Variation in surface forms showing underlying form oddities

[Travis B.]

I was thinking about something today, specifically that I have winter [ˈwɘ̃ːʁˤ] and twenty [ˈtʲsʲʰwʌ̃ːj] but splinter [ˈsʲpɰɘ̃ɾ̃ʁ̩ˤ(ː)]~[ˈsʲpɰɘ̃ntʁ̩ˤ(ː)] and dentist [ˈd̥ɜ̃ntɘsʲ(tʲ)] most of the time. The thing about this, though, is that the frequency of /nt/ reduction to [ɾ̃] or outright elision ought to be equal across the board, assuming these have underlying /nt/ (the adjacent vowels do not seem relevant here).

So how do we handle cases where they are differences in surface forms' distribution which cannot be explained by conventional analyses? In my case one could just say screw the conventional analysis, bring on the phonemic vowel length and nasalization—but the problem is there are far too many coincidences in surface forms and in what wordforms are consciously accessible to people which imply that there is not phonemic vowel length or nasalization—for starters, vowel length and nasalization do not seem to be consciously accessible to speakers of the dialect here.

So if there is not phonemic vowel length or nasalization, what then? Simply ditching phonemicity does not seem to be a viable alternative, because from all signs people do have phonemic awareness, or why would only certain wordforms be permitted in coinages, and why would loanwords get remodeled to one's native dialect's phonemic system even in cases where the phones in question could be produced natively in other environments?

[Richard W]

Could literacy be complicating matters here? Dentist is more bookish than the other words, and to me splinter feels vaguely bookish, even though I think it is learnt via speech, at least as a word for something that gets under one's skin.

[bradrn]

Could literacy be complicating matters here? Dentist is more bookish than the other words, and to me splinter feels vaguely bookish, even though I think it is learnt via speech, at least as a word for something that gets under one's skin.

I find this somewhat hard to believe. For me both ‘dentist’ and ‘splinter’ are both fairly common words. The former is more common than the latter, and of course both ‘winter’ and ‘twenty’ are far more basic items of vocabulary, but I don’t see them as ‘bookish’ by any means.

[Travis B.]

The thing is how do we capture these sorts of things in terms of phonemic analysis? We can say that one form is more "learnèd" than another, but ultimately how is this factored in, especially since conventional phonemic analysis, an analysis with both phonemic vowel quantity and phonemic vowel nasalization, and a purely phonetic analysis all seem untenable? I am almost thinking that an analysis that mixes levels is necessary, where people are consciously aware of phonemes and apply this to coinages, loanwords, and learned words, but people are simultaneously not consciously aware of the actual realization of their native words, especially core vocabulary, which may very well violate their own perceived phonemic system.

[Richard W]

Could literacy be complicating matters here? Dentist is more bookish than the other words, and to me splinter feels vaguely bookish, even though I think it is learnt via speech, at least as a word for something that gets under one's skin.

I find this somewhat hard to believe. For me both ‘dentist’ and ‘splinter’ are both fairly common words. The former is more common than the latter, and of course both ‘winter’ and ‘twenty’ are far more basic items of vocabulary, but I don’t see them as ‘bookish’ by any means.

Dentist is further from being a 'word of one syllable' than winter and twenty.

[Travis B.]

Could literacy be complicating matters here? Dentist is more bookish than the other words, and to me splinter feels vaguely bookish, even though I think it is learnt via speech, at least as a word for something that gets under one's skin.

I find this somewhat hard to believe. For me both ‘dentist’ and ‘splinter’ are both fairly common words. The former is more common than the latter, and of course both ‘winter’ and ‘twenty’ are far more basic items of vocabulary, but I don’t see them as ‘bookish’ by any means.

Dentist is further from being a 'word of one syllable' than winter and twenty.

The thing is that the intervocalic elision phenomenon does not seem limited to short or core words, e.g. I normally elide the /t/ in Saturday and I often elide the /n/ in Oconomowoc - but some words seem to resist it for unclear reasons.

[Travis B.]

A good example of how it's not just core vocabulary that is subject to this is Internet [ˈɘ̃ːʁ̃ˤˌnɜʔ]~[ˈɘ̃ɾ̃ʁ̩̃ˤːˌnɜʔ].

[Rounin Ryuuji]

The Internet is pretty core to our lives now, however.

[Travis B.]

Interestingly enough I tend to have less reduction in Interwebs [ˈɘ̃ɾ̃ʁ̩ˤːwɜːps]~[ˈɘ̃ntʁ̩ˤːwɜːps].

[Zju]

Sorry, but how many people pronounce winter as [ˈwɘ̃ːʁˤ]? Which dialect is that and is it documented somewhere? The closest pronunciation I could find on forvo was more like [winɹ] or something. Ditto for twenty.

[Travis B.]

Sorry, but how many people pronounce winter as [ˈwɘ̃ːʁˤ]? Which dialect is that and is it documented somewhere? The closest pronunciation I could find on forvo was more like [winɹ] or something. Ditto for twenty.

I live in southeastern Wisconsin, and no, I have not conducted a study or referenced any on this with documented evidence, but listened to people around me speaking unprompted. I would not trust something like Forvo for this anyways, because people always tend to speak a bit more carefully when they are prompted to pronounce particular words, especially if they are being recorded, myself included. (I speak much more carefully on the phone or Internet voice chat than I do in everyday speech.)

About [winɹ], the typical standard NAE pronunciation for winter IRL is [ˈwɪ̃ɾ̃ɻ̩(ː)] as far as I can tell. Regressive assimilation of vowels to the nasality of following consonants and pronouncing unstressed intervocalic /nt/ as [ɾ̃] are both normal NAE features. Where my dialect differs here is that it elides the /nt/ readily, which is really not that strange for NAE, since it is common for NAE varieties to elide unstressed intervocalic flaps in uncareful speech, that it lowers and centralizes /ɪ/ (which is a typically NCVS feature), that it has a bunched /r/ (what I describe as [ʁˤ], because it is the best way I can find to render it in this environment, which really is not strange for NAE), and then after the elision it merges the two syllables into a single syllable because hiatus is forbidden after /ɪ/ (again, restrictions on where hiatus is allowed are typical of English).