Pronunciation: -dth

Guest   Tue Aug 12, 2008 11:49 am GMT
OK, I'm having a really hard time pronouncing "-dth" at the end of words like hundredth or thousandth. How do native speakers pronouce this? It seems I always have to silence the d or I end up sounding completely retarded.
Guest   Tue Aug 12, 2008 12:31 pm GMT
It's a /dT/ sound. It is a bit of a tongue twister :).

hundredth /hVndr@dT/
thousandth /TaUz@ndT/
Damian in Edinburgh   Tue Aug 12, 2008 3:32 pm GMT
Easy peasy lemon squeezy - as an example take the word "breadth".......as in length and breadth dimensions - simply say "bread" (as in a loaf) and tack on to the end of it the "th" sound as in the word "thin" but in doing so you automatically modulate the sound of the letter "d" so that you scarcely voice it. If you eliminate the "d" altogether you find yourself saying "breath" - as in exhaled air.
Travis   Tue Aug 12, 2008 5:18 pm GMT
Final historical /dθ/ is somewhat weird in my dialect, as it actually results in [θ] preceded by a *long* vowel (whereas normally /θ/ would always be preceded by a short vowel here). Hence I myself have:

width [ˈwɪːθ]
breadth [ˈb̥ʁɛːθ]
hundredth [ˈhʌ̃ːndʒ̥ɹ̠ɨːθ]

Likewise, final historical /ndθ/ results in [nːθ] preceded by a long vowel here, as in:

thousandth [ˈθɑɔ̯zɨ̃ːnːθ]

However, in quite careful speech, final historical /dθ/ and /ndθ/ may be realized as [θː] and [nθː] or [nd̥θ] preceded by long vowels here, even though such may be at least partly a spelling pronunciation of sorts.
Travis   Tue Aug 12, 2008 5:26 pm GMT
Whoops, "breadth" should be [ˈb̥ʁɜːθ] above.
greg   Tue Aug 12, 2008 9:22 pm GMT
Travis : « hundredth [ˈhʌ̃ːndʒ̥ɹ̠ɨːθ] ».

Ich finde dieses (in Bezug auf Englisch) unwahrscheinliche Nasallaut — nämlich /ʌ̃ː/ (IPA) = /V~:/ (X-Sampa) — gänzlich erquicklich. Das Vokalnasalierungsmuster läßt an Altfranzösisch denken.
Travis   Tue Aug 12, 2008 9:40 pm GMT
It is not that my dialect is really all too atypical from most English dialects in this regard; actually, all English dialects have some level of allophonic nasalization of vowels before nasal consonants. Conversely, vowel nasalization is definitely weaker in my dialect than in, say, French, Portuguese, or Polish, like that in practically all English dialects.

The only thing is that I consistently mark vowel nasalization in my transcriptions because it is very common for /n/ which is not in the onset of a stressed syllable to be elided in my dialect and to just leave the preceding vowel nasalized. Furthermore, there are a few cases where one will have a surface form containing [nː] or [mː] will follow an *unnasalized* vowel, due to reflecting historical /d(V)n/ and /b(V)m/, where V is any vowel; hence merely marking vowel nasalization where /n/ has been elided is not enough, even though such vowel nasalization is generally stronger than that where /n/ has not been lost.
Travis   Tue Aug 12, 2008 9:45 pm GMT
For example, take how I commonly pronounce "lieutenant" (to use an example from a thread on Unilang in which I posted), which is as [ɰ(ˡ)uˈtʰɜ̃ːnɨ̃ʔ] or [ɰ(ˡ)uˈtʰɜ̃ːɨ̯̃ʔ]. To not mark vowel nasalization would leave one with [ɰ(ˡ)uˈtʰɜːnɨʔ] or [ɰ(ˡ)uˈtʰɜːɨ̯ʔ], which is inaccurate because in Real Life one can hear that there had been both intervocalic /n/ and /n/ before /t/ in my dialect. Yet, at the same time, I commonly pronounce "problem" as [ˈpʰʁɑːmː]; if I were to never mark vowel nasalization before realized nasal consonants, one would probably assume that I actually pronounced that as [ˈpʰʁɑ̃ːmː], except that I do not nasalize the vowel in such in Real Life, and the lack of nasalization is actually audible in practice.
greg   Wed Aug 13, 2008 9:46 am GMT
Travis : « For example, take how I commonly pronounce "lieutenant" (to use an example from a thread on Unilang in which I posted), which is as [ɰ(ˡ)uˈtʰɜ̃ːnɨ̃ʔ] or [ɰ(ˡ)uˈtʰɜ̃ːɨ̯̃ʔ]. »

Mensch ! Früher habe ich immer — und ahnungsloserweise... — gedacht, daß En <lieutenant> wahlweise /lɛftɛnənt/ = /lEftEn@nt/ oder /lɪvtɛnənt/ = /lIvtEn@nt/ im Vereinigte Königreich geäußert worden (gegen /luːtɛnənt/ = /lu:tEn@nt/ in Nordamerika). Die Wahrheit ist also ein bißchen komplexer (!), wie es scheint.

Apropos : steht [ɰ] = [M\] für einen Halbvokal ? Und hat sich /l/ in [ɰ] (durch Halbvokalisierung des initialen Mitlauts) bei regressiver Assimilation an /u/ geglichen ?
Travis   Wed Aug 13, 2008 4:00 pm GMT
>>Travis : « For example, take how I commonly pronounce "lieutenant" (to use an example from a thread on Unilang in which I posted), which is as [ɰ(ˡ)uˈtʰɜ̃ːnɨ̃ʔ] or [ɰ(ˡ)uˈtʰɜ̃ːɨ̯̃ʔ]. »

Mensch ! Früher habe ich immer — und ahnungsloserweise... — gedacht, daß En <lieutenant> wahlweise /lɛftɛnənt/ = /lEftEn@nt/ oder /lɪvtɛnənt/ = /lIvtEn@nt/ im Vereinigte Königreich geäußert worden (gegen /luːtɛnənt/ = /lu:tEn@nt/ in Nordamerika). Die Wahrheit ist also ein bißchen komplexer (!), wie es scheint.

Apropos : steht [ɰ] = [M\] für einen Halbvokal ? Und hat sich /l/ in [ɰ] (durch Halbvokalisierung des initialen Mitlauts) bei regressiver Assimilation an /u/ geglichen ?<<

(I am speaking in English rather than German here just because I am far more accustomed to speaking of linguistics topics in English than German, and because I do not have the time to craft a more elaborate reply in German.)

Basically, my pronunciation [ɰ(ˡ)uˈtʰɜ̃ːnɨ̃ʔ] or [ɰ(ˡ)uˈtʰɜ̃ːɨ̯̃ʔ] can be rather directly derived from a historical NAE /luːtɛnənt/. The only matter is that my dialect has been affected by a whole range of sound shifts, some of which are actually very common in modern NAE dialects and some of which are more dialect specific. Of the changes that occurred, in no particular order, the following are the more common ones which show up in very many modern NAE dialects:

1. The loss of historical English phonemic vowel length and its replacement with allophonic vowel length.
2. The merger of all reduced vowels as [ɨ̃] before /n/ except in word-initial position.
3. The backing of all cases of /l/ to [ɫ].
4. The elision of /n/ before /t/ in syllable codas, leaving the preceding vowel nasalized.
5. The realization of coda /t/ not followed by another obstruent within the same word or a vowel in the following word as [ʔ]. Note that in many dialects this is not entirely consistent even in that regard, and many speakers of my own dialect actually lack this change or at least do not express it consistently.

The following less common and more dialect-specific changes have also occurred:

6. Allophonic vowel length has in turn taken on quasiphonemic properties due to further sound changes which obscure the original allophony of such.
7. The backing of historical /ɛ/ to [ɜ]. This is a change associated with the NCVS.
8. The loss of coronal articulation and vocalization in all positions of [ɫ], which has in turn switched to one of [ɰˡ], [ɰ], [ɯ̯], [wˡ], [w], or [ʊ̯] depending on position, stress, and just who is speaking. Such changes actually seem to be quite common in modern NAE dialects in coda positions, but what makes such particularly innovative is that such applies in *all* positions, which is more reminiscent of the fate of historical /ɫ/ in most dialects of Polish. Also note that, no, this is not due to assimilation to the /u/ at all.
9. And last but not least, in the latter pronunciation, the elision of intervocalic /n/ where the following vowel is unstressed, leaving the preceding vowel nasalized and resulting in a nasal diphthong if the preceding vowel is not one of /e/, /i/, /o/, /u/, or /ɒ/. Note that this is not a consistent change at this point, and varies heavily depending on who is speaking and what register they are speaking in.
Travis   Wed Aug 13, 2008 4:58 pm GMT
>>3. The backing of all cases of /l/ to [ɫ].<<

That should be:

>>3. The velarization of all cases of /l/, and not just those in coda positions, as [ɫ].<<
Travis   Wed Aug 13, 2008 5:14 pm GMT
>>Apropos : steht [ɰ] = [M\] für einen Halbvokal ? Und hat sich /l/ in [ɰ] (durch Halbvokalisierung des initialen Mitlauts) bei regressiver Assimilation an /u/ geglichen ?<<

Oh, and yes, [ɰ] in my transcriptions is a semivowel, albeit one closer than [ɯ̯], which is merely a nonsyllabic vowel.
Travis   Wed Aug 13, 2008 5:17 pm GMT
>>Also note that, no, this is not due to assimilation to the /u/ at all.
9. And last but not least, in the latter pronunciation, the elision of intervocalic /n/ where the following vowel is unstressed, leaving the preceding vowel nasalized and resulting in a nasal diphthong if the preceding vowel is not one of /e/, /i/, /o/, /u/, or /ɒ/.<<

The above should instead be:

>>Also note that, no, this is not due to assimilation to the /u/ at all.
9. And last but not least, in the latter pronunciation, the elision of intervocalic /n/ where the following vowel is unstressed, leaving the preceding vowel nasalized and resulting in a nasal diphthong if the preceding vowel is not one of /e/, /i/, /o/, /u/, /ɒ/, /ae̯/, /əe̯/, or /ɑɔ̯/.<<
greg   Wed Aug 13, 2008 9:13 pm GMT
Travis : ich habe deine Erklärung neu auf Französisch formuliert (und für französischsprachige Neulinge auch weiter expliziert).




Travis : « 5. The realization of coda /t/ not followed by another obstruent within the same word or a vowel in the following word as [ʔ]. Note that in many dialects this is not entirely consistent even in that regard, and many speakers of my own dialect actually lack this change or at least do not express it consistently. »

Le phonème /t/ en finale (Auslaut, fonema final) est réalisé [ʔ] en API = [?] en X-Sampa (coup de glotte, Knacklaut, saltillo) par Travis pourvu que :
1] ce /t/ ne précède pas une autre constrictive (Geräuschlaut, constrictiva) et demeure donc effectivement en finale du mot <lieutenant> ;
2] ce /t/ ne soit suivi d'aucune voyelle à l'initiale du mot subséquent, auquel cas la réalisation [ʔ] = [?] serait impossible pour Travis.

Le coup de glotte, noté [ʔ] en API et [?] en X-Sampa, n'est pas phonémique en français mais on peut le retrouver dans le syntagme <les haricots> quand le locuteur distingue bien l'article du substantif → [leʔaʁiko], à différencier de [leaʁiko] quand le locuteur enchaîne la prononciation de l'article et du substantif d'un trait, sans marquer de césure. Noter que les réalisations phonétiques [leʔaʁiko] & [leaʁiko] correspondent toutes deux à un seul schéma phonologique (= phonématique) : /leaʁiko/.
En clair : /leaʁiko/ → [leʔaʁiko] ou [leaʁiko], suivant la libre prononciation adoptée par le locuteur.




Travis : « 4. The elision of /n/ before /t/ in syllable codas, leaving the preceding vowel nasalized. »

Le phonème /n/ — qui précède la consonne finale /t/ (réalisée [ʔ] = [?]) — n'est pas réalisé du tout : il est élidé (= supprimé).

Ainsi la succession phonologique (= phonématique) /...nt/ positionnée en finale du mot <lieutenant> est-elle rendue par un son (= phone) unique : [...ʔ] = [...?]. Et en effet les trois graphèmes <...nant> qui terminent le mot <lieutenant> sont bien prononcés [...nɨ̃ʔ] par Travis.

À la disparition du phonème /n/ s'ajoute, en guise de compensation consécutive, la nasalisation de la voyelle qui précède /n/ (voir ci-dessous).




Travis : « 2. The merger of all reduced vowels as [ɨ̃] before /n/ except in word-initial position. »

La tilde sur le voyelle [ɨ] signifie que celle-ci a été nasalisée, ici sous l'effet de la disparition du phonème suivant /n/ (voir ci-dessus). Autrement dit : [ɨn] → [ɨ̃]. On retrouve un phénomène analogue en français, avec [ɔ] à la place de [ɨ], quand on prononce <gasconne> & <gascon> successivement dans cet ordre : [gaskɔn] (féminin) → [gaskɔ̃] (masculin).

Travis indique le processus [ɨn] → [ɨ̃] s'applique plus généralement à toutes les voyelles inaccentuées à partir du moment où celles-ci ne sont pas placées dans la syllabe initiale du mot.




Travis : « 3. The backing of all cases of /l/ to [ɫ]. »

Les personnes familiarisées avec les langues d'Oc et en particulier le provençal connaissent bien ce phénomène de vélarisation (Velarisierung, posteriorización) du /l/ (très présent en ancien français aussi).
Travis   Wed Aug 13, 2008 9:44 pm GMT
Travis : ich habe deine Erklärung neu auf Französisch formuliert (und für französischsprachige Neulinge auch weiter expliziert).




Travis : « 5. The realization of coda /t/ not followed by another obstruent within the same word or a vowel in the following word as [ʔ]. Note that in many dialects this is not entirely consistent even in that regard, and many speakers of my own dialect actually lack this change or at least do not express it consistently. »

Le phonème /t/ en finale (Auslaut, fonema final) est réalisé [ʔ] en API = [?] en X-Sampa (coup de glotte, Knacklaut, saltillo) par Travis pourvu que :
1] ce /t/ ne précède pas une autre constrictive (Geräuschlaut, constrictiva) et demeure donc effectivement en finale du mot <lieutenant> ;
2] ce /t/ ne soit suivi d'aucune voyelle à l'initiale du mot subséquent, auquel cas la réalisation [ʔ] = [?] serait impossible pour Travis.

Le coup de glotte, noté [ʔ] en API et [?] en X-Sampa, n'est pas phonémique en français mais on peut le retrouver dans le syntagme <les haricots> quand le locuteur distingue bien l'article du substantif → [leʔaʁiko], à différencier de [leaʁiko] quand le locuteur enchaîne la prononciation de l'article et du substantif d'un trait, sans marquer de césure. Noter que les réalisations phonétiques [leʔaʁiko] & [leaʁiko] correspondent toutes deux à un seul schéma phonologique (= phonématique) : /leaʁiko/.
En clair : /leaʁiko/ → [leʔaʁiko] ou [leaʁiko], suivant la libre prononciation adoptée par le locuteur.<<

I would say that the [ʔ] in such words is not epenthic at all, though, as it will even show up in the last word in an utterance. Rather, what it is is that there is a range of realizations of /t/ at the end of a word, which are as either [(ʔ)t] or [ʔ], if not followed by a vowel or /j/ or, in careful speech, even sometimes when followed by a vowel or /j/; [ɾ̥] or just nothing at all (but leaving the preceding vowel short) if followed by a vowel in normal everyday speech; and [(ʔ)tʃ] or [ʔ] if followed by /j/ in everyday speech. Similar rules apply word-internally as well.

>>Travis : « 4. The elision of /n/ before /t/ in syllable codas, leaving the preceding vowel nasalized. »

Le phonème /n/ — qui précède la consonne finale /t/ (réalisée [ʔ] = [?]) — n'est pas réalisé du tout : il est élidé (= supprimé).

Ainsi la succession phonologique (= phonématique) /...nt/ positionnée en finale du mot <lieutenant> est-elle rendue par un son (= phone) unique : [...ʔ] = [...?]. Et en effet les trois graphèmes <...nant> qui terminent le mot <lieutenant> sont bien prononcés [...nɨ̃ʔ] par Travis.

À la disparition du phonème /n/ s'ajoute, en guise de compensation consécutive, la nasalisation de la voyelle qui précède /n/ (voir ci-dessous).<<

The main difference from, say, French in this regard is that underlying /n/ has not been irrevocably lost in such positions. In careful speech, an actual articulated [n] will show up in such positions, and not as a mere spelling pronunciation either. But yes, the actual non-realization of [n] itself does strengthen the nasalization of the preceding vowel in such cases (even though such is always present to some degree).

>>Travis : « 2. The merger of all reduced vowels as [ɨ̃] before /n/ except in word-initial position. »

La tilde sur le voyelle [ɨ] signifie que celle-ci a été nasalisée, ici sous l'effet de la disparition du phonème suivant /n/ (voir ci-dessus). Autrement dit : [ɨn] → [ɨ̃]. On retrouve un phénomène analogue en français, avec [ɔ] à la place de [ɨ], quand on prononce <gasconne> & <gascon> successivement dans cet ordre : [gaskɔn] (féminin) → [gaskɔ̃] (masculin).

Travis indique le processus [ɨn] → [ɨ̃] s'applique plus généralement à toutes les voyelles inaccentuées à partir du moment où celles-ci ne sont pas placées dans la syllabe initiale du mot.<<

Actually, said merger, known as the Lennon-Lenin merger, applies regardless of whether /n/ itself is elided. Even in careful speech, historical /ən/ and /ɪn/ are not distinguished in the dialect here except word-initially, while in careful speech it is normal to pronounce [n] in such words. And likewise, it is common for both to word-finally and in some other cases become just [n̩(ː)], aside from before /t/, indicating that /n/ is still present in the underlying forms in general regardless of register.

>>Travis : « 3. The backing of all cases of /l/ to [ɫ]. »

Les personnes familiarisées avec les langues d'Oc et en particulier le provençal connaissent bien ce phénomène de vélarisation (Velarisierung, posteriorización) du /l/ (très présent en ancien français aussi).<<

This is also a common phenomenon in many West Germanic dialects outside of just English itself; it just happens that RP and Standard German lack such generalized velarization of historical /l/.