/t/ versus /d/ and /tS/ versus /dZ/ alternation in NAE

[Travis B.]

One thing I had noticed a while back at least in the English in southeastern Wisconsin was that there certain words that seem to have had /d/ rather than an expected /t/ or have had /dʒ/ rather than an expected /tʃ/. These were particularly noticeable to me because said English varieties normally preserve the distinction between historical /t/ and /d/ and between historical /tʃ/ and /dʒ/ very well for NAE varieties. There are very few environments in them where the members of either pair ever become ambiguous or are liable to become the other member of the pair. Even when the consonants themselves become ambiguous, vowel quantity and, to a lesser extent, quality still encodes whether the next historical obstruent had been fortis or lenis, regardless of its synchronic realization.

As a result I am wondering about the diachronics of the members of these pairs' crossing over in NAE. Some of the pairs are somewhat obvious in why they crossed over, but others are seemingly more complex or obscure in what just happened in their crossing over.

When not stated obvious below, I presume that all the changes marked below apply to all less conservative NAE varieties. I am only using my own pronunciations because they are the ones I am most familiar with personally, and also because my own dialect typically preserves these kinds of distinctions better than most NAE varieties do.

Here are some cases that I have been thinking of at the moment, but if anyone else thinks of any other notable cases, please mention them:

/t/ > /d/:

kindergarten; I have [ˈkʰɪ̃ːndʁ̩ˤːˌɡɑːʁˤ(ɾ)ɨ̃(ː)n]~[ˈkʰɪ̃ːnːʁ̩ˤːˌɡɑːʁˤ(ɾ)ɨ̃(ː)n], indicating /d/, instead of the expected [ˈkʰɪ̃ːndʁ̩ˤːˌɡɑʁˤʔn̩(ː)]~[ˈkʰɪ̃ːnːʁ̩ˤːˌɡɑʁˤʔn̩(ː)] that would indicate /t/. I presume this is a case of analogy with English garden overpowering orthography, and in my own dialect there likely is also dialect borrowing at work, where it likely had passed through dialects that more easily permit the changing of intervocalic /rt/ to /rd/.

/tʃ/ > /dʒ/, likely linked to historical /t/ > /d/:

congratulate and related words like congratulations; I have [kʰŋːˈɡɰˤɛ̞ːd̥ʒ̊ɯ̞ːʟ̞eʔ]~[kʰŋːˈɡɰˤɛ̞ːd̥ʒ̊ɯ̞ːɰeʔ], indicating /dʒ/, rather than [kʰŋːˈɡɰˤɛ̞tʃɯ̞ːʟ̞eʔ]~[kʰŋːˈɡɰˤɛ̞tʃɯ̞ːɰeʔ] that would indicate /tʃ/. Note congrats [kʰŋːˈɡɰˤɛ̞ʔts], where either said crossing never occurred or where said crossing over was somehow reversed either under the influence of orthography or dialect borrowing.

/dʒ/ > /tʃ/:

sandwich; General American and many other NAE varieties have the change of /dʒ/ > /tʃ/ here. My own dialect noticeably lacks this change except when dialect borrowing or orthographic influence is at work, preserving /dʒ/ here as reflected by [ˈsɛ̞̃ːnwɨːtʃ]~[ˈsɛ̞̃ːwɨːtʃ] and also by sandwiches [ˈsɛ̞̃ːnwɨːd̥ʒ̊ɨːs]~[ˈsɛ̞̃ːwɨːd̥ʒ̊ɨːs]. Note that [ˈsɛ̞̃ːndwɨʔtʃ]~[ˈsɛ̞̃ːnːwɨʔtʃɨːs] and sandwiches [ˈsɛ̞̃ːndwɨtʃɨːs]~[ˈsɛ̞̃ːnːwɨtʃɨːs], reflecting /tʃ/, are also attested in formal speech my dialect, and likely reflect dialect borrowing and/or orthographic influence.

[The Unseen]

I refuse to believe you have a pharyngealized uvular rhotic.

[Travis B.]

I refuse to believe you have a pharyngealized uvular rhotic.

Would you believe it if I called it a bunched /r/ without coronal articulation (except after another coronal)? And in that case, would you know a better way to transcribe it phonetically (without the cop-out of just writing [r])?

Anyways, this is completely irrelevant to what I had been speaking of in my original post.

[makvas]

Honestly you over-analyze phones to the extent it's hard to read your post.

Otherwise I don't see what's so strange about this "phenomenon"...

[Travis B.]

Honestly you over-analyze phones to the extent it's hard to read your post.

No, I don't. Rather, I just do actual phonetic transcription, as opposed to doing what most people do, which is a mish-mash of marking phonemic and phonetic features. Conversely, when I mark historical phonemes, I mark historical phonemes, and I do not confuse synchronic realizations with them.

Too bad most people, even here, seem to be utterly incapable of actually reading proper phonetic transcription. It is quite irritating when some of them rather throw around meaningless accusations of "overanalyzing" in response to actual phonetic transcription when they find themselves unable to actually understand them.

Note that in this post my marking of vowel quantity is very much topical, as it preserves the historical fortisness/lenisness of following consonants very well. It actually matters for some of these words, such as sandwich; from a naive, superficial transcription would assume that both my dialect and General American have /tʃ/ due to the surface realization in both being [tʃ], while it is obvious from vowel quantity that my dialect actually has /dʒ/ for sandwich despite any surface realizations of /tʃ/ or /dʒ/.

Otherwise I don't see what's so strange about this "phenomenon"...

I was just wondering about the diachronics of certain words where, rather than any kind of mere neutralization having taken case amongst the members of each pair, one member of a pair unambiguously became the other member of the pair. I never claim to have "discovered" this or whatnot, but rather these particular words' diachronics were simply rather interesting to me, and I could only guess at the answers for them offhand myself.

[tezcatlip0ca]

I refuse to believe you have a pharyngealized uvular rhotic.

Would you believe it if I called it a bunched /r/ without coronal articulation (except after another coronal)? And in that case, would you know a better way to transcribe it phonetically (without the cop-out of just writing [r])?

[M\r\`_-)_G_-_w]?

[Nortaneous]

I refuse to believe you have a pharyngealized uvular rhotic.

Would you believe it if I called it a bunched /r/ without coronal articulation (except after another coronal)? And in that case, would you know a better way to transcribe it phonetically (without the cop-out of just writing [r])?

[M\r\`_-)_G_-_w]?

Shouldn't that be [M\_-r\`)_G_-_w]? Bunched r is definitely further back than velar.

[Soap]

I suspect kindergarten was loaned from German and read as if it were "kindergarden" (hence the very common misspelling) and that it never had /t/. At least not in America. Maybe they say it with /t/ in Britain or Aus or someplace, but the word has only been in use in English for about 150 years, according to EtymOnline, so whatever they say in Britain doesnt really have any relevance to America.

Congradulate ... no idea. Only that I can attest that people say it that way in New England too.

[Travis B.]

I suspect kindergarten was loaned from German and read as if it were "kindergarden" (hence the very common misspelling) and that it never had /t/. At least not in America. Maybe they say it with /t/ in Britain or Aus or someplace, but the word has only been in use in English for about 150 years, according to EtymOnline, so whatever they say in Britain doesnt really have any relevance to America.

That is my suspicion as well, but I really have no real documentation supporting it besides my own speculation, hence my wondering in my original post.

[Travis B.]

I refuse to believe you have a pharyngealized uvular rhotic.

Would you believe it if I called it a bunched /r/ without coronal articulation (except after another coronal)? And in that case, would you know a better way to transcribe it phonetically (without the cop-out of just writing [r])?

[M\r\`_-)_G_-_w]?

Shouldn't that be [M\_-r\`)_G_-_w]? Bunched r is definitely further back than velar.

What I have varies considerably POA-wise; prevocalically what I have is velar, also postalveolar if preceded by a coronal, and elsewhere (intervocalically and postvocalicallly) it is uvular. It also certainly is pharyngealized. It is not rounded at all in the normal sense of the term; it is only ever "rounded" at all word-initially or, less commonly, in other prevocalic environments, and then it has a bit of a "compressed lips" thing (think Japanese /u/) going on rather than normal rounding, and even that is highly inconsistent and very optional.

[makvas]

Bunched r is definitely further back than velar.

My "bunched r" is somewhere between palatal and velar. In fact it might be something like [1`_^]

[TaylorS]

Bunched r is definitely further back than velar.

My "bunched r" is somewhere between palatal and velar. In fact it might be something like [1`_^]

My R varies between /r\_?\/ and /R_?\/ depending on the surrounding phones.

[zompist]

I think I share all three pronunciations: kindergarden, congrajulate, sandwidjes. That would be suburban Chicago, not far from your home dialect anyway. (Either sandwich or sandwij sounds OK to me.)

Though "garden" may have an influence on the first, there's not really any need for an extra explanation-- post-stress t normally gets realized as [d] here. 'Carted' and 'carded' are homophones for me.

[eodrakken]

Though "garden" may have an influence on the first, there's not really any need for an extra explanation-- post-stress t normally gets realized as [d] here. 'Carted' and 'carded' are homophones for me.

Do you also have [d] in "carton", though? Carted and carded have [d] for me too, but not carton or kindergarten; they both have [?] because of the syllabic nasal.

[Nortaneous]

IMD:

carted: /karɾᵻd/ [ˈkʰɑɰ̠͡ɻɽɪ̽d̚]
carded: /karɾᵻd/ [ˈkʰɑɰ̠͡ɻɽɪ̽d̚]
carton: /kartən/ [ˈkʰɑɰ̠ʔn̠̩]~[kʰɑɰ̠͡ɻʔɳ]
kindergarten: /kɪndərgardən/ [ˈkʰɪndɰ̟̩ˌɡɑɰ̠͡ɻɖɳ̩] (where ɰ̟̩ (M\_+=) is probably the thing Zoris mentioned, except with no retroflexion, or at least a laminal, not apical, postalveolar secondary articulation)
congratulate: /kəngrædʒəlet/ [kŋ̩ˈɡɰ̠͡ɻˠ̠ʷæːdʒɫ̩ˌeːʔ͡t̚]

[Travis B.]

I think I share all three pronunciations: kindergarden, congrajulate, sandwidjes. That would be suburban Chicago, not far from your home dialect anyway. (Either sandwich or sandwij sounds OK to me.)

Though "garden" may have an influence on the first, there's not really any need for an extra explanation-- post-stress t normally gets realized as [d] here. 'Carted' and 'carded' are homophones for me.

It is interesting that, as close to you as Milwaukee, carted and carded are definitely not homophones, except maybe for some middle-aged and older middle and upper-class people who speak General American; to me at least they are definitely distinct, being:

carted: [ˈkʰʌʁˤ(ɾ̥)ɨːd̥]
carded: [ˈkʰɑːʁˤ(ɾ)ɨːd̥]

Diachronically, in this case probably what happened is that intervocalic /rt/ and /rd/ never neutralized; this later influenced the outcome of Canadian Raising after it spread into the Milwaukee area and the replacement of historical phonemic vowel quantity with frozen allophonic vowel quantity.

It is still interesting that, at least our two datapoints, there is an isogloss with regard to the neutralization of flapped /t/ and /d/ somewhere between Milwaukee and Chicago. However, to say anything definitive about this one would have to look at traditional Chicago dialect, as you are from the suburbs yourself.

Changes: Increased transcription size.

[makvas]

is probably the thing Zoris mentioned, except with no retroflexion, or at least a laminal, not apical, postalveolar secondary

There's no retroflexion in my /r/, I was using xsampa "`" to mean rhoticized, not retroflex.

[zompist]

Though "garden" may have an influence on the first, there's not really any need for an extra explanation-- post-stress t normally gets realized as [d] here. 'Carted' and 'carded' are homophones for me.

Do you also have [d] in "carton", though? Carted and carded have [d] for me too, but not carton or kindergarten; they both have [?] because of the syllabic nasal.

Same for me on "carton", good catch.

[Soap]

Vermont says kindergarten with a voiceless t? Is that common in Vermont or just your idiolect?

[Mecislau]

No, I don't. Rather, I just do actual phonetic transcription, as opposed to doing what most people do, which is a mish-mash of marking phonemic and phonetic features. Conversely, when I mark historical phonemes, I mark historical phonemes, and I do not confuse synchronic realizations with them.

Most professional linguists I've read do the same thing. When discussing a particular aspect of phonetics, they only go into detail when it's absolutely necessarily, because superfluous information hurts more than it helps. Hell, you even see professionals use mixed orthographic/phonetic transcription all the time (eg, "sandwi[dʒ]es"). Why? Because it's emphasizing just the relevant information.

Too bad most people, even here, seem to be utterly incapable of actually reading proper phonetic transcription. It is quite irritating when some of them rather throw around meaningless accusations of "overanalyzing" in response to actual phonetic transcription when they find themselves unable to actually understand them.

Good Communication 101: Write to your audience. You're trying to pose a question to the ZBB, but you yourself realize that few people know how to interpret extremely detailed phonetic transcriptions. Yet, you do it anyways, and then blame the people here for being "utterly incapable of actually reading proper phonetic transcription". I'm sure it's not your intent, but it makes you look like an ass.

Note that in this post my marking of vowel quantity is very much topical, as it preserves the historical fortisness/lenisness of following consonants very well. It actually matters for some of these words, such as sandwich; from a naive, superficial transcription would assume that both my dialect and General American have /tʃ/ due to the surface realization in both being [tʃ], while it is obvious from vowel quantity that my dialect actually has /dʒ/ for sandwich despite any surface realizations of /tʃ/ or /dʒ/.

None of which you stated in your original post. You can't expect us to be psychic and know the consequences of every single feature of your transcription. It would be a lot more effective overall if you used broader transcription scheme for most of your post, and then included this very detail separately, as a proof that this particular [tʃ] is actually /dʒ/. That way you clear all of the superfluous information out of the areas where it's not relevant and distracting.

And even if the vowel quality is topical in certain circumstances (and, as you explained in the above quote, it does appear to be), that doesn't justify the very narrow transcription everywhere else where it isn't relevant.


Being able to use narrow transcription accurately is a good skill, but so is knowing when it's simply not appropriate.




EDIT: I should add that in a thread like this, a narrower transcription is perhaps more appropriate than it would be elsewhere. It was mostly your comment that came off as blaming everyone else for the miscommunication that triggered this post.

[Thomas Winwood]

Maybe they say it with /t/ in Britain

"Kindergarten" is an exclusively American concept - the UK has nurseries or preschools. If I used the word at all it'd be a spelling pronunciation.

[eodrakken]

Vermont says kindergarten with a voiceless t? Is that common in Vermont or just your idiolect?

Me? I'm from San Francisco originally. I really can't remember if others there have [?] in kindergarten, or if it's just me (maybe the spelling has subconsciously influenced my pronunciation, but I never thought about it before this). I asked the nearest native Vermonter, and she has [d], FWIW.

[Travis B.]

No, I don't. Rather, I just do actual phonetic transcription, as opposed to doing what most people do, which is a mish-mash of marking phonemic and phonetic features. Conversely, when I mark historical phonemes, I mark historical phonemes, and I do not confuse synchronic realizations with them.

Most professional linguists I've read do the same thing. When discussing a particular aspect of phonetics, they only go into detail when it's absolutely necessarily, because superfluous information hurts more than it helps. Hell, you even see professionals use mixed orthographic/phonetic transcription all the time (eg, "sandwi[dʒ]es"). Why? Because it's emphasizing just the relevant information.

In this post I could probably have omitted everything but the topical consonant's realization, the length of the preceding vowel, and any preglottalization or lack thereof, yes; I did instead to go and bold the relevant parts of each transcription rather than merely cutting them out.

To me, however, there is a difference between reducing the detail of a phonetic transcription, while keeping it phonetic, and mixing phonetic and phonemic transcriptions. I myself normally do not include all phonetic detail, and typically omit much of what I am aware of but which would make transcriptions only more complex, such as long vowels IMD being typically laxer and having different phonation from short vowels, or where IPA is defined ambiguously but where further precision is not needed for what I am describing; ironically enough I have actually had to add rather than remove precision in response to questions about my transcriptions, such as the details of my realization of /r/ and /l/. On the other hand, I could probably omit much of the description of alternate realizations in my posts, though I prefer to include it as it often better describes what shows up in actual recordings while more "normal" pronunciations are actually primarily explicitly careful ones.

By actually mixing phonemic and phonetic transcription I am rather referring to the rather maddening tendency for people to use what is essentially phonemic transcription and then add selected phonetic features, often in an arbitrary or seemingly rule-driven away, rather than going from actual phones and then removing irrelevant detail. Transcriptions of this sort are irritating because they seem to reflect a lack of knowledge of the actual phones being transcribed and may at times be confusing, as they often indicate things that would be quite notable were they how someone actually pronounced them but which may actually be just an artifact of inconsistently marking a mixture of phonemes and phones.

Too bad most people, even here, seem to be utterly incapable of actually reading proper phonetic transcription. It is quite irritating when some of them rather throw around meaningless accusations of "overanalyzing" in response to actual phonetic transcription when they find themselves unable to actually understand them.

Good Communication 101: Write to your audience. You're trying to pose a question to the ZBB, but you yourself realize that few people know how to interpret extremely detailed phonetic transcriptions. Yet, you do it anyways, and then blame the people here for being "utterly incapable of actually reading proper phonetic transcription". I'm sure it's not your intent, but it makes you look like an ass.

It seems to me like "extremely detailed phonetic transcriptions" in this context means simply phonetic transcriptions, that is, transcriptions fundamentally based in phones, in the first place. It seems like many individuals here have this idea of phonetic transcription of English that is based primarily in, as I mentioned before, a mechanical transformation, often seemingly memorized, of phonemes into what look like phones, yet have little ability to actually reason about phones and phonology outside such a mechanistic paradigm.

This is, of course, very bothersome when I am trying to ask for comments about English phonology. I have in many cases gotten complaints about how I am "showing off how special my dialect is" even though what I describe is actually phonological features that, to myself, seem to be quite widespread in North American English varieties as actually spoken, and I am only using my own dialect as an example as I can speak of it definitively without having to find suitable informants and extract the desired phonetic detail from them without their attempting to "correct" themselves*. This indicates that the reader obviously has no understanding of both what I am even asking for comments about and how NAE varieties are actually spoken, and prefers to cover for their own ignorance with baseless mockery and cynicism.

* This is actually very hard to do in an repeatable, reliable fashion while recording the subject in practice. And even when this is done successfully, it usually requires painstaking phonetic analysis after the fact, especially since one normally does not have full (or any) access to articulatory information this way.

All this makes it so that I often do not even bother to ask for comments about these sorts of matters even when I very well want further information on them and know that there are individuals here who would likely be capable of providing useful further information on the matter from their past posts, simply as I do not always feel like having to respond to the aforementioned baseless mockery and cynicism. Often I ask for comments on these sorts of matters specifically because I know that my own knowledge of them is limited, and I want further clarification as to the distribution of a feature or phenomenon that I suspect is widespread but which I have seen little real documentation of. Very often, these comments confirm that something is as widespread as I suspect it is, but in other cases have made it clear that a particular feature is actually synchronically limited in distribution or have raised as many questions as answered them, especially due to having to take others' transcriptions with a degree of skepticism, particularly with regard to things such as voicing or vowel quantity.

Note that in this post my marking of vowel quantity is very much topical, as it preserves the historical fortisness/lenisness of following consonants very well. It actually matters for some of these words, such as sandwich; from a naive, superficial transcription would assume that both my dialect and General American have /tʃ/ due to the surface realization in both being [tʃ], while it is obvious from vowel quantity that my dialect actually has /dʒ/ for sandwich despite any surface realizations of /tʃ/ or /dʒ/.

None of which you stated in your original post. You can't expect us to be psychic and know the consequences of every single feature of your transcription. It would be a lot more effective overall if you used broader transcription scheme for most of your post, and then included this very detail separately, as a proof that this particular [tʃ] is actually /dʒ/. That way you clear all of the superfluous information out of the areas where it's not relevant and distracting.

And even if the vowel quality is topical in certain circumstances (and, as you explained in the above quote, it does appear to be), that doesn't justify the very narrow transcription everywhere else where it isn't relevant.

Actually, I devoted an entire good-sized paragraph in my original post to the significance of vowel quantity in this particular post and just why I was marking it here, despite the post being nominally about historical phoneme distribution and not about phonetic and phonological detail.

[Dewrad]

tl;dr - I enjoy the feeling of being superior, thank you very much.

[Travis B.]

tl;dr - I enjoy the feeling of being superior, thank you very much.

Or, in other words, you enjoy being snarky and cynical for their own sake.