Linguistic Miscellany Thread

[bradrn]

Fair enough-- Catford calls [j] an ultrashort vowel. But then, why isn't [in] a big problem? Why isn't that [iən]?

It seems to me that people want to rescue their theories, without even hearing counter-evidence, by inserting vowels that aren't there. And then get snotty about how their analysis cannot possibly be wrong.

To be completely clear, I’m not arguing that you don’t have /jn̩/. I was a little skeptical in the beginning, but that was all.

As Darren notes, many idiolects (including my own) ‘repair’ it by adding a schwa. But then it’s no longer [jn] any more, but [jən].

It shouldn't be necessary, but it seems to be: I am talking about my own idiolect, not Darren's or yours.

And over there I was talking about my own idiolect, not yours.

Hey zompist, sorry for keeping nagging you, but which one of these is close enough to your idiolect? Maybe e.g. the one by ausg? And is it just me, or does ynarakit pronounce onion as [ɐnjɛ]?

To me ynarakit sounds more like [ɐnjɛ̃], with nasalisation. Perhaps the recording got cut off early.

But also… it sounds to me like most of these recordings have a very clear [jən], with a distinct schwa. There are a few US recordings which sound more like a true [jn], though.

[zompist]

Hey zompist, sorry for keeping nagging you, but which one of these is close enough to your idiolect? Maybe e.g. the one by ausg? And is it just me, or does ynarakit pronounce onion as [ɐnjɛ]?

I think ausg is closest to what I'm talking about. bananaman, Matt3799, and Neptunium all come close.

The Americans except for Matt sound like [jɪn].
TopQuark seems to have [iən]. Weirdly long for a j.
Lizj and eliasg sound like [jɛn], while lurkfish sounds to me like [jæn].

With some of these, it can be hard to decide, but the above is my best guess.

[bradrn]

Hey zompist, sorry for keeping nagging you, but which one of these is close enough to your idiolect? Maybe e.g. the one by ausg? And is it just me, or does ynarakit pronounce onion as [ɐnjɛ]?

I think ausg is closest to what I'm talking about.

I had a listen to this, and couldn’t quite decide what’s happening in it, so I made a spectrogram. Here’s what I eventually came up with (sorry for the large image):



So, there’s a few interesting things here:

• There’s a bit of what sounds like [ɪ] to me, then a shorter transition region, then [n]. The time of [ɪ] plus the transition is roughly the same amount of time as the [ä].
• I labelled the transition as [ə] because that’s what it sounded like to me, but the formants make it clear that it’s really more like [ɪ͡ə]. (Presumably it’s nasalised too.)
• Judging from the topmost plot, that ‘transition’ region seems to be very slightly louder than the [ɪ], though by such a small amount it may be insignificant.
• Most interestingly, the transition from /n/ to /j/ is almost instantaneous, but the following transition from /j/ to /n/ clearly takes up a significant proportion of the vowel (perhaps one-third).

Calling this [jən] doesn’t sit quite right with me, because the supposed [j] is longer than the [ə]… but at the same time, [jn̩] feels inadequate to describe what’s happening here. Plausibly, the best transcription may be [änɪə̆n].

Whatever the transcription, I think the best explanation is that this phonologically is /jn̩/, but that’s so highly marked sonority-wise that the language ‘tries really hard’ to get rid of it (so to speak). That manifests itself as the phonology spitting out an extra vowel which is different enough from /j/ to act as a separate syllable nucleus, but ultra-short enough to make it clear that it doesn’t correspond to any underlying phoneme. And all this must be done in the phonology, not as a result of articulatory factors, because the [nj] shows no signs of any such transitional vowel.

(Hmm, I hope that all made sense… it makes sense to me, at least!)

EDIT: fixed typo

[Richard W]

Whatever the transcription, I think the best explanation is that this phonologically is /jn̩/, but that’s so highly marked sonority-wise that the language ‘tries really hard’ to get rid of it (so to speak).

Why does the phonology even have /jn̩/? Wouldn't /jən/ do just as well?

[Zju]

Right, that makes sense as to how /jn̩/ would be realised. I was curious to see a spectrogram of it.

[bradrn]

Whatever the transcription, I think the best explanation is that this phonologically is /jn̩/, but that’s so highly marked sonority-wise that the language ‘tries really hard’ to get rid of it (so to speak).

Why does the phonology even have /jn̩/? Wouldn't /jən/ do just as well?

In a few ways, the [ə] just doesn’t look ‘nucleus-ey’ enough to me. I’ve already mentioned that it’s half the length of the neighbouring [ɪ], which is enough to make me suspicious. Also, if it was underlyingly /ə/, I’d expect a much clearer distinction from the preceding [ɪ] and following [n]; instead, formant-wise, it just looks like what you get if you gradually transition from the one to the other.

[Richard W]

Whatever the transcription, I think the best explanation is that this phonologically is /jn̩/, but that’s so highly marked sonority-wise that the language ‘tries really hard’ to get rid of it (so to speak).

Why does the phonology even have /jn̩/? Wouldn't /jən/ do just as well?

In a few ways, the [ə] just doesn’t look ‘nucleus-ey’ enough to me. I’ve already mentioned that it’s half the length of the neighbouring [ɪ], which is enough to make me suspicious. Also, if it was underlyingly /ə/, I’d expect a much clearer distinction from the preceding [ɪ] and following [n]; instead, formant-wise, it just looks like what you get if you gradually transition from the one to the other.

Even for a schwa? Or are your schwas not prone to deletion?

[bradrn]

Why does the phonology even have /jn̩/? Wouldn't /jən/ do just as well?

In a few ways, the [ə] just doesn’t look ‘nucleus-ey’ enough to me. I’ve already mentioned that it’s half the length of the neighbouring [ɪ], which is enough to make me suspicious. Also, if it was underlyingly /ə/, I’d expect a much clearer distinction from the preceding [ɪ] and following [n]; instead, formant-wise, it just looks like what you get if you gradually transition from the one to the other.

Even for a schwa? Or are your schwas not prone to deletion?

Hmm… not sure, actually. My gut feeling is that a true /jən/ would be much more obvious as such on a spectrogram, but it’s hard to tell.

In terms of testing this theory, at the moment I can only think of one word which might begin with /jən/, that being the contracted form of you know. But when I ran a quick spectrogram I honestly couldn’t tell what was happening there. Adding to the uncertainty is that the above spectrogram wasn’t even my voice, so who knows how that person would say a true /jən/.

(And, you know, on reflection, I’m wondering if I don’t actually realise ‘you know’ as /jnɞ͡ʉ/… when I say it really quickly in isolation it seems to have no schwa at all, and certainly less of a schwa than ‘onion’ does. This would support my earlier theory, in that this word already has a vocalic nucleus, so there’s less need to insert an extra high-sonority phone.)

[zompist]

I had a listen to this, and couldn’t quite decide what’s happening in it, so I made a spectrogram. Here’s what I eventually came up with (sorry for the large image):

Very interesting! I wonder if you could do the same for one of the others, especially the ones I said didn't come close to mine? (E.g. the first UK speaker, or the Americans besides Matt.)

[Travis B.]

For final /ən/ after /t/ I consistently have [n̩], after /d/ I have [n̩] (or the two merge together as [nː]), but otherwise I have allophony between [n̩ ɘ̃n ɘ̃]. [ɘ̃] seems to be inconsistently favored before vowels and semivowels, otherwise it seems to be free variation. Hence for onion /ʌnjən/ I have [ˈʌ̃ːnjn̩(ː)]~[ˈʌ̃ːnjɘ̃(ː)n]~[ˈʌ̃ːnjɘ̃(ː)].

[bradrn]

I had a listen to this, and couldn’t quite decide what’s happening in it, so I made a spectrogram. Here’s what I eventually came up with (sorry for the large image):

Very interesting! I wonder if you could do the same for one of the others, especially the ones I said didn't come close to mine? (E.g. the first UK speaker, or the Americans besides Matt.)

I just had a look at xyzzyva, who sounds to me like he has a distinct /ə/. Here’s the spectogram:



So, it looks like this person has a true diphthong [ɪ͡ə], going smoothly from [ɪ] to [ə]. Still, that yields a more distinct [ə] than the last person had.

To investigate this further in my own speech, I recorded myself saying the following words:

1. sign /sɑjn/
2. ire /ɑjɐ/
3. iron, ion /ˈɑjən/
4. ionisation /ˌɑjənɑjˈzæjʃən/
5. onion /ˈɐnjən/
6. bottom /ˈbɒtəm/
7. button /ˈbɐtən/
8. cabin /ˈkæbən/
9. bnei /bnei/ (Hebrew, ‘sons of’)

For now I won’t bother sharing all these spectograms, but here’s the conclusions I have so far:

• Words 5–8 show a clear [ə] vowel in the nucleus, before the so-called ‘syllabic consonant’. In particular, onion very obviously has a true [jən].
• By contrast, the Hebrew word bnei shows an abrupt transition from [b] to [n], so clearly I am capable of pronouncing such clusters without an intermediate vowel.
• Words 2–3 are triphthongs. The quality of the central vowel is hard to determine, but I believe ire is [ɑɪ̆ɐ] and iron/ion are [ɑĕən].
• On the other hand, ionisation has smoothed out the triphthong, yielding [ɑ͡ɪn-]. Similarly, there is no schwa in sign [sɑ͡ɪn].

So, in sum: it appears that I very consistently vocalise my syllabic consonants. (Which makes sense, given that I have /l/-vocalisation too.) I am perfectly capable of pronouncing these consonant clusters without schwa, but when the phonological conditions are right, schwa gets inserted anyway.

[zompist]

So, it looks like this person has a true diphthong [ɪ͡ə], going smoothly from [ɪ] to [ə]. Still, that yields a more distinct [ə] than the last person had.

Thanks for looking! My impression is that he has [jɪn] while ausg has [jn]. It's pretty clear that he has a longer and more gradual vowel and a much shorter [n].

I may be sensitive to the ɪ/ə contrast because it's significant in my speech-- also Travis's, if I recall correctly.

And yeah, when you look closely diphthongs are a little journey through the formant space-- Labov talks about this.

[Darren]

I just had a look at xyzzyva, who sounds to me like he has a distinct /ə/. Here’s the spectogram:

How do you make these lovely spectograms?

[bradrn]

I just had a look at xyzzyva, who sounds to me like he has a distinct /ə/. Here’s the spectogram:

How do you make these lovely spectograms?

Using Praat.

[Travis B.]

I may be sensitive to the ɪ/ə contrast because it's significant in my speech-- also Travis's, if I recall correctly.

The only place where I have such a contrast is before /r/ (i.e. I contrast /ɪr/ and /ɜr/~/ər/ as [ɪʁˤ] and [ʁ̩ˤ]). I perceive the two as being distinct, but they do not form a true phonemic contrast in any other cases because I cannot form any unstressed syllables other than those exceptions in which the two are contrastive, i.e. I have allophony between [ə] and [ɘ] even though I perceive the latter as akin to stressed /ɪ/ not before /r/, which for me is a [ɪ̞̈]. Hence I perceive Finland [ˈfɪ̞̈̃ːnɰɘ̃ːnt] as /ˈfɪnlɪnd/ even though it is better indicated as /ˈfɪnlənd/ because I have no actual contrast between /ɪ/ and /ə/ in unstressed syllables when not before /r/. Because I am able to perceive this difference made me unaware of that I have the weak vowel merger for the longest time even though I actually can form no such contrast.

[bradrn]

Finland [ˈfɪ̞̈̃ːnɰɘ̃ːnt]

Wait, your /l/ is [ɰ] even in onsets? I didn’t realise there was any dialect of English where /l/-vocalisation had gone that far.

(For me, it’s [ɫ] in the onset and [ɰ~w] in the coda.)

[Travis B.]

Finland [ˈfɪ̞̈̃ːnɰɘ̃ːnt]

Wait, your /l/ is [ɰ] even in onsets? I didn’t realise there was any dialect of English where /l/-vocalisation had gone that far.

(For me, it’s [ɫ] in the onset and [ɰ~w] in the coda.)

When I am not speaking carefully and it is not geminate, yes. When I am speaking carefully and it is in an onset, or when it is geminate, it tends to be [ʟ̞].

[jal]

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

According to Etymonline, Old French already had an "arduous journey" sense, but it also mention in the "travel" entry "The semantic development may have been via the notion of "go on a difficult journey," but it also may reflect the difficulty of any journey in the Middle Ages".

If you want to invent a schwa before the jump, why not invent it before, too? It's the exact same tongue movements in reverse.

Theoretically, yes. But though I can easily glide my tongue forward across the pallet, I can't for the life of me reverse it - it jumps or stutters automatically. As for "onion", I have a smooth glide forward of my tongue against the pallet from alveolar "n" to palettal "j" with something of a [ɲ] in the mix, then a short lowering of the tongue (but far from a full schwa), but that may be just it opening for the oral "j", then an upward movement again, though I think it's mostly the tip of my tongue closing for the "n". I think the second "n" is more alveopalettal than the first, which it definitely more alveolar, but also articulated more with the tip of my tongue than the first, that's more bunched (probably because of the "o" in front).


JAL

[bradrn]

If you want to invent a schwa before the jump, why not invent it before, too? It's the exact same tongue movements in reverse.

Theoretically, yes. But though I can easily glide my tongue forward across the pallet, I can't for the life of me reverse it - it jumps or stutters automatically.

I think you’ll find that if you say [nĭn], your tongue glides both forwards and back with no difficulty. But as soon as you think ‘onion’ while doing it, your brain notices that that final /n̩/ is ‘meant’ to be syllabic, and automatically inserts something schwa-like to make that phonetically clear.

[jal]

I think you’ll find that if you say [nĭn], your tongue glides both forwards and back with no difficulty.

It doesn't glide at all actually, it just attaches, detaches, and attaches again, at the alveolar ridge.

But as soon as you think ‘onion’ while doing it, your brain notices that that final /n̩/ is ‘meant’ to be syllabic, and automatically inserts something schwa-like to make that phonetically clear.

"Onion" has something of raised schwa, or at least something that feels like a schwas, because of the palatalization of the first /n/. That's what causes the glide. I think the "schwa" is just the opening then closing of the channel for pronouncing the /j/.


JAL