English questions

[Travis B.]

On that note, should it be regarded as a criterion for phonemicity that a hypothetical form should be able to be borrowed, coined, or created via onomatopoeia as long as it fits into a variety's phonotactics? This is a big part of why I presently am hesitant to regard things like vowel quantity and nasality as phonemic in my dialect because they cannot be created de novo but rather can only be derived from historical underlying forms, and borrowed and coined words behave like such historical underlying forms. Conversely, I do regard consonant quantity as phonemic in my dialect specifically because it can be borrowed (e.g. the geminate in pizza, which is contrastive with the lack of a geminate in Nazi) and in cases cannot be explained in terms of historical underlying forms except by appealing to likely ahistorical back-formed 'phonemic' forms (e.g. the geminate in raccoon or in Kyoto).

[Lērisama]

Does anyone else's English dialects have phonemic contrasts between 'rhotic' vowels and sequences of non-'rhotic' vowels and /r/?

Yes, but not in the same words. I have synchronic length distinction, the same the the one between former VC and VrC before /rV/ as well, in words like ⟨merry⟩ [ˈmɛɹ̠ɪj] and ⟨Mary⟩ [ˈmɛːɹ̠ɪj], and near-minimal pairs in words like ⟨mirror⟩ [ˈmɪɹ̠ə] and ⟨nearer⟩ [ˈnɪːɹ̠ə]. I'm not sure how to analyse this. It doesn't work neatly with the ‘extended VC’ analysis, where long vowels are /Vr/¹, because you'd have to posit as geminate /r/ or a phonemic syllable boundary, while the face value vowel length analysis isn't particularly nice with hiatus.

¹ Intrusive /r/, patterning in the same way as intrusive /j/ and /w/, as well as my native speaker intuition² both point to this
² Weirdly, native speakers here seem to treat diphthongs as phonemic, but long vowels as /Vr/, although this may be more to do with the spelling than anything else

[Travis B.]

Does anyone else's English dialects have phonemic contrasts between 'rhotic' vowels and sequences of non-'rhotic' vowels and /r/?

Yes, but not in the same words. I have synchronic length distinction, the same the the one between former VC and VrC before /rV/ as well, in words like ⟨merry⟩ [ˈmɛɹ̠ɪj] and ⟨Mary⟩ [ˈmɛːɹ̠ɪj], and near-minimal pairs in words like ⟨mirror⟩ [ˈmɪɹ̠ə] and ⟨nearer⟩ [ˈnɪːɹ̠ə]. I'm not sure how to analyse this. It doesn't work neatly with the ‘extended VC’ analysis, where long vowels are /Vr/¹, because you'd have to posit as geminate /r/ or a phonemic syllable boundary, while the face value vowel length analysis isn't particularly nice with hiatus.

¹ Intrusive /r/, patterning in the same way as intrusive /j/ and /w/, as well as my native speaker intuition² both point to this
² Weirdly, native speakers here seem to treat diphthongs as phonemic, but long vowels as /Vr/, although this may be more to do with the spelling than anything else

I personally am opposed to /Vr/ analyses of non-rhotic varieties because they require /rr/ geminates in varieties that otherwise (to my knowledge) lack geminates and also complicate the analysis of lettER as it involves intrusive-r (outside of non-rhotic NAE varieties) yet is short rather than long. Additionally, many of the cases where they require /rr/ geminates are cases where in reality the preceding vowel was always long.

[Travis B.]

I should note that in classic NAE varieties with a full Mary-merry-marry merger all vowels before /r/ are 'rhotic vowels', in that they form a distinct vowel system that is independent of the other vowels and do not undergo the sound changes that other vowels undergo. This is shown with the NCVS where 'rhotic vowels' did not undergo the NCVS with the exception that in many NCVS-affected varieties START got fronted (which is not true of the dialect here).

[Lērisama]

I personally am opposed to /Vr/ analyses of non-rhotic varieties because they require /rr/ geminates in varieties that otherwise (to my knowledge) lack geminates and also complicate the analysis of lettER as it involves intrusive-r (outside of non-rhotic NAE varieties) yet is short rather than long. Additionally, many of the cases where they require /rr/ geminates are cases where in reality the preceding vowel was always long.

I am aware of your opposition to it, and I did note exactly that problem¹, although I don't think that LETTER is a problem, because unstressed NURSE is realised as a schwa, and the analysis would require all vowels to have a following consonant anyway, so why except final /ə/? I just think the problems with assuming vowel length is phonemic are marginally worse: of ⟨mirror⟩ and ⟨nearer⟩ would you choose /mɪrə nɪːrə/, /mɪə nɪːə/, /mɪə nɪːrə/ or /mɪrə nɪːrə/ – all of them would lead to surface [ˈmɪɹ̠ə ˈnɪːɹ̠ə]?


¹

you'd have to posit as geminate /r/ or a phonemic syllable boundary

[Travis B.]

I personally am opposed to /Vr/ analyses of non-rhotic varieties because they require /rr/ geminates in varieties that otherwise (to my knowledge) lack geminates and also complicate the analysis of lettER as it involves intrusive-r (outside of non-rhotic NAE varieties) yet is short rather than long. Additionally, many of the cases where they require /rr/ geminates are cases where in reality the preceding vowel was always long.

I am aware of your opposition to it, and I did note exactly that problem¹, although I don't think that LETTER is a problem, because unstressed NURSE is realised as a schwa, and the analysis would require all vowels to have a following consonant anyway, so why except final /ə/? I just think the problems with assuming vowel length is phonemic are marginally worse: of ⟨mirror⟩ and ⟨nearer⟩ would you choose /mɪrə nɪːrə/, /mɪə nɪːə/, /mɪə nɪːrə/ or /mɪrə nɪːrə/ – all of them would lead to surface [ˈmɪɹ̠ə ˈnɪːɹ̠ə]?


¹

you'd have to posit as geminate /r/ or a phonemic syllable boundary

On second thought, my preferred analysis for your mirror versus nearer would be /mɪrə nɪːrə/ while my preferred analysis of just near would be /nɪːr/ not /nɪr/. This enables there to be separate treatment of NURSE and lettER as /ɜːr/ and /ər/ without one being treated as a 'stressed version' or 'unstressed version' of the other (because why should NURSE be singled out as a stressed lettER when other vowels can be reduced to lettER as well?). This also avoids the problems of potential analyses where all intervocalic rhotics after long vowels are analyzed as non-consonants. Of course, I would tend to regard lettER as the morpheme-final counterpart to commA.

[Lērisama]

On second thought, my preferred analysis for your mirror versus nearer would be /mɪrə nɪːrə/ while my preferred analysis of just near would be /nɪːr/ not /nɪr/. This enables there to be separate treatment of NURSE and lettER as /ɜːr/ and /ər/ without one being treated as a 'stressed version' or 'unstressed version' of the other (because why should NURSE be singled out as a stressed lettER when other vowels can be reduced to lettER as well?). This also avoids the problems of potential analyses where all intervocalic rhotics after long vowels are analyzed as non-consonants. Of course, I would tend to regard lettER as the morpheme-final counterpart to commA.

So your plan is both vowel length and /Vr/? That seems rather overengineered to me. I see three main problems with it, starting with the most severe:

• If I understand you correctly, you're treating COMMA & LETTER differently, which is properly wrong. I couldn't tell you which words belong to which set without visualising the spelling, and I'm not aware of any non-rhotic dialect that distinguishes them
• What do you do with historical /VrC/ words? Phonemicising them as /VːC/ is what I'd expect, but then exactly what are the environments in which you'd assume an underlying /Vːr/. Without knowing your reasoning behind it, it feels like you're assuming postvocalic /r/ based on your rhotic-dialect-based intuition of them, which this dialect doesn't have
• I don't like the fact that you in effect have a set of long vowels that can only go before /r/ in almost complimentary distribution with a set of short vowels that can't, but that's more of an aesthetic problem

I worry at this point that we might be using slightly different reference dialects, so I have the following table of what I'm assuming for vowels, in case that is a problem anywhere

More: show

Lexical sets	My current analysis	Stressed open syllable¹²	Stressed closed syllable	Unstressed final	Unstressed non-final
TRAP	æ³	X	✓	X	X
STRUT	ʌ⁴	X	✓	X	X
DRESS	ɛ	X	✓	X	X
LOT CLOTH	ɔ	X	✓	X	X
KIT	ɪ	X	✓	X	✓
FOOT	ɵ	X	✓	X	X
COMMA LETTER	ə	X	X	✓	✓
BATH PALM START	ɑr⁵ ɑː	✓	✓	X	X
SQUARE	ɛr ɛː	✓	✓	X	X
THOUGHT NORTH FORCE	ɔr ɔː	✓	✓	X	X
NEAR	ɪr ɪː	✓	✓	X	X
CURE⁶	ɵː	✓	✓	X	X
NURSE	ər əː	✓	✓	X	X
MOUTH	a⁷w	✓	✓	X	X
PRICE	ɑ⁵j	✓	✓	X	X
FACE	ɛj	✓	✓	X	X
CHOICE	ɔj	✓	✓	X	X
GOAL⁸	ɔw	✓	✓	X	X
FLEECE HAPPY	ɪj	✓	✓	✓	X
GOOSE	ɵw	✓	✓	X	X
GOAT	əw	✓	✓	X	X

¹ Assuming a single consonant after a stressed vowel closes a syllable, because that makes this significantly easier
² Including before /r/
³ If I were starting over, I'd use /a/, but that'd be needlessly confusing here
⁴ Would use /ɑ/, because it's as close to the phonetic value and matches the lengthened/rhotic version in START, but for the same reason as above
⁵ Same as STRUT, but see above at ⁴
⁶ Falling out of use in favour of /ɔr/ɔː/. Only found after /j/
⁷ Same as TRAP, but see above at ³
⁸ i.e. GOAT + /l/, now marginally phonemic for me, I think, unless you want a geminate /l/ in [ˈhɔwɫɪj] ⟨wholely/holey⟩

Edit: I may have got carried away with the table and left a bullet point unfinished

Edit edit: thank you Quinterbeck, you are indeed correct about the error

[quinterbeck]

I worry at this point that we might be using slightly different reference dialects, so I have the following table of what I'm assuming for vowels, in case that is a problem anywhere

More: show

Lexical sets	My current analysis	Stressed open syllable¹²	Stressed closed syllable	Unstressed final	Unstressed non-final
TRAP	æ³	X	✓	X	X
STRUT	ʌ⁴	X	✓	X	X
DRESS	ɛ	X	✓	X	X
LOT CLOTH	ɔ	X	✓	X	X
KIT	ɪ	X	✓	X	✓
FOOT	ɵ	X	✓	X	X
COMMA LETTER	ə	X	X	✓	✓
BATH PALM START	ɑr⁵ ɑː	✓	✓	X	X
SQUARE	ɛr ɛː	✓	✓	X	X
THOUGHT NORTH FORCE	ɔr ɔː	✓	✓	X	X
NEAR	ɪr ɪː	✓	✓	X	X
CURE⁶	ɵː	✓	✓	X	X
NURSE	ər əː	✓	✓	X	X
MOUTH	a⁷w	✓	✓	X	X
PRICE	ɑ⁵j	✓	✓	X	X
FACE	ɛj	✓	✓	X	X
CHOICE	ɔj	✓	✓	X	X
GOAL⁸	ɔw	✓	✓	X	X
FLEECE HAPPY	ɪj	X	✓	✓	X
GOOSE	ɵw	✓	✓	X	X
GOAT	əw	✓	✓	X	X

¹ Assuming a single consonant after a stressed vowel closes a syllable, because that makes this significantly easier
² Including before /r/
³ If I were starting over, I'd use /a/, but that'd be needlessly confusing here
⁴ Would use /ɑ/, because it's as close to the phonetic value and matches the lengthened/rhotic version in START, but for the same reason as above
⁵ Same as STRUT, but see above at ⁴
⁶ Falling out of use in favour of /ɔr/ɔː/. Only found after /j/
⁷ Same as TRAP, but see above at ³
⁸ i.e. GOAT + /l/, now marginally phonemic for me, I think, unless you want a geminate /l/ in [ˈhɔwɫɪj] ⟨wholely/holey⟩

Is there an error in your FLEECE/HAPPY row? I would put a tick in the first column also, cf. bee, tree, plea, key, free etc.

[Raphael]

Can you write "businesspeople" as one word? My Firefox spellcheck says yes, but my Libreoffice spellcheck says no.

[Travis B.]

On second thought, my preferred analysis for your mirror versus nearer would be /mɪrə nɪːrə/ while my preferred analysis of just near would be /nɪːr/ not /nɪr/. This enables there to be separate treatment of NURSE and lettER as /ɜːr/ and /ər/ without one being treated as a 'stressed version' or 'unstressed version' of the other (because why should NURSE be singled out as a stressed lettER when other vowels can be reduced to lettER as well?). This also avoids the problems of potential analyses where all intervocalic rhotics after long vowels are analyzed as non-consonants. Of course, I would tend to regard lettER as the morpheme-final counterpart to commA.

So your plan is both vowel length and /Vr/? That seems rather overengineered to me. I see three main problems with it, starting with the most severe:

• If I understand you correctly, you're treating COMMA & LETTER differently, which is properly wrong. I couldn't tell you which words belong to which set without visualising the spelling, and I'm not aware of any non-rhotic dialect that distinguishes them
• What do you do with historical /VrC/ words? Phonemicising them as /VːC/ is what I'd expect, but then exactly what are the environments in which you'd assume an underlying /Vːr/. Without knowing your reasoning behind it, it feels like you're assuming postvocalic /r/ based on your rhotic-dialect-based intuition of them, which this dialect doesn't have
• I don't like the fact that you in effect have a set of long vowels that can only go before /r/ in almost complimentary distribution with a set of short vowels that can't, but that's more of an aesthetic problem

From some more thought, I would only analyze an /r/ before another vowel or morpheme-finally, so, say, father would not have /r/ but words such as draw and, actually, comma itself would end in /r/. The rule would simply be that long vowels and the schwa could not exist in hiatus or morpheme-finally, and /r/ would be elided when not preceding a vowel (ignoring word boundaries).

I worry at this point that we might be using slightly different reference dialects, so I have the following table of what I'm assuming for vowels, in case that is a problem anywhere

The main difference between how I've seen things and how you detailed things is that I see intervocalic consonants (and certain consonant sequences that effectively behave in such environments as if they were single consonants such as /nt/ and /rt/) before unstressed vowels as ambisyllabic rather than belonging to a particular syllable, which predicts many behaviors of my own dialect and other NAE varieties very well.

[Lērisama]

Can you write "businesspeople" as one word? My Firefox spellcheck says yes, but my Libreoffice spellcheck says no.

I did a double take and then decided it was fine, so that fits.

[Raphael]

Can you write "businesspeople" as one word? My Firefox spellcheck says yes, but my Libreoffice spellcheck says no.

I did a double take and then decided it was fine, so that fits.

Thank you.

[jal]

Can you write "businesspeople" as one word? My Firefox spellcheck says yes, but my Libreoffice spellcheck says no.

Instead of trusting spelling checkers, you could check a dictionary 🤷‍♂️.


JAL

[Raphael]

Instead of trusting spelling checkers, you could check a dictionary .


JAL

You are right, of course. And now I'm feeling rightly embarrassed.

[Travis B.]

In contemporary usage, most cases where you can have the final morpheme -man outside of surnames can be replaced with -person, and likewise -men with -people.

[Herra Ratatoskr]

Just curious, does anyone else use -folk as a less formal plural of -person? So formally it would be "businessperson/businesspeople", but casually it would be "businessperson/businessfolk"? Or am I just being weird?

[Travis B.]

Just curious, does anyone else use -folk as a less formal plural of -person? So formally it would be "businessperson/businesspeople", but casually it would be "businessperson/businessfolk"? Or am I just being weird?

Using -folk in the place of -people sounds to me like one is deliberately trying to be, well, folksy.

[Raphael]

Just curious, does anyone else use -folk as a less formal plural of -person? So formally it would be "businessperson/businesspeople", but casually it would be "businessperson/businessfolk"? Or am I just being weird?

"businessfolk" does sound rather weird to me.

[Travis B.]

Just curious, does anyone else use -folk as a less formal plural of -person? So formally it would be "businessperson/businesspeople", but casually it would be "businessperson/businessfolk"? Or am I just being weird?

"businessfolk" does sound rather weird to me.

Same here.

[Herra Ratatoskr]

Just curious, does anyone else use -folk as a less formal plural of -person? So formally it would be "businessperson/businesspeople", but casually it would be "businessperson/businessfolk"? Or am I just being weird?

"businessfolk" does sound rather weird to me.

Same here.

Huh, neat to know. Thanks for the responses, fellow Zbeebfolk!