[Please note that in the end this is not essentially a question about English!]

Intervocalic /t/ in Gen Am English may be realised as a voiced alveolar tap, [ɾ].

In words like entertain or ninety where the /t/ occurs after an /n/, it is also often voiced in General American. Given that the nasal stop, /n/, usually involves a full blockage of the air leaving the vocal tract, and that a canonical /t/ is homorganic with a canonical /n/, it is reasonable to suppose that the realisation of the voiced /t/ might actually be an alveolar plosive instead of a tap. So my first question is: is it? Of course, I'm aware that the /nt/ sequence in such words may actually be reduced to a single nasal tap, but I'm only interested in the situation where the /n/ and /t/ are both realised.

My second question, however, is whether it's actually possible for a tap to follow directly on from a (nasal or oral) stop made with the same articulators. So, for example, is it even physically possible for [ɾ] to follow an 'unreleased' alveolar [t]? And is it possible for [ɾ] to directly follow an /n/ if the tongue does not lose contact with the alveolar ridge between the two segments?

  • It's hard for me to understand your first question, since you write phonemes, yet seem to be talking about sounds. Is your "voiced /t/" actually [d]? For your second question, there is a different definitional problem, which is whether you can call an alveolar consonant a flap when it is not immediately preceded by a non-alveolar sound. I don't think you can.
    – Greg Lee
    Commented Aug 23, 2015 at 6:26
  • @GregLee Well, the most common voiced allophone of /t/ is [ɾ], but a voiced allophone of /t/ could of course have many different forms. I could have written "When an underlying /t/ follows an underlying /n/, the resulting allophone of /t/ may be voiced, perhaps as the result of a phonological rule". Of course I can't write "A canonical [t]" to contrast a canonical /t/ with, for example a non-canonical realisation of /t/ for example [ɾ], or [ʔ]. So my first question can be rephrased like this ... Commented Aug 23, 2015 at 10:38
  • @GregLee ... "When we get /nt/ followed by a weak vowel, we might expect the allophone of /n/ to be [n] and therefore, because this is homorganic with a canonical /t/, [t], we might expect the voiced version of /t/ to be a plosive with a full alveolar stricture as opposed to [ɾ]. Is this this happens in those cases where /n/ is realised by [n]?" Commented Aug 23, 2015 at 10:47
  • In my speech, /nt/ before unstressed vowel is a nasal flap, [t], or [nt]. Never [n] or [nd]. Does that answer the question? I don't know why you expect there to be a "voiced version of /t/" other than the voiced nasal flap. Following David Stampe's analysis, the voiced flap allophone arises by this route: intervocalic syllable offset [t] becomes sonorant (i.e. a flap), then that becomes voiced by assimilation. [t] does not voice, but rather the flap derived from /t/ voices.
    – Greg Lee
    Commented Aug 23, 2015 at 13:48
  • Any info on why this Q was downvoted? I can't improve it without some!! Commented Oct 14, 2018 at 22:52

2 Answers 2


When flapping applies to /t/ in "entertain", "ninety", /n/ actually deletes and the preceding vowel becomes nasalized, so you have [ɛ̃ɾ̃ɹ̩ˈtʰɛjn, ˈnãj̃ɾ̃ɪj]. The explanation for the correlation between these processes is that a syllable-final nasal in contact with a following homorganic consonant can delete, but deletion causes nasalization of the preceding vowel. When the nasal consonant deletes, /t/ then becomes intervocalic. There may be dialect variation in whether or not deletion+nasalization takes places before just voiceless stops, or before voiced stops as well. The test would be whether "banter" and "bander" can be neutralized-- they don't in my dialect, although both /t/ and /d/ flap in underlyingly-intervocalic context (writer, rider). Restricting ourselves to just the variants where n is realized as a nasal, then t will be realized as a stop (voiceless, not voiced) and not as a flap, since when preceded by a nasal, /t/ is not intervocalic.

Given the nature of a lingual tap (flap), the tip of the tongue has to be away from the palate and then strike the palate, so between any lingual consonant and a lingual flap, there has to be a moment when the first consonant's constriction is released and the tongue is retracted, and then it can strike the palate.

  • For me "banner" and "banter" are neutralized -- both have a medial nasal flap, but remain distinct from "bander", whose "nd" can simplify to an "n" which is not subject to flapping.
    – Greg Lee
    Commented Aug 23, 2015 at 6:12
  • Actually, the vowel preceding the nasal would be nasalized whether the nasal is deleted or not. It's just more obvious when the nasal's gone. English routinely nasalizes vowels preceding nasal consonants; the velic flap is an independent articulator and can antitipate the approaching obstruent by opening. This becomes obvious to anyone working with Indonesian languages, which routinely nasalize vowels following nasal consonants. Learning to do this precisely requires suppressing the English norm, which is very hard.
    – jlawler
    Commented Aug 23, 2015 at 13:35
  • 2
    @jlawler, you might want to look at Cohn's phonetic research on supposed nasalization in English (which compares English, French and Sundanese). She shows that vowels are not actually nasalized before in English, though there is some right-edge nasal leakage. Unfortunately she did not capture examples like reduced banter.
    – user6726
    Commented Aug 23, 2015 at 15:55
  • 1
    Well, of course not. The purpose in English is to anticipate the velic opening required for the nasal obstruent, and full nasalization is not necessary for that -- just an opening of the velic at some point before the obstruent is articulated -- what you call "right-edge nasal leakage". I'm not sure of the Sundanese purpose, but if it's anything like Indonesian or Minangkabau, it will have significant left-edge nasal leakage in vowels after the obstruent.
    – jlawler
    Commented Aug 23, 2015 at 16:43

Note: I don't know too much about phonetics, so this answer is mainly about phonology.

For me, there is a difference between the pronunciation of <nt> in the two words "ninety" and "seventy" and the pronunciation of <nt> in words with regular spellings, like entertain, center, winter, twenty. In words with regular spelling-pronunciation correspondences (which I suppose should be analyzed as having /nt/ phonemically), I can have /ɾ̃/ or /nt/ (or just nasalization followed by /t/, probably), but not [nd].

In "ninety" and "seventy", I feel like I have [nd], which I would analyze as /nd/, not as /nt/. (I say "feel like" because I would imagine that I actually sometimes produce these words without any phonetic plosive [d], but these pronunciations are not salient or obvious to me the way that the possible pronunciation of "twenty" without a [t] is.) I would guess that the use of /nd/ is somehow based on analogy with the pronunciation of other "-ty" words, although I don't know exactly in what way. Perhaps, based on the /t/~/d/ alternation in -ed-suffixed words, words like fifty and sixty were misperceived as having an underlying /d/ that was only voiceless because of the preceding voiceless fricative.

For evidence that some other people use these special pronunciations of seventy and ninety, see this reddit thread (which also mentions the word "carpenter" as a possible candidate for /nd/, which I guess I would agree with).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.