Weak t

« previous post | next post »

Yesterday I claimed that "if you examine a random stretch of English speech closely, within 10 seconds (of speech time, not your analysis time) you'll notice an instance of an interesting general pattern that has never been systematically studied in the linguistic literature." And I promised to test that claim on a clip from the S-Town podcast, which I happened to have at hand because I was adding it to a portfolio of English intonational examples. (To make it a fair test — though of course this is an illustration of my claim, not a proof — I'm looking for an "interesting general pattern" that's not in the area of intonation.)

But first, let's take a look at another sample from the intonation portfolio, President Donald Trump's weekly address for 3/10/2017. Mr. Trump starts this way:

Listen to the word "scientists":

And just the last syllable of that word:

The dictionary will give you the pronunciation of that syllable as [tɪsts], but the president's performance is really [tɪsss]. Some would describe this as "t/d deletion", but that's a misleading analysis, in my opinion. One clue: the [s] segment is about 300 milliseconds long, which is about the same length as the expected [sts] cluster.

The word "scientists" spans the time period from 8.70 to 9.937 seconds, so I win the over/under on this sample.

Now let's check the S-Town clip. Here's that first clause:

oh it's wonderful at homecoming to go to the turnip green supper

And the first two words:

You'll notice that "it's" is pronounced [ɪss].

Now "at home-":

And just "at" with the initial consonant of "home":

There the final /t/ of "at" has assimilated to — or perhaps merged with — the initial /h/ of "homecoming".

One more: the sequence "-coming to go":

There the initial /t/ of "to" has become a voiced flap, as in might in a word like "twenty".

These are all examples of the weakening (or "lenition") of non-pre-stress onset consonants. Different aspects of this phenomenon in American English are known as "flapping of t" and "t/d deletion", but it should be seen as part of a larger pattern, which in fact is not limited to coronal consonants like /t/. For example, here's Donald Trump's pronunciation of "entrepreneurs":

And now just the two medial syllables:

At the start of the third syllable, what the dictionary tells you is a /p/ is pronounced as a (short) [b] — basically for the same reason that /t/ sometimes emerges variously as [s], [h], [ɾ], etc. (And something interesting happens to the /r/ in that third syllable as well, but that's another story…)

In fairness to the field, I should admit that lenition of non-onset consonants in English has been noted by phoneticians from time to time. But more commonly, a couple of subcases are treated, separately, as the symbolic substitution rules of "flapping" and "t/d deletion". And the application to labial and velar consonants is rarely noted.

For lagniappe, listen carefully to Donald Trump's pronunciation of "contributions":

Now just the third syllable:

This one is about the phonetics of Engish vowel pronunciations — that vowel is a beautifully monophthongal high front rounded vowel [y], as in French bûche nf "log, numbskull". (And again, something interesting has happened to the /r/ in the /tr/ cluster at the start of the second syllable…)

As I said, this is an illustration, not a proof. And I've been known to improve my odds by setting the over/under at 40 seconds rather than 10 seconds. But still…


  1. Andrej Bjelaković said,

    April 6, 2017 @ 7:06 am

    "And something interesting happens to the /r/ in that third syllable as well, but that's another story…"

    I hope you'll tell us in the comments.

  2. Martha said,

    April 6, 2017 @ 9:11 am

    I had a friend in high school (and years later, a student, a different guy) named Slavik. I always caught myself calling him "Ssavik."

  3. Ross Presser said,

    April 6, 2017 @ 9:16 am

    The S-Town clip, it sounds to me like she's saying "It was wonderful at homecoming", not "Oh it's wonderful at homecoming."

  4. BZ said,

    April 6, 2017 @ 10:51 am

    Is Trump's use of "Our fellow Americans" to begin an address unusual? Although it is closely followed by "we", it seems clear that the "we" refers to the fellow Americans plus Trump. I suppose if I were a conspiracy theorist, I could theorize that Trump was speaking on behalf of some secret group that actually celebrates "women's history month", but in normal analysis I would expect "My fellow Americans" or just "Fellow Americans" at the start.

    [(myl) I believe that what he says is "My fellow Americans", which is a pretty standard opening for such addresses.]

  5. Pflaumbaum said,

    April 6, 2017 @ 10:52 am

    @Andrej Bjelakovic –

    I think he might have already told us in an earlier post: http://languagelog.ldc.upenn.edu/nll/?p=3349

    In fact the "-tr-" here and in "entrepreneur sound pretty much like plain /c/ to me – I can't hear the /r/.

  6. Jonathan Smith said,

    April 6, 2017 @ 11:18 am

    Is this connected to what happens with aspirates where in AAVE especially we can have word-final /kʰ/ > [g] and perhaps even deletion where in standard varieties there is at least glottal closure at the phonetic level? I am thinking before vowels/resonants, so "look like", etc. Maybe stress environment must also be specified…

  7. Ryan said,

    April 6, 2017 @ 6:53 pm


    That's the second syllable, not the third syllable. I think what Mark is referring to is the silencing of the /r/ after the /p/. If not, well, I've always pronounced the word that way.

  8. Geoff said,

    April 6, 2017 @ 8:18 pm

    Concerning [y]: I've been watching 'The story of film', a 15-hour documentary voiced by an Irishman (Mark Cousins). He has a beautiful [y], which is especially cute when, still in full-blown Irish, he drops it into foreign language phrases where it doesn't belong (Le Chien Andalou, A Bout de Souffle ….)

  9. Andrew Usher said,

    April 6, 2017 @ 9:13 pm

    Well, I was already aware of all these phenomena except the 'at home' one, which I'd have dismissed as merely glottaling of 't'. But close listening to the shortest clip seems to show no glottal stop – on the other hand 'h' is a partial glottal closure, is it not?

    Although dictionaries don't transcribe it and it's not talked about much, I am sure that in English 'tr' and 'dr' are affricates just as much as 'ch' and 'j'. When a bunched r is used sibilance developes, and can sometimes seem to even hide the initial 'r' – I don't think 'train' and 'chain' are actually homophones for anyone, though it's hard to understand what blocks the merger. Trump's 'contribution' really does sound as if it could be 'conchibution'.

    In my own speech, there is no sibilance but it's still easily seen to be an affricate.

    The silencing of the _second_ 't' in 'entrepreneur' is a different phenomenon – it's just a mistake of the same type as 'liberry'.

    k_over_hbarc at yahoo.com

  10. Katie said,

    April 7, 2017 @ 12:11 am

    I'm amazed at how much I learn about speech when, from time to time, I attempt a narrow transcription of real speech. I'm surprised no one ever made me do this in a phonetics class.

  11. Akito said,

    April 7, 2017 @ 12:43 am

    I am sure that in English 'tr' and 'dr' are affricates just as much as 'ch' and 'j'.

    I wonder about this. Mandarin has retroflex affricates in 出差 /chuchai/ [͡ʈʂʰu͡ʈʂʰaɪ]. Similarly sounding English words true try seem to have much longer [ʂ]s.

  12. MD said,

    April 7, 2017 @ 3:38 pm

    @ Geoff: Based on that description alone, I would say he's from Ulster (the northern historical province of Ireland which includes Northern Ireland). Am I right or am I right?

RSS feed for comments on this post