.

Oh, good point (and nicely put).

Here any "unstressed" syllables that receive STRUT are actually secondary-stressed. Same with KIT and DRESS.

Ah, but does the second syllable of hiccup (say) have secondary stress (if it even does; I'm not at all convinced) because it has an unreduced vowel or vice versa? And how do those who perceice schwa and STRUT to be the same vowel perceive their pronunciation of hiccup?

In my speech schwa is marginally stressable anyway: I often use it in pronouncing apparently vowelless acronyms, and /məz/ is my normal pronunciation of Ms. Also for some reason the word but seems to be losing its strong form, and can be /bət/ even when stressed.

[anteallach]

You could but this requires contrastive syllabification at some level to account for e.g. starry /ˈstɑːri/ vs tarry /ˈtari/, since positing a geminate /r/ (and only geminate /r/) is unjustified. Either way it's strikingly ad-hoc.

Doesn't require contrastive syllabification, only requires relevance of morpheme boundaries.

No, because there are examples where the ones with the long vowels are monomorphemic, most famously Mary as opposed to merry. If not using contrastive syllabification you might have to make them /merri(j)/ and /meri(j)/ respectively, which amusingly is the reverse of where the orthography puts the double r.

[KathTheDragon]

You could but this requires contrastive syllabification at some level to account for e.g. starry /ˈstɑːri/ vs tarry /ˈtari/, since positing a geminate /r/ (and only geminate /r/) is unjustified. Either way it's strikingly ad-hoc.

Doesn't require contrastive syllabification, only requires relevance of morpheme boundaries.

No, because there are examples where the ones with the long vowels are monomorphemic, most famously Mary as opposed to merry. If not using contrastive syllabification you might have to make them /merri(j)/ and /meri(j)/ respectively, which amusingly is the reverse of where the orthography puts the double r.

Except here the geminate /r/ is ad-hoc, since there are no other geminates. I think you can still say "Mary" is /ˈmejri/ with /ej/ > [ɛː] before /r/, however.

[bradrn]

You could bring my Australian English idiolect down to 6 vowels /i ʊ e æ a ɔ/ without too much of a stretch

/i ʊ e æ a ɔ/ = [i ʊ e æ~ɛ ä ɔ] (KIT/happY FOOT DRESS TRAP STRUT LOT)
/aj æj æw ij ʊw ʊj ɔw/ = [ɑe̯ æi̯ æu̯ ɪi̯ ʉu̯ ʊi̯ ɒʉ̯] (PRICE FACE MOUTH FLEECE GOOSE CHOICE GOAT)
/iɹ ʊɹ eɹ æɹ aɹ ɔɹ/ = [ɪː ʊː eː æː äː ə(ː)] (NEAR THOUGHT/NORTH/FORCE SQUARE BAD PALM/START NURSE/lettER/commA)

This requires you to count most instances of [ə] as unstressed /ɔɹ/ which is wack but doesn’t create any inconsistencies, and stressed [əː] is treated as /ɔɹ/ which would otherwise never occur. Otherwise it kinda makes sense cause it's /ɹ/-intrusive (counting [æː] as /æɹ/ only works cause it only occurs pre-consonantally) and non-rhotic long vowels are noticeably diphthongised.

I just saw this, and only a slight variation is necessary for my Australian English:

/i ʊ e æ a o/ (KIT/happY FOOT DRESS TRAP/BAD STRUT/lettER/commA LOT/THOUGHT)
/aj æj æw ij ʊw ʊj ow/ (PRICE FACE MOUTH FLEECE GOOSE CHOICE GOAT)
/iɹ ʊɹ eɹ aɹ oɹ/ (NEAR NURSE SQUARE PALM/START NORTH/FORCE)

Sample: /ol hjʊwman bijiŋz aɹ born fɹij ænd ijkwal in diɡniti ænd ɹajts/. Sounds a bit strange, but actually not overly so.

[Nortaneous]

I think you can still say "Mary" is /ˈmejri/ with /ej/ > [ɛː] before /r/, however.

Yes, this is reasonable, cf. /ʌw/ > [oˤ] before /l/

[Travis B.]

Oh, good point (and nicely put).

Here any "unstressed" syllables that receive STRUT are actually secondary-stressed. Same with KIT and DRESS.

Ah, but does the second syllable of hiccup (say) have secondary stress (if it even does; I'm not at all convinced) because it has an unreduced vowel or vice versa? And how do those who perceice schwa and STRUT to be the same vowel perceive their pronunciation of hiccup?

In my speech schwa is marginally stressable anyway: I often use it in pronouncing apparently vowelless acronyms, and /məz/ is my normal pronunciation of Ms. Also for some reason the word but seems to be losing its strong form, and can be /bət/ even when stressed.

For me at least, hiccup takes COMMA in its second syllable.

[Pabappa]

i can imagine a world where kindergarten teachers give us phonemic stressed /ə/ in [bət] to stop the children from laughing so much. but I dont think that particular word is given syntactic stress often enough to make a difference, and also that kids will laugh anyway.

personally, i have stressed schwa in pull full wool bull, which is extremely marginal, but a schwa is a schwa, and it was enough for me to think growing up that **that** vowel was the stressed form of the schwa. I had seen pronunciation guides in school textbooks, and had some very basic linguistic knowledge even then, but I didnt know the IPA until 5th grade, so i knew that pull full wool bull were part of a wider set of words that included book, took, etc and for some speakers also other words like roof and room.

i assume that's the general pronunciation around New England because I couldnt have gotten it anywhere else and my accent has always been fairly average for where i grew up. but i dont like to quibble about matters of analysis when one system is as good as another ... i would only object to someone telling me that COMMA=STRUT is the only valid analysis, or that pull full wool bull are just showing the stressed allophones of syllabic consonants (which i'd call circular reasoning), and so on.

[Travis B.]

Pull wool full bull for me all have a rounded nucleus and a rounded offglide (from vocalized /l/) or even just a rounded overlong monophthong (where vocalized /l/ merges with the nucleus), so those cannot simply be cases of stressed schwas (since schwas followed by /l/ are unrounded for me).

[Richard W]

Pull wool full bull for me all have a rounded nucleus and a rounded offglide (from vocalized /l/) or even just a rounded overlong monophthong (where vocalized /l/ merges with the nucleus), so those cannot simply be cases of stressed schwas (since schwas followed by /l/ are unrounded for me).

Do (m)any people rime them with gull dull hull cull?

Having /'ʌ' ~ ə/ has been claimed as a British difference form American. (Writing /ʌ/ for /ɐ/ annoys me, especially when I see it applied to Asian languages.)

[Travis B.]

Pull wool full bull for me all have a rounded nucleus and a rounded offglide (from vocalized /l/) or even just a rounded overlong monophthong (where vocalized /l/ merges with the nucleus), so those cannot simply be cases of stressed schwas (since schwas followed by /l/ are unrounded for me).

Do (m)any people rime them with gull dull hull cull?

I certainly don't.

Having /'ʌ' ~ ə/ has been claimed as a British difference form American. (Writing /ʌ/ for /ɐ/ annoys me, especially when I see it applied to Asian languages.)

"/ʌ/" for /ɐ/ particularly annoys me because I have true [ʌ] for /ʌ/.