Wednesday, 8 June 2011




EFL teachers in Argentina are assiduous in covering every imaginable detail of English (RP-style) pronunciation, and rightly hold me to task if anything in LPD is not entirely clear. Veronica Varela wrote
Would you be so kind to explain to me very briefly why is it that in the dictionary an optional t is [found after] n? Or perhaps you can tell me where I can read this from. For example:

against ə ˈɡentst ə ˈɡeɪntst

I replied, briefly,
This is a feature of some people's pronunciation that we call Plosive Epenthesis. There is a discussion in my blog for 27 Jan 2009.

It applies whenever a nasal is followed by a voiceless fricative within
the same syllable (as I analyse syllables).

This phenomenon is not well described in standard textbooks or English phonetics. This may be because in Daniel Jones’s day it was less characteristic of RP than it is now, or perhaps because it was felt to be inappropriate for EFL. If you look up mince in Jones-Gimson-Ramsaran editions of EPD you will find it transcribed only with -ns. If you fossick around in the Explanations at the beginning of the book, however, you will find, under ‘Other variants of sound-distribution occurring in RP but not as a rule noted in this Dictionary’,
(2) Insertion of a « t » between « n » and « s », or of « d » between « n » and « z », e.g. « fents » for « fens » (fence), « ˈfrendzi » for « ˈfrenzi » (frenzy).
It is not until Roach took over as editor that we find -nts explicitly in entries in the body of the book.

Kenyon & Knott, too, in their Pronouncing Dictionary of American English (1944) mention epenthesis only in their Introduction, under Addition or Omission of Consonants (p. xliv).

In present-day pronunciation, it seems to me, most Americans and plenty of English people (including some RP speakers) pronounce mince as mɪnts rather than mɪns. (There may also be glottal reinforcement or replacement: mɪnʔts, mɪnʔs.)

Presumably this necessarily makes mince a homophone of mints, giving rise to all those jokes about ‘one day my prints will come’.

Cruttenden goes so far as to say (Gimson’s Pron., p. 199)
Few RP speakers regularly maintain the distinction between /ns/ and /nts/ which is widespread in regional speech, e.g. distinguishing the final clusters in mince—mints, tense—tents, assistance—assistants, dance—plants, /nts/ tending to be used in all cases.

In the blog posting I referred to, Mariano Mazzeo (also in Argentina) commented
[When I] have to explain Epenthesis. I usually resort to my own home-made explanation: “When a nasal sound is followed by a voiceless fricative, a voiceless plosive which is homorganic to the nasal may be included between them.” Then I point out a few exceptions, comment briefly on the possibility of epenthesis before a voiced sound and that’s about it.

Mariano was correct to make the generalization that plosive epenthesis happens (for those who use it) when ANY nasal is followed by ANY voiceless fricative, though only in certain syllabification environments. In my 2009 posting I mentioned fen(t)s, ˈkɒn(t)ʃən(t)s, ˈhʌm(p)fri, ˈgæŋ(k)stə (fence, conscience, Humphrey, gangster). So the nasal can be labial or velar as well as alveolar; the fricative can be any of f θ s ʃ.

In the examples just given the nasal+fricative sequence (or cluster) is always word-final or followed in the next syllable by a weak vowel: that is, these are cases where, in my syllabification analysis (which I know some people disagree with), the nasal+fricative cluster is syllable-final.

In LPD I do not allow for epenthesis in words where (in my analysis) there is a syllable boundary between the nasal and the fricative, e.g. ˌɪnˈsaɪd, kənˈsɪdə (inside, consider): I took the view that this absolutely blocks epenthesis.

Interestingly, epenthesis DOES occur in cases such as warmth wɔːm(p)θ and Benson ˈben(t)sn̩, despite the presence of a morpheme boundary between the nasal and the fricative.

To my surprise I find that Cruttenden, in the current seventh edition of Gimson’s Pronunciation of English, says (p. 199, fn 50)
Similar epenthesis may occasionally take place [in other cases] e.g. confusion /kəmp`fjuːʒn/, convert /kəmb`vɜːt/ (=[kəɱb`vɜːt]), anthem /`æntθəm/ (=[`ænt̪θəm]), mansion /`mæntʃən/.

The last two examples are fine, but I don’t believe epenthesis is a regular (or even an occasional) possibility in the first two. It seems to me that the undoubted syllable boundary between the nasal and the fricative in confusion blocks (or at least strongly disfavours) epenthesis there, just as in inside or beanstalk. This applies a fortiori in the case of convert, where the fricative is not even voiceless.


  1. I have most of these epenetheses [sic], but definitely distinguish between gangster and gankster 'an online player who habitually ganks or (over)kills other players'.

    I quite agree that compfusion and combvert are impossible, and in fact pronouncing these words with [ɱ] is just as impossible for me: [m] and [n] contrast strongly before labiodentals.

  2. You also did "some day my prints will come" on 25 August 2010, where you say

    «Speakers who have this epenthetic plosive normally do so whenever a nasal is followed by a voiceless fricative within the same syllable. If the nasal and fricative are in different syllables (e.g. inside, uncertain, consider — you have to buy into my views on syllabification here), there is no epenthesis. So Dan can get a model for plain -ns- by considering his consider

    I said then and I still say that I am continually hearing it in that very word. And words like "inform", "confirm" etc can have an analogous epenthesis.

    Like you I take the view that a syllable boundary between the nasal and the fricative absolutely blocks epenthesis, but only for myself can I say that with any certainty. And for the sort of exceptions I have mentioned, it would rejoice me greatly to have Cruttenden's support, but what you now report him as saying doesn't seem to be it. Why ɱ for 'convert' but not for 'confusion'? And why are the p and b not homorganic with the ɱ in both of them?

    I remember when I first encountered ɱ (I think in my earliest reading of DJ) the example given for it was 'comfort'. I thought it was not worth having for that, but all those years ago it was obvious the example should have been something like 'influence'. And it seems to me that the tendency to epenthesize or glottalize in such words has come on strongly since then.

  3. I'm having two difficulties with Cruttenden's narrowly trascribed [kəɱb`vɜːt].

    1. Why is the labiodental [ɱ] stuck right before a labial [b]? Or is the [b] supposed to be labiodental?

    2. I'm having some difficulty articulating this form without making the [bv] sequence a kind of affricate. This would then be more naturally written as [kəɱ`bvɜːt].

  4. These difficulties are what I was complaining about. As I said, if it's ɱ, why are the p and b not homorganic with it, i.e. p̪, b̪, in both of these examples ('confusion' and 'convert')? If he's transcribing this narrowly, the unadorned p and b do look, and would sound, if you did succeed in articulating them, most peculiar. Your suggestion of using an affricate, with the syllabification [kəɱˈbvɜːt] goes some way towards articulatory plausibility, but if you think of the syllabification of his unproblematic examples 'anthem' [ˈænt̪θəm] (where he does mark the dentalization) and 'mansion' [mæntʃən] (I've put everything in square brackets as the epenthetic bits aren't functional), and ˈkʌmpfət, the t and p are strikingly syllable-final (they can be glottalized or just ʔ), and the fricatives remain strikingly syllable-initial.

    We all seem to agree that the least we can say is that epenthesis is less likely before the stressed syllable, as in John's 'consider' and my 'inform' and 'confirm', and therefore Cruttenden's 'confusion' and 'convert', but I think the same syllable division still obtains in such cases. So the reason Cruttenden's examples didn't seem to be the support they might have been was not the syllable division in them, but the apparent claim that the labialization can go as far as kəmpˈfjuːʒn and kəmbˈvɜːt. If he had specified kəɱp̪fjuːʒn and kəɱb̪ˈvɜːt, it would have been powerful support for my relatively innocuous claim to have been hearing not only ˈɪntfluənts (ˈɪnʔfluənʔs) and ˈɪɱʔfluənts, but also kəntˈsɪdə (kənʔˈsɪdə), ɪntˈfɔːm (ɪnʔˈfɔːm, ɪɱp̪ˈfɔːm, ɪɱʔˈfɔːm) and kəntˈfɜːm (kənʔˈfɜːm, kəɱp̪ˈfɜːm, kəɱʔˈfɜːm).

  5. Very interesting. I was aware of the t-epenthesis but for some reason I never spotted p-epenthesis.

    You're right that many Americans (and also us Canadians, don't forget us!!) pronounce mince and mints as homophones with /t/. It turns out I do also pronounce /p/ in comfort and warmth but plosive epenthesis isn't popping up in convert, influence or even conscience. The latter I pronounce /'kanʃənts/, maybe because of intervening syllable boundary between /n/ and /ʃ/.

  6. p-epenthesis is very common. For example, the two detectives Dupont and Dupond in the Tintin comics are called in English Thompson and Thomson, both < Tom's son but pronounced alike.

  7. JW (Of confusion' and 'convert') «I don’t believe epenthesis is a regular (or even an occasional) possibility»
    JC: «I quite agree that 'compfusion' and 'combvert' are impossible»

    I don't see how you can so sweepingly rubbish Cruttenden's observations. I went into a more detailed analysis of the state of play with respect to this, but all I offered by way of examples was some more observations of my own, and as usual at too great a length for anyone to read it. However spellings like Thompson are also powerful evidence, so I've done a few googles. It's so stunning that the spectacular misspellings involved have any currency at all that they must be the tip of a really significant iceberg of epenthesizers who can spell as well as epenthesize, and the two don't necessarily go together even for RP speakers these days!

    For my example 'influence', LPD is all for ˈɪnᵗf lu‿ənᵗs, which has therefore to be represented as having the n followed by the f within the same syllable (as John analyses syllables), and even for people who do buy into his views on syllabification, that may be a bit more problematic than his parallel treatment of my other initial-stressed example 'comfort' as ˈkʌmᵖf |ət. I have already said I think Cruttenden's syllabification in the transcriptions of 'confusion' and 'convert' is more plausible because the epenthetic plosives are strikingly syllable-final (they can be glottalized or just ʔ), and the fricatives remain strikingly syllable-initial. And all this is particularly striking when the epenthesis comes not after the stressed syllable but before it, which is I think the other thing that makes Cruttenden's examples so implausible to John.

    And this correlates with the Google results:

    +intfluence 1,080 results (And I find MS Word Autocorrect has that in its file of misspellings to correct, and I certainly didn't put it there!)

    +intfluential 100 results (Not common enough for Autocorrect to have in its file. Stunning, what?)

    But I was concentrating on the pre-stress epentheses which I think you disbelieve because they are so relatively uncommon, and also on unvoiced fricatives following them, as I too have difficulty believing Cruttenden's kəmbˈvɜːt. And what I searched was misspellings with intrusive t, as although I find his examples less incredible if I transcribe them as kəɱp̪fjuːʒn and kəɱb̪ˈvɜːt, I think if these sequences do occur, they would be more likely to be interpreted by native speaker-spellers as having t or d after the prefix, especially if they are glottalized. As a belt-and-braces job I did search 'compfuse' and 'combvert', but obviously the results are meaningless. They seem to be some sort of portmanteau words. Of course all these results are a bit confused by nonce formations etc.

    John's examples:
    +intside 39,800 results!
    +contsider 353 results

    My examples:
    +intform 42,200 results!
    +contfirm 525 results

    Cruttenden's examples:
    +contfusion 406 results (+contfuse 170 results)
    +condvert 166 results

  8. In the U.S. state of Utah, where I'm currently studying, strong plosive epenthesis occurs after [l] as well as [n]: words like 'salsa' and 'Chelsea' become [sɑlʔtsə] and [tʃɛlʔtsi]. 'Also' is normally realized as [ɑlʔtsɔʊ] but is sometimes hypercorrected to something like [ɑssɔʊ]. Words like 'infinite' and 'information' normally become [ɪɱʔfɨnɪʔ] and [ɪɱfɚmeɪʃən] but in close speech (speaking at church, for example) they are clearly and deliberately pronounced as [ɪntfɨnɪʔ] and [ɪntfɚmeɪʃən].

  9. As before: when I say something is impossible, I mean it's impossible for me, not that it's out of the range of variation for English in general.

  10. On the note about Argentine teachers, in the movie Evan Almighty, both Steve Carell and the actor playing his son say Argentinian with /g/.

  11. More generally, there's also the "else" variant /'ɛlts/, but that's lexical, right?

  12. John Cowan,
    Yes I remember now you did say that before. My apologies. But _why_ when you say something is impossible without further specification do you mean it's only impossible for you? Do all your fans unfailingly remember about this curiously circumscribed impossibility? It's a bit difficult to remember, seeing that it appears alongside "impossible for me" as in "compfusion and combvert are impossible, and in fact pronouncing these words with [ɱ] is just as impossible for me", as above.

    I do not doubt that my use of "impossible" may on occasion be no less circumscript or even circumspect, but I do try to be specific, and I ring the changes on my specifications of specificity as best I can, in attempt to avoid unduly tedious formulae. No doubt I fail in that attempt too.

    At any rate JW appears not to have the same editorial policy on this as you, and here expresses his reaction to Cruttenden's observation as disbelief, so my querying of his rubbishing them stands.

    No, /elts/ is not lexical. There's also ˈɔːltsəʊ and others I have mentioned on here. I hope I can say that without circumscription.