Linguistic Miscellany Thread

[bradrn]

I bring up Goldberg because her framework allows meanings to be assigned to constructions rather than just words; she also recognizes prototype effects. I think the prototype here is <person A> doesn't <do some positive thing> like <person B who does it really well> with some slippage about the positivity.

I’ve always liked Construction Grammar, and it’s probably the best option here. But this template may need to be loosened a bit, since other structures show this implicature too:

You don’t kiss like him.
You don’t kiss like he does.
You don’t kiss the same way as him.
You don’t kiss the same way he does.

If I’m not mistaken, the implication gradually weakens as you go down this list — I’d be most likely to say the last sentence here if I wanted to make a straight comparison.

Incidentally, I feel this example also highlights the weakness of Construction Grammar, which is that it ignores compositionality. ‘A doesn’t X like B’ may has a strong implication, but its basic meaning is predictable purely from its structure as a clause with an adverbial. That’s something which, as far as I’m aware, Construction Grammar doesn’t even attempt to model.

(I seem to recall someone on this board (possibly dhok?) mentioning a derivative theory which could handle these things, but I never looked into it properly and am not sure I can find the post again.)

[zompist]

Incidentally, I feel this example also highlights the weakness of Construction Grammar, which is that it ignores compositionality. ‘A doesn’t X like B’ may has a strong implication, but its basic meaning is predictable purely from its structure as a clause with an adverbial. That’s something which, as far as I’m aware, Construction Grammar doesn’t even attempt to model.

I'd put it the other way around: Construction Grammar takes up where compositionality leaves off. The thing about Daniel's example is precisely that the extra implication (B kisses well) is not predictable from any of its parts. The raw statement is simply what your later examples had: B's way is not identical to A's way.

[Moose-tache]

One of my pet peeves is when people treat the word "schwa" as an alternate name for the STRUT vowel

[bradrn]

One of my pet peeves is when people treat the word "schwa" as an alternate name for the STRUT vowel

On the other hand, it appears that some dialects, especially American, genuinely do merge them: https://www.englishspeechservices.com/b ... n-english/. (Not mine, though.)

[Moose-tache]

To me, "schwa" means [ə], a phonetic form that is very common as a reduction of various vowels, including but not limited to the STRUT vowel. But if you pronounce the STRUT vowel carefully, it's no longer [ə]. You might as well call the DRESS vowel "schwa," because it can be realized as [ə] in unstressed syllables if someone speaks hastily enough.

I guess you could call it "schwa" if we all agree to just treat schwa as a synonym for STRUT, and abandon its connection to any specific phone. But that would annoy me, personally, so let's not do it.

Dr. Lindsey's article is interesting, because it holds true for me as well, but with the symbols reversed: abandon /ə/ as a separate phoneme, and just use /ʌ/ for both vowels in words like "Russia."

[bradrn]

To me, "schwa" means [ə], a phonetic form that is very common as a reduction of various vowels, including but not limited to the STRUT vowel. But if you pronounce the STRUT vowel carefully, it's no longer [ə]. You might as well call the DRESS vowel "schwa," because it can be realized as [ə] in unstressed syllables if someone speaks hastily enough.

I guess you could call it "schwa" if we all agree to just treat schwa as a synonym for STRUT, and abandon its connection to any specific phone. But that would annoy me, personally, so let's not do it.

I suppose the problem is that ‘schwa’ can really mean either of two things. From a phonetic perspective, it refers to the sound [ə], which in these dialects is used for both STRUT and reduced vowels. From a phonological perspective, it refers specifically to those reduced vowels, with STRUT being a separate archiphoneme (even if on the phonetic level it can be realised the same way).

Dr. Lindsey's article is interesting, because it holds true for me as well, but with the symbols reversed: abandon /ə/ as a separate phoneme, and just use /ʌ/ for both vowels in words like "Russia."

To me, that sounds like you just don’t have the merger. For me, ‘Russia’ is also /ˈɹɐʃɐ/. By contrast, the XKCD comic contains two vowels: /dɐɡz stɐk kəz əv‿ə ˈtɐnl̩ əbˈstɹɐkʃən/.

[zompist]

One of my pet peeves is when people treat the word "schwa" as an alternate name for the STRUT vowel

I'm not sure I can out-pedant you, but I can sure try. I count five separate vowels in xkcd's text:

ə in was, a, of, obs-
ʌ in up, Doug, stuck, etc.
syllabic n in obstruction, onions
syllabic l in tunnel
ʊ in ugh

[Zju]

Now I wanna hear a recording of 'onion' pronounced with a syllabic [n̩]. ([ˈʌn.jn̩]?.. [ˈʌn.n̩]??)

[Travis B.]

I am with Moose-tache here -- STRUT is very much distinct from commA in the dialect here -- and furthermore, the /ə/ phoneme is very often [ɘ], which I perceive as being close to /ɪ/ [ɪ̈], for me.

[malloc]

One of my pet peeves is when people treat the word "schwa" as an alternate name for the STRUT vowel

Meanwhile for the longest time I was genuinely confused by descriptions of English vowels listing the schwa and STRUT vowels as separate sounds. They always seemed identical to me and I thought the distinction represented some esoteric theoretical principle in phonology that eluded me.

[Travis B.]

Meanwhile for the longest time I was genuinely confused by descriptions of English vowels listing the schwa and STRUT vowels as separate sounds. They always seemed identical to me and I thought the distinction represented some esoteric theoretical principle in phonology that eluded me.

The schwa is an unstressed unrounded mid or close-mid central vowel, whereas STRUT is a stressed unrounded open-mid back vowel (think START but a bit closer). This is especially true in Inland North dialects like my own. (Note that it also is distinct from DRESS, which here is a stressed unrounded open-mid central vowel, even though strictly speaking it does not contrast with it as the schwa is always unstressed (except in some monosyllabic grammar words such as the where stressed schwa may be seen) and DRESS is always stressed (and is only found word-finally in the neologism teh), unless you consider the case of a stressed syllabic /r/ as containing a stressed schwa.)

[zompist]

Now I wanna hear a recording of 'onion' pronounced with a syllabic [n̩]. ([ˈʌn.jn̩]?.. [ˈʌn.n̩]??)

[ˈʌn.jn̩]. "Nation" and "onion" end in the same sound. ([n̩], I mean— the [j] is just in onion.)

[bradrn]

Meanwhile for the longest time I was genuinely confused by descriptions of English vowels listing the schwa and STRUT vowels as separate sounds. They always seemed identical to me and I thought the distinction represented some esoteric theoretical principle in phonology that eluded me.

The schwa is an unstressed unrounded mid or close-mid central vowel, whereas STRUT is a stressed unrounded open-mid back vowel (think START but a bit closer). This is especially true in Inland North dialects like my own.

In your dialect, perhaps, but not in others’. My STRUT is certainly not a back vowel, for instance. And it sounds like malloc has the merger.

Now I wanna hear a recording of 'onion' pronounced with a syllabic [n̩]. ([ˈʌn.jn̩]?.. [ˈʌn.n̩]??)

[ˈʌn.jn̩]. "Nation" and "onion" end in the same sound. ([n̩], I mean— the [j] is just in onion.)

[jn̩] is a weird syllable, sonority-wise. Perhaps this is a stupid question, but how do you know it’s not [in]? For me, both of those words end clearly in [ən].

[zompist]

[ˈʌn.jn̩]. "Nation" and "onion" end in the same sound. ([n̩], I mean— the [j] is just in onion.)

[jn̩] is a weird syllable, sonority-wise.

Why is it any odder than "cure"? Or for that matter "strike"?

Perhaps this is a stupid question, but how do you know it’s not [in]? For me, both of those words end clearly in [ən].

Because, as I said, the ending is identical to that in "nation." A lot of people, including lexicographers, seem to be under the impression that syllabic r/l/n includes two segments, [ə] and the consonant, and this just isn't the case in my speech.

[bradrn]

[ˈʌn.jn̩]. "Nation" and "onion" end in the same sound. ([n̩], I mean— the [j] is just in onion.)

[jn̩] is a weird syllable, sonority-wise.

Why is it any odder than "cure"? Or for that matter "strike"?

Er, what’s odd about [kʰjʊɜ̆] and [s̠ʈ͡ʂˤɑɪ̆k]?

(Apart from the pharyngealisation, that is, but then again I never have been quite sure how my /r/ is realised.)

If what you’re getting at is the sonority violation… it isn’t just that, it’s that it yields a V+C syllable where the syllabic item is the consonant rather than the vowel. That’s pretty unusual. (Although then again, I suppose Salishan and Berber languages have even weirder syllabic goings-on.)

Perhaps this is a stupid question, but how do you know it’s not [in]? For me, both of those words end clearly in [ən].

Because, as I said, the ending is identical to that in "nation." A lot of people, including lexicographers, seem to be under the impression that syllabic r/l/n includes two segments, [ə] and the consonant, and this just isn't the case in my speech.

I’ve never been quite sure about this in my own speech. Sometimes it sounds like [ən], and other times it’s more like [n̩]. Most probably they’re in free variation, but it’s hard to know.

[Darren]

(Although then again, I suppose Salishan and Berber languages have even weirder syllabic goings-on.)

Salishan languages do seem to follow the sonority hierarchy, bizarrely enough. Nuxalk for instance only syllabifies fricatives if there's nothing more sonorous in the syllable (according to the aptly-named "Good Enough Nucleus Hypothesis"), and syllabic plosives only occur as syllables on their own. So you don't get any syllables like */st̩ʰ/ or */ws̩/; they'd be syllabified as /s̩.t̩ʰ/ and /us/.

[bradrn]

(Although then again, I suppose Salishan and Berber languages have even weirder syllabic goings-on.)

Salishan languages do seem to follow the sonority hierarchy, bizarrely enough. Nuxalk for instance only syllabifies fricatives if there's nothing more sonorous in the syllable (according to the aptly-named "Good Enough Nucleus Hypothesis"), and syllabic plosives only occur as syllables on their own. So you don't get any syllables like */st̩ʰ/ or */ws̩/; they'd be syllabified as /s̩.t̩ʰ/ and /us/.

Huh, now that’s interesting. It inspired me to check my Tashlhiyt source, which suggests that this is not true in Tashlhiyt: e.g. /iʃkd/ appears to be syllabified as /iʃ̩.kd̩/, with /ʃ/ a nucleus and /i/ an onset.

[Moose-tache]

I'm not sure I can out-pedant you, but I can sure try. I count five separate vowels in xkcd's text:

ə in was, a, of, obs-
ʌ in up, Doug, stuck, etc.
syllabic n in obstruction, onions
syllabic l in tunnel
ʊ in ugh

A fun activity!
I agree with your phonetic analysis (except I have STRUT for "ugh," not FOOT). I consider the schwa in "was" to be a simple reduction caused by lack of stress; the STRUT reasserts itself if you're asked to repeat yourself slowly and clearly. The syllabic consonants are another matter. It would be tidy and elegant to say they are simply the surface realization of unstressed vowel + nasal or lateral. But since there's no way to retrieve the vowel (e.g. I still have syllabic L when I enunciate "tunnel" very carefully), that explanation is a bit... theory for the sake of theory. I hate it when Chompsky-ites say "This kangaroo is actually a bat at a more fundamental level, if you agree to pretend that it is," so I am reluctant to commit the same sin here.

[Richard W]

I'm not sure I can out-pedant you, but I can sure try. I count five separate vowels in xkcd's text:

ə in was, a, of, obs-
ʌ in up, Doug, stuck, etc.
syllabic n in obstruction, onions
syllabic l in tunnel
ʊ in ugh

A fun activity!
I agree with your phonetic analysis (except I have STRUT for "ugh," not FOOT). I consider the schwa in "was" to be a simple reduction caused by lack of stress; the STRUT reasserts itself if you're asked to repeat yourself slowly and clearly. The syllabic consonants are another matter. It would be tidy and elegant to say they are simply the surface realization of unstressed vowel + nasal or lateral. But since there's no way to retrieve the vowel (e.g. I still have syllabic L when I enunciate "tunnel" very carefully), that explanation is a bit... theory for the sake of theory. I hate it when Chompsky-ites say "This kangaroo is actually a bat at a more fundamental level, if you agree to pretend that it is," so I am reluctant to commit the same sin here.

And then if one lacks syllabic consonants, as a fair few people do, then it only takes the merger of /ə/ and /ʌ/ to have but a single vowel. Their distinction seems weird in my speech, mostly one of aperture, but I can just make out the vowel difference between the second syllables of gallop and hiccough. The difference may be reinforced by vowel length, with a rather specialised chrone. At the moment, the minimal pair occurs in gallop and Gallup, but that's probably idiosyncratic and unstable.

[Otto Kretschmer]

How did a French word for work (travail) become an English word for travel?