The "How do You Pronounce X" Thread

[ˈd̪ʲɛ.gɔ kɾuˑl̪]

When I saw Pole's post, I recollected one more question to English speakers.

Some people call me linguist-amateur. I like speaking about languages. I have spoken about them since I was four and I have had one specific way of naming Polish phonemes. It's the main realization of the phoneme plus /i/ or /ɨ/ (some argue this is single phoneme /i/), so /ɕ/ is [ɕiː], /n/ is [n̪ɪ̞ː] etc. But if I think about a foreign phoneme, e.g. /ŋ/, I hear in my mind [ŋɪ̞ː], not something like [ŋäː], what can be heard on Wikipedia. How do you think about a certain phoneme?

[KathTheDragon]

I found one candidate so far: the OED does list /ˈbʊrəʊ/ as a possible British pronunciation of "burro". It's given after /ˈbʌrəʊ/, though, so I don't know if it's at all common.

I do have this.

[finlay]

That reminds me, I guess there is a gap in British English for the sequence /ʊr/? It seems like in theory, the general phonotactics of BrEng should allow it, like /ɛr/ etc, but I guess there are a lot of gaps relating to the distribution of /ʊ/ anyway, it's such an infrequent vowel phoneme (I also can't think of any words with /ʊb/).

Historically English English had /ʊr/ (derived from /uːr/ - earlier /ʊr/ had already merged into what is now /ɜː/ - which was lost in the process), but in very many English English varieties this has been lowered and (of course) nonrhoticized so as to merge with /ɔː/ (shall we call this the sure-shore merger?) IIRC. Contrast with many NAE varieties, where /ʊr/ in many cases was either re-tensed to [u˞] (e.g. in tour) or reduced to [ɜ˞] (e.g. in cure for many), even though some dialects did also lower it to [ɔ˞] (i.e. our sure-shore merger).

RP CURE (non-merged with NORTH/FORCE) is usually written /ʊə(r)/. I was wondering if there were any words with distinct /ʊr/, like how "serious" (with /ɪər/) contrasts with "Sirius" (with /ɪr/) in most British English accents.

I found one candidate so far: the OED does list /ˈbʊrəʊ/ as a possible British pronunciation of "burro". It's given after /ˈbʌrəʊ/, though, so I don't know if it's at all common.

The thing that's happening here is that in American English these vowels get rhotic colouring, but they don't in British English, so those are /si:.ri.əs/ and /sɪ.ri.əs/ (for me anyway - i think scottish english has even less rhotic colouring than english english), but in AmE they're both /sir.riəs/ (simplifying a lot). i think the r is ambisyllabic, actually.

as for don't, want, etc... I naturally pronounce want as /wʌnt/, and i only started using /ɒ/ when i realised this was unusual when i was living in england. the other thing is when i'm angry and want to put stress on "don't", i say /do.ənt/, breaking the syllable just as we were talking about in 'hire' and 'vile'. no idea how common that is.

[Travis B.]

The thing that's happening here is that in American English these vowels get rhotic colouring, but they don't in British English, so those are /si:.ri.əs/ and /sɪ.ri.əs/ (for me anyway - i think scottish english has even less rhotic colouring than english english), but in AmE they're both /sir.riəs/ (simplifying a lot). i think the r is ambisyllabic, actually.

I personally question some analyses of rhotic NAE as having rhotic vowels because that would imply that intervocalic r actually belongs to the preceding syllable not the following one, which goes against the general concept of that syllable onsets should be as large as possible except when there is a very good reason for them not to be (i.e. that such would result in a disallowed onset... but /r/ is a perfectly good onset in English). Rather, I favor analyses where vowels are colored by following r but the r is not part of the vowel itself. (And after all, why should the fact that a vowel is colored by an adjacent consonant mean that the consonant is part of the vowel itself?)

(And hell, in the speech here, /ɑ/ is also colored by following /w h kw gw/ and preceding /r w h/ as well... so does that mean following /w h kw gw/ and preceding /r w h/ are also part of the vowel?)

[Sumelic]

The thing that's happening here is that in American English these vowels get rhotic colouring, but they don't in British English, so those are /si:.ri.əs/ and /sɪ.ri.əs/ (for me anyway - i think scottish english has even less rhotic colouring than english english), but in AmE they're both /sir.riəs/ (simplifying a lot). i think the r is ambisyllabic, actually.

I personally question some analyses of rhotic NAE as having rhotic vowels because that would imply that intervocalic r actually belongs to the preceding syllable not the following one, which goes against the general concept of that syllable onsets should be as large as possible except when there is a very good reason for them not to be (i.e. that such would result in a disallowed onset... but /r/ is a perfectly good onset in English). Rather, I favor analyses where vowels are colored by following r but the r is not part of the vowel itself. (And after all, why should the fact that a vowel is colored by an adjacent consonant mean that the consonant is part of the vowel itself?)

(And hell, in the speech here, /ɑ/ is also colored by following /w h kw gw/ and preceding /r w h/ as well... so does that mean following /w h kw gw/ and preceding /r w h/ are also part of the vowel?)

John Wells has argued against using the maximal-onset principle in the syllabification of English. It's true that /r/ is a valid onset in English, but in general, /ɪ/ /ɛ/ /æ/ /ʊ/ are not valid rhymes, and in many varieties of English these vowels can occur before intervocalic /r/. Syllabifying these as /ɪr/ /ɛr/ /ær/ /ʊr/ saves the principle that lax vowels in stressed syllables are only possible when the syllable is closed.

[Travis B.]

John Wells has argued against using the maximal-onset principle in the syllabification of English. It's true that /r/ is a valid onset in English, but in general, /ɪ/ /ɛ/ /æ/ /ʊ/ are not valid rhymes, and in many varieties of English these vowels can occur before intervocalic /r/. Syllabifying these as /ɪr/ /ɛr/ /ær/ /ʊr/ saves the principle that lax vowels in stressed syllables are only possible when the syllable is closed.

Well wait a second - /ɪ ɛ æ ʊ/ can occur in open syllables, they just cannot occur word-finally (except for /ɪ/ in RP) or before vowels except in interjections, unless you are arguing that English still has non-morphological geminates.

[Sumelic]

John Wells has argued against using the maximal-onset principle in the syllabification of English. It's true that /r/ is a valid onset in English, but in general, /ɪ/ /ɛ/ /æ/ /ʊ/ are not valid rhymes, and in many varieties of English these vowels can occur before intervocalic /r/. Syllabifying these as /ɪr/ /ɛr/ /ær/ /ʊr/ saves the principle that lax vowels in stressed syllables are only possible when the syllable is closed.

Well wait a second - /ɪ ɛ æ ʊ/ can occur in open syllables, they just cannot occur word-finally (except for /ɪ/ in RP) or before vowels except in interjections, unless you are arguing that English still has non-morphological geminates.

I also specified "stressed". What word in RP or GA has one of /ɪ ɛ æ ʊ/ in a stressed open syllable? (Obviously ignoring words where these vowels come before /r/.) Do you syllabify words like "letter" as "le.tter" rather than "lett.er"? It's more-or-less homophonous for me with "let her".

[KathTheDragon]

A syllabification /lɛt.ə/ for me sounds forced and awkward. I'm pretty sure I have /lɛ.tə/.

[finlay]

That's why i used the word "ambisyllabic". Why do we have to choose one or the other? It's both.

[Travis B.]

I also specified "stressed". What word in RP or GA has one of /ɪ ɛ æ ʊ/ in a stressed open syllable? (Obviously ignoring words where these vowels come before /r/.) Do you syllabify words like "letter" as "le.tter" rather than "lett.er"? It's more-or-less homophonous for me with "let her".

Yes. Coda /t/ is frequently glottalized or preglottalized even when followed by a vowel (i.e. across word boundaries) in the English here, so letter cannot be "lett.er", as letter is never pronounced with glottalization or preglottalization. And no, it is not homophonous with let her pronounced with h-dropping, because that is very frequently pronounced with glottalization or preglottalization.

And seriously, you have to have a very good reason to reject the maximal onset rule. And justifying this by saying that certain vowels in English cannot exist in open syllables makes no sense, since how does one justify insisting that these vowels cannot exist in open syllables in the first place? It almost seems like one is trying to formulate a phonological rule for modern English varieties based on the phonology of Late Middle English, where these vowels seemingly could not exist in open syllables due to open syllable lengthening... but wait this was not true in the first place, due to vowel shortening/laxing in words of three or more syllables, so even LME could have short vowels in open syllables!

[Sumelic]

letter cannot be "lett.er", as letter is never pronounced with glottalization or preglottalization.

I'm pretty sure [lɛʔə] is possible for some British English speakers.

And no, it is not homophonous with let her pronounced with h-dropping, because that is very frequently pronounced with glottalization or preglottalization.

Did you miss the "for me" in that sentence? It is a fact that I naturally produce a pronunciation of "let her" that is aurally indistinguishable for me from "letter". Obviously they are not always neutralized; there may ever be speakers for whom they are never neutralized, and I can't even prove that a complete neutralization is possible for me, since I haven't analyzed my speech. But it's a fact that there is definitely a possible perceptual neutralization. Aside from my own perceptions, a minor piece of evidence I can offer for this is the "___er? I hardly know her!" snowclone (e.g. "Letter? I hardly know her!"), which only works if the "er" of the first word is thought to be perceptually separated from the preceding consonant.

And seriously, you have to have a very good reason to reject the maximal onset rule.

Why? What does it help to explain in an analysis of English?

Did you even read Wells? I can understand disagreeing with the idea, but I don't understand why you consider it ridiculous. To me, it seems like the kind of idea that even if it's wrong, has obvious motivations (like, say, the analysis of English long vowels as sequences of short vowel + /j/, /w/, or /h/). It's not like I just came up with this idea on my own; you can see that syllabifying an intervocalic consonant with the preceding vowel when it is short is common enough and old enough to be in the Cambridge English Dictionary's transcriptions (I don't know who is responsible).

And justifying this by saying that certain vowels in English cannot exist in open syllables makes no sense, since how does one justify insisting that these vowels cannot exist in open syllables in the first place?

The same way you justify not starting a syllable with /ks/ in words like "axis". No monosyllabic English word starts with /ks/, so we assume it is not a valid syllable onset (this assumption could be revised if there was sufficient reason to later). No monosyllabic English word ends in a short vowel (setting aside, as you said, interjections in certain accents) so we assume this is not a valid syllable rhyme, except for in situations where we are forced by further considerations to use this type of syllabification (e.g. in unstressed syllables).

[Travis B.]

letter cannot be "lett.er", as letter is never pronounced with glottalization or preglottalization.

I'm pretty sure [lɛʔə] is possible for some British English speakers.

I did state in the English here, i.e. a Inland North dialect, mind you.

And no, it is not homophonous with let her pronounced with h-dropping, because that is very frequently pronounced with glottalization or preglottalization.

Did you miss the "for me" in that sentence? It is a fact that I naturally produce a pronunciation of "let her" that is aurally indistinguishable for me from "letter". Obviously they are not always neutralized; there may ever be speakers for whom they are never neutralized, and I can't even prove that a complete neutralization is possible for me, since I haven't analyzed my speech. But it's a fact that there is definitely a possible perceptual neutralization. Aside from my own perceptions, a minor piece of evidence I can offer for this is the "___er? I hardly know her!" snowclone (e.g. "Letter? I hardly know her!"), which only works if the "er" of the first word is thought to be perceptually separated from the preceding consonant.

You did say for me, but you did seem to imply that the analysis you had put forth applies to English as a whole.

And seriously, you have to have a very good reason to reject the maximal onset rule.

Why? What does it help to explain in an analysis of English?

Um it perfectly explains the distribution of preglottalization in the English here.

Did you even read Wells? I can understand disagreeing with the idea, but I don't understand why you consider it ridiculous. To me, it seems like the kind of idea that even if it's wrong, has obvious motivations (like, say, the analysis of English long vowels as sequences of short vowel + /j/, /w/, or /h/). It's not like I just came up with this idea on my own; you can see that syllabifying an intervocalic consonant with the preceding vowel when it is short is common enough and old enough to be in the Cambridge English Dictionary's transcriptions (I don't know who is responsible).

I consider it ridiculous because, as I said, the distribution of preglottalization in the dialect here perfectly follows the maximal onset rule. And you are arguing from authority here, I should note.

And justifying this by saying that certain vowels in English cannot exist in open syllables makes no sense, since how does one justify insisting that these vowels cannot exist in open syllables in the first place?

The same way you justify not starting a syllable with /ks/ in words like "axis". No monosyllabic English word starts with /ks/, so we assume it is not a valid syllable onset (this assumption could be revised if there was sufficient reason to later). No monosyllabic English word ends in a short vowel (setting aside, as you said, interjections in certain accents) so we assume this is not a valid syllable rhyme, except for in situations where we are forced by further considerations to use this type of syllabification (e.g. in unstressed syllables).

The thing is I do not have any reason to believe that medial open syllables need to behave the same way as final open syllables, and thus just because something cannot be found in a final open syllable does not mean that it cannot be found in a medial open syllable (e.g. diachronically final open syllables have been more prone to reduction than medial open syllables).

[Travis B.]

letter cannot be "lett.er", as letter is never pronounced with glottalization or preglottalization.

I'm pretty sure [lɛʔə] is possible for some British English speakers.

I did state in the English here, i.e. a Inland North dialect, mind you.

And no, it is not homophonous with let her pronounced with h-dropping, because that is very frequently pronounced with glottalization or preglottalization.

Did you miss the "for me" in that sentence? It is a fact that I naturally produce a pronunciation of "let her" that is aurally indistinguishable for me from "letter". Obviously they are not always neutralized; there may ever be speakers for whom they are never neutralized, and I can't even prove that a complete neutralization is possible for me, since I haven't analyzed my speech. But it's a fact that there is definitely a possible perceptual neutralization. Aside from my own perceptions, a minor piece of evidence I can offer for this is the "___er? I hardly know her!" snowclone (e.g. "Letter? I hardly know her!"), which only works if the "er" of the first word is thought to be perceptually separated from the preceding consonant.

You did say for me, but you did seem to imply that the analysis you had put forth applies to English as a whole.

And seriously, you have to have a very good reason to reject the maximal onset rule.

Why? What does it help to explain in an analysis of English?

Um it perfectly explains the distribution of preglottalization in the English here.

Did you even read Wells? I can understand disagreeing with the idea, but I don't understand why you consider it ridiculous. To me, it seems like the kind of idea that even if it's wrong, has obvious motivations (like, say, the analysis of English long vowels as sequences of short vowel + /j/, /w/, or /h/). It's not like I just came up with this idea on my own; you can see that syllabifying an intervocalic consonant with the preceding vowel when it is short is common enough and old enough to be in the Cambridge English Dictionary's transcriptions (I don't know who is responsible).

I consider it ridiculous because, as I said, the distribution of preglottalization in the dialect here perfectly follows the maximal onset rule. And you are arguing from authority here, I should note.

And justifying this by saying that certain vowels in English cannot exist in open syllables makes no sense, since how does one justify insisting that these vowels cannot exist in open syllables in the first place?

The same way you justify not starting a syllable with /ks/ in words like "axis". No monosyllabic English word starts with /ks/, so we assume it is not a valid syllable onset (this assumption could be revised if there was sufficient reason to later). No monosyllabic English word ends in a short vowel (setting aside, as you said, interjections in certain accents) so we assume this is not a valid syllable rhyme, except for in situations where we are forced by further considerations to use this type of syllabification (e.g. in unstressed syllables).

The thing is I do not have any reason to believe that medial open syllables need to behave the same way as final open syllables, and thus just because something cannot be found in a final open syllable does not mean that it cannot be found in a medial open syllable (e.g. diachronically final open syllables have been more prone to reduction than medial open syllables).

[Sumelic]

Thanks for the reply.

letter cannot be "lett.er", as letter is never pronounced with glottalization or preglottalization.

I'm pretty sure [lɛʔə] is possible for some British English speakers.

I did state in the English here, i.e. a Inland North dialect, mind you.

Sorry, I missed your "in the English here" which qualified the "never". This is a good piece of evidence, although I don't think it's absolutely compelling by itself.

You did say for me, but you did seem to imply that the analysis you had put forth applies to English as a whole.

That was a miscommunication. I'm not saying it applies to all accents, only that this piece of evidence for the "lett.er" analysis exists in my accent and apparently in a number of other people's accents.

Did you even read Wells? I can understand disagreeing with the idea, but I don't understand why you consider it ridiculous. To me, it seems like the kind of idea that even if it's wrong, has obvious motivations (like, say, the analysis of English long vowels as sequences of short vowel + /j/, /w/, or /h/). It's not like I just came up with this idea on my own; you can see that syllabifying an intervocalic consonant with the preceding vowel when it is short is common enough and old enough to be in the Cambridge English Dictionary's transcriptions (I don't know who is responsible).

I consider it ridiculous because, as I said, the distribution of preglottalization in the dialect here perfectly follows the maximal onset rule. And you are arguing from authority here, I should note.

"Did you even read Wells? " was not meant as an argument; it was a real question. You seem to be either ignoring or dismissing the arguments he presents (e.g. aspiration, clipping, r-allophony) so I'm unsure if you've seen them or not.

And justifying this by saying that certain vowels in English cannot exist in open syllables makes no sense, since how does one justify insisting that these vowels cannot exist in open syllables in the first place?

The same way you justify not starting a syllable with /ks/ in words like "axis". No monosyllabic English word starts with /ks/, so we assume it is not a valid syllable onset (this assumption could be revised if there was sufficient reason to later). No monosyllabic English word ends in a short vowel (setting aside, as you said, interjections in certain accents) so we assume this is not a valid syllable rhyme, except for in situations where we are forced by further considerations to use this type of syllabification (e.g. in unstressed syllables).

The thing is I do not have any reason to believe that medial open syllables need to behave the same way as final open syllables, and thus just because something cannot be found in a final open syllable does not mean that it cannot be found in a medial open syllable (e.g. diachronically final open syllables have been more prone to reduction than medial open syllables).

Special reduction processes have only applied historically to final unstressed syllables, I think; and everyone agrees that short vowels can occur without a coda consonant in unstressed syllables.

I just thought of another thing I'd like to know about your viewpoint: would you syllabify "singing" as /sɪ.ŋɪŋ/ or /sɪŋ.ɪŋ/?

[Travis B.]

Did you even read Wells? I can understand disagreeing with the idea, but I don't understand why you consider it ridiculous. To me, it seems like the kind of idea that even if it's wrong, has obvious motivations (like, say, the analysis of English long vowels as sequences of short vowel + /j/, /w/, or /h/). It's not like I just came up with this idea on my own; you can see that syllabifying an intervocalic consonant with the preceding vowel when it is short is common enough and old enough to be in the Cambridge English Dictionary's transcriptions (I don't know who is responsible).

I consider it ridiculous because, as I said, the distribution of preglottalization in the dialect here perfectly follows the maximal onset rule. And you are arguing from authority here, I should note.

"Did you even read Wells? " was not meant as an argument; it was a real question. You seem to be either ignoring or dismissing the arguments he presents (e.g. aspiration, clipping, r-allophony) so I'm unsure if you've seen them or not.

I think I have read Wells a very long time ago (I remember his analysis of English long vowels as sequences of short vowels plus /j w h/), but I do not remember it too well.

About aspiration, the explanation I have had IMD is simply that all fortis plosives at the start of a stressed or initial onset (even though the initial /t/ in today, tomorrow, and tonight is frequently an exception) are aspirated.

Note that there are exceptions to the maximal onset rule like where morpheme boundaries result in consonants syllabifying with the preceding syllable and the exception that a few words' pronunciations like conservative pronunciations of Wisconsin have stressed syllables attract following sibilants to themselves rather than having them group with the following syllable (as in my pronunciation of Wisconsin).

I am not familiar with his analyses of clipping or r-allophony though.

And justifying this by saying that certain vowels in English cannot exist in open syllables makes no sense, since how does one justify insisting that these vowels cannot exist in open syllables in the first place?

The same way you justify not starting a syllable with /ks/ in words like "axis". No monosyllabic English word starts with /ks/, so we assume it is not a valid syllable onset (this assumption could be revised if there was sufficient reason to later). No monosyllabic English word ends in a short vowel (setting aside, as you said, interjections in certain accents) so we assume this is not a valid syllable rhyme, except for in situations where we are forced by further considerations to use this type of syllabification (e.g. in unstressed syllables).

The thing is I do not have any reason to believe that medial open syllables need to behave the same way as final open syllables, and thus just because something cannot be found in a final open syllable does not mean that it cannot be found in a medial open syllable (e.g. diachronically final open syllables have been more prone to reduction than medial open syllables).

Special reduction processes have only applied historically to final unstressed syllables, I think; and everyone agrees that short vowels can occur without a coda consonant in unstressed syllables.

Yes, but special processes also happened to final stressed open syllables, namely open syllable lengthening, which did not always happen to word-internal stressed open syllables (due to trisyllabic laxing).

I just thought of another thing I'd like to know about your viewpoint: would you syllabify "singing" as /sɪ.ŋɪŋ/ or /sɪŋ.ɪŋ/?

I would syllabify it as /sɪŋ.ɪŋ/, because /ŋ/ is not a valid onset in any English variety I am aware of.

[Sumelic]

I think I have read Wells a very long time ago (I remember his analysis of English long vowels as sequences of short vowels plus /j w h/), but I do not remember it too well.

About aspiration, the explanation I have had IMD is simply that all fortis plosives at the start of a stressed or initial onset (even though the initial /t/ in today, tomorrow, and tonight is frequently an exception) are aspirated.

Note that there are exceptions to the maximal onset rule like where morpheme boundaries result in consonants syllabifying with the preceding syllable and the exception that a few words' pronunciations like conservative pronunciations of Wisconsin have stressed syllables attract following sibilants to themselves rather than having them group with the following syllable (as in my pronunciation of Wisconsin).

I am not familiar with his analyses of clipping or r-allophony though.

Oh. I linked to the relevant article, but the link was just on Wells's name, so it might not have been clear: "Syllabification and allophony"

I also found a later analysis here by Ricardo Bermúdez-Otero that seeks to explain the same phenomena in a way that respects the maximizing-onsets principle: http://www.bermudez-otero.com/amphichronic.pdf (from page 5 on is most relevant)

special processes also happened to final stressed open syllables, namely open syllable lengthening, which did not always happen to word-internal stressed open syllables (due to trisyllabic laxing).

I'm actually not sure that open syllable lengthening applied word-finally. Did any Old English words end in unreduced, stressed short vowels?

I would syllabify it as /sɪŋ.ɪŋ/, because /ŋ/ is not a valid onset in any English variety I am aware of.

How do you determine what is a valid onset? By looking at what clusters can occur word intially? Or would you say it can be established through introspection?

[Travis B.]

I would syllabify it as /sɪŋ.ɪ/, because /ŋ/ is not a valid onset in any English variety I am aware of.

How do you determine what is a valid onset? By looking at what clusters can occur word intially? Or would you say it can be established through introspection?

Precisely because /ŋ/ cannot occur word-initially, and there is no reason to believe that there are valid medial onsets which are not valid initial onsets.

[ˈd̪ʲɛ.gɔ kɾuˑl̪]

Maybe I'm not aware of something, but shouldn't syllabification show real division of speech? When I see your conversation I think you write mostly about formal distribution of sounds to syllables. To explain, I syllabify Polish words shouting them in my mind and seeing how do I divide given word. Based on this I go from formal division of singing /sɪŋ.ɪŋ/ based on phonotactics of final and initial syllables to real (for me), /sɪ.ŋɪŋ/, based on my [sɪː˦.ŋɪːŋ˥]. So, if I were you I would try it. Will you shout [sɪːŋ˦.ɪːŋ˥]?

[Travis B.]

Maybe I'm not aware of something, but shouldn't syllabification show real division of speech? When I see your conversation I think you write mostly about formal distribution of sounds to syllables. To explain, I syllabify Polish words shouting them in my mind and seeing how do I divide given word. Based on this I go from formal division of singing /sɪŋ.ɪŋ/ based on phonotactics of final and initial syllables to real (for me), /sɪ.ŋɪŋ/, based on my [sɪː˦.ŋɪːŋ˥]. So, if I were you I would try it. Will you shout [sɪːŋ˦.ɪːŋ˥]?

Syllables mean nothing phonetically - they are purely phonological in nature.

[ˈd̪ʲɛ.gɔ kɾuˑl̪]

Maybe I'm not aware of something, but shouldn't syllabification show real division of speech? When I see your conversation I think you write mostly about formal distribution of sounds to syllables. To explain, I syllabify Polish words shouting them in my mind and seeing how do I divide given word. Based on this I go from formal division of singing /sɪŋ.ɪŋ/ based on phonotactics of final and initial syllables to real (for me), /sɪ.ŋɪŋ/, based on my [sɪː˦.ŋɪːŋ˥]. So, if I were you I would try it. Will you shout [sɪːŋ˦.ɪːŋ˥]?

Syllables mean nothing phonetically - they are purely phonological in nature.

Okay, I thought differently.

[Imralu]

I think my /t/ in "yesterday" is dental, maybe even interdental. Phonetically, of course, it's /ˈjestəˌdeɪ/ (using the standard phonemic notation of RP) but it seems to come out of my mouth as something like [ˈjest̪ə̆ˌɾæe̯] or even [ˈjesθ(ə̆)ˌɾæe̯]. Looking in the mirror, I even see my tongue.

I think it might be some weird effect of being stuck between an s and a flapped d with a schwa that's doing its best to disappear. (They seem to try to do that before flaps. When I say "g'day" it's basically just [gɾæe̯]. Could future Aussie end up with two rhotics, contrasting tray/today and grey/g'day?) It also happens to the t in today when I say "as today" or "plus today". Does anyone else have anything like this?

[Znex]

Yep, it happens the same for me as a dental [t̪], although I don't have a flap in "yesterday". I have flaps in those other words with flaps you gave though.

It could just be what's happening to Australian /st/ though: it's becoming less alveolar in nature.

[KathTheDragon]

Maybe I'm not aware of something, but shouldn't syllabification show real division of speech? When I see your conversation I think you write mostly about formal distribution of sounds to syllables. To explain, I syllabify Polish words shouting them in my mind and seeing how do I divide given word. Based on this I go from formal division of singing /sɪŋ.ɪŋ/ based on phonotactics of final and initial syllables to real (for me), /sɪ.ŋɪŋ/, based on my [sɪː˦.ŋɪːŋ˥]. So, if I were you I would try it. Will you shout [sɪːŋ˦.ɪːŋ˥]?

Syllables mean nothing phonetically - they are purely phonological in nature.

I would disagree, but maybe that's just me.

[Pole, the]

Maybe I'm not aware of something, but shouldn't syllabification show real division of speech? When I see your conversation I think you write mostly about formal distribution of sounds to syllables. To explain, I syllabify Polish words shouting them in my mind and seeing how do I divide given word. Based on this I go from formal division of singing /sɪŋ.ɪŋ/ based on phonotactics of final and initial syllables to real (for me), /sɪ.ŋɪŋ/, based on my [sɪː˦.ŋɪːŋ˥]. So, if I were you I would try it. Will you shout [sɪːŋ˦.ɪːŋ˥]?

Syllables mean nothing phonetically - they are purely phonological in nature.

There might be a prosodic difference in mora-timed languages, where coda consonants counts as morae, while onset consonants (and complex onsets) do not. In such a language /sɪ.ŋɪŋ/ would be moraified as /sɪ-ŋɪ-ŋ/ (3 morae), while /sɪŋ.ɪŋ/ would be moraified as /sɪ-ŋ-ɪ-ŋ/ (4 morae).

[Imralu]

Yep, it happens the same for me as a dental [t̪], although I don't have a flap in "yesterday". I have flaps in those other words with flaps you gave though.

It could just be what's happening to Australian /st/ though: it's becoming less alveolar in nature.

Hmm, honestly I wouldn't have been surprised if it was just me because I tend to slur in some funny ways.

I'd never noticed that in terms of a general /st/ thing before. It definitely doesn't happen like this with words like "still" without a following flap, although that "t" is further forward/down than in "till", becoming dental but not interdental. I've just uses the highly scientific method of sticking my finger into my mouth (yesterday was just looking in the mirror) to see exactly where my tongue is. In "yesterday" the tip of my tongue hits the gap between the top and bottom teeth and goes slightly through. It also happens in "stutter" so my theory of the schwa seems to be out, but the following alveolar seems to be important.