/d/ for intervocalic /t/ in American dictionaries.

[Space60]

Some American English dictionaries I've seen will represent intervocalic /t/ as /d/. They give the pronunciation of "letter" as /lEd@r/. What do you think about this?

[Zaarin]

Flapping. Virtually all Americans have [ɾ] for /t d/ between vowels, including syllabic resonants (except /n̩/, where at least some of us have [ʔ] -- so butter [bʌɾɹ̩] but button [bʌʔn̩]).

[Kuchigakatai]

Some American English dictionaries I've seen will represent intervocalic /t/ as /d/. They give the pronunciation of "letter" as /lEd@r/. What do you think about this?

I see it as nothing but a welcome and righteous development towards the establishment of the true pronunciation that the LORD intended. Death to the difference between "liter" and "leader", "wetting" and "wedding", "matter" and "madder".

[Estav]

The original post is kind of vague. I guess you're asking about the suitability of /d/ as a transcription for lenited original /t/? For me, [ɾ] in the context of English sounds the same as [d], so it's a fitting transcription in that sense. Even though linguistic material that is easily available online seems to prefer to describe the lenition in terms of "tapping" or "flapping", that's not a salient quality to my ears, while the voicing is.

Some accents retain a distinction in the length and/or quality of the preceding vowel in at least some cases, so in that regard, using a symbol that is distinct from /d/, such as /t̬/, might be preferable. But at the same time, the vowel distinctions apparently aren't always in accordance with the etymological identity of the consonant: for example, I've heard that some speakers pronounce some words that etymologically should have /aɪd/, like "cider" or "spider", with the [əɪ] vowel that would only be expected before /t/. And in any case, dictionaries also give the spelling of a word, which makes it obvious whether the original consonant was /t/ or /d/ regardless of whether the transcriptions consistently mark that information.

[Linguoboy]

I've heard that some speakers pronounce some words that etymologically should have /aɪd/, like "cider" or "spider", with the [əɪ] vowel that would only be expected before /t/.

This is why the versions of the flapping rule for NAE I'm familiar with usual include reference to morpheme boundaries. (Having said that, though, I have centralisation in spider but not cider, so English is basically just a mess.)

[Travis B.]

Death to those who would write /t/ as "/d/" in American English dictionaries!

At least in the dialect here, /t/ and /d/ are distinguished in unstressed intervocalic positions even when elided. This is because /t/ is preceded by a short vowel while /d/ is preceded by a long vowel, and when elision takes place, cases that had /t/ receive long vowels while cases that had /d/ receive overlong vowels.

Also, in the dialect here, /aɪ/ before intervocalic /t/ is always [əe̯] whereas /aɪ/ before intervocalic /d/ is [ae̯] except in words like spider, cider, Ida, idle and Idaho, where then it is [əe̯].

[Space60]

https://www.google.com/search?source=hp ... I_S0ZqJzGY

The Google dictionary represents flap t as /d/.

They give /ˈwôdər,ˈwädər/ as the pronunciation of "water".

[anteallach]

Also, in the dialect here, /aɪ/ before intervocalic /t/ is always [əe̯] whereas /aɪ/ before intervocalic /d/ is [ae̯] except in words like spider, cider, Ida, idle and Idaho, where then it is [əe̯].

"Words like" those in what sense? Is there some rule here (before intervocalic /d/ not before a morpheme boundary?) or what?

[Travis B.]

Also, in the dialect here, /aɪ/ before intervocalic /t/ is always [əe̯] whereas /aɪ/ before intervocalic /d/ is [ae̯] except in words like spider, cider, Ida, idle and Idaho, where then it is [əe̯].

"Words like" those in what sense? Is there some rule here (before intervocalic /d/ not before a morpheme boundary?) or what?

The rule seems to be that /aɪ/ is [əe̯] before fortis obstruents and before lenis plosives followed directly by /ə/ (when not realized as [ɘ], e.g. Ida, Idaho), /ər/ (e.g. tiger, spider), /əl/ (e.g. idle), /ɔ/ (e.g. Midol), or /oʊ/ (e.g. Fido) unsplit by a morpheme boundary. Note that the vowel following matters, because Biden (as in the former vice-president) and Heidi have [ae̯] (and that isn't an effect of borrowing, since Schneider has [əe̯]). There are exceptions, though; liger, Tiber, Midol, and Fido may optionally have [ae̯] for me at least.

[gestaltist]

For me, [ɾ] in the context of English sounds the same as [d]

As exemplified by [ɾɔɹ] for "door"?

[missals]

The rule seems to be that /aɪ/ is [əe̯] before fortis obstruents and before lenis plosives followed directly by /ə/ (when not realized as [ɘ], e.g. Ida, Idaho), /ər/ (e.g. tiger, spider), /əl/ (e.g. idle), /ɔ/ (e.g. Midol), or /oʊ/ (e.g. Fido) unsplit by a morpheme boundary. Note that the vowel following matters, because Biden (as in the former vice-president) and Heidi have [ae̯] (and that isn't an effect of borrowing, since Schneider has [əe̯]). There are exceptions, though; liger, Tiber, Midol, and Fido may optionally have [ae̯] for me at least.

What you've said sounds mostly correct, but the further away you get from the original pre-voiceless-consonant environment, the more individual variation you get. I, for example, always have /əj/ on Tiber, and I and a lot of other speakers have it in certain positions before morpheme boundaries - high school and bicycle, most typically.

Also in common with many speakers, I have it before all coda /r/ - but here, it is sensitive to morpheme boundaries, resulting in a plethora of minimal pairs, such as raised lyre vs. unraised liar, hire vs. higher, shire vs. shyer - partly a consequence of the dephonemicization of schwa between a dipthong and a following liquid, i.e. a vile-vial merger, resulting in words like lyre being disyllabic like liar. But I also have minimal pairs like raised pliers (the tool) vs. unraised pliers (or plyers) (ones who ply), and raised pacifier (for babies) vs. unraised pacifier (one who pacifies), showing that the ultimate factor conditioning the presence of /əj/ before /r/ is morpheme boundary, not syllable break, given that pliers, for example, has always had two syllables.

[Travis B.]

I uniformly have [əe̯] before historical /r/ and [ae̯] before historical /ər/, regardless of morpheme boundaries (e.g. Meyer has [ae̯] despite being monomorphemic); note though that I may also distinguish such words by pronouncing words with historical /r/ without breaking into disyllables.

[Nortaneous]

The rule seems to be that /aɪ/ is [əe̯] before fortis obstruents and before lenis plosives followed directly by /ə/ (when not realized as [ɘ], e.g. Ida, Idaho), /ər/ (e.g. tiger, spider), /əl/ (e.g. idle), /ɔ/ (e.g. Midol), or /oʊ/ (e.g. Fido) unsplit by a morpheme boundary.

Further evidence that orthographic final -o is actually /əw/. I wonder if anyone has raising there but not before /ɔ/. (For lenis plosives, I only have raising before /-dər/, so spider, cider, but not tiger, idle, Idaho, fire, etc. Title and tidal contrast.)

What about before -y? (e.g. tidy)

[Travis B.]

What about before -y? (e.g. tidy)

I have [ae̯] in tidy.

[Travis B.]

I should note that aside from the cases with /ər/ such as tiger and Tiber, it seems all the lenis plosive cases involve /d/. I should also note that there are clear discrepancies such as fiber, which always has [ae̯].

[Zaarin]

Am I weird for having raising in the other direction? For me /ai/ is [ɐɪ̯] before a voiceless consonant, [aɪ̯] otherwise (regardless of what's in the following syllable).

[missals]

I should note that aside from the cases with /ər/ such as tiger and Tiber, it seems all the lenis plosive cases involve /d/. I should also note that there are clear discrepancies such as fiber, which always has [ae̯].

Oh, wow, then I am much different, then. I have raising on cyber (and all derivatives), Khyber, fiber, libel, Niger, Nigerien, Nigeria, Nigerian, Nigel, Nigella, and Rigel.

[Travis B.]

I should note that aside from the cases with /ər/ such as tiger and Tiber, it seems all the lenis plosive cases involve /d/. I should also note that there are clear discrepancies such as fiber, which always has [ae̯].

Oh, wow, then I am much different, then. I have raising on cyber (and all derivatives), Khyber, fiber, libel, Niger, Nigerien, Nigeria, Nigerian, Nigel, Nigella, and Rigel.

Of those I have it in cyber- and optionally in Khyber and Niger- (with the apparent notable exception of Niger itself).