November 19, 2019

Stop-Affrication, or Stop, In The Name of Affricates!

I have spoken about glide affrication before - the process that makes "y" sound like "j" and "w" sound like "v." But did you know there's also affrication of stops in English? It makes "t" sound like "ts" and "d" sound like "dz." It can also make "k" sound like "khhh" and "p" like "pf".

What causes stop affrication? Well, it's a natural result of 'lengthening' a stop. A stop, by definition, is when a bunch of air builds up behind part of your mouth and is then suddenly released, causing a 'pop' sound. How can you lengthen this release of air? You can't, but you can continue pushing air from the lungs through, which causes a friction - hence affriccation. You can think of it like a stop combined with a fricative that's formed at the same part of the mouth.

Wells's Accents of English associates this sort of affrication with Cockney English speakers. But I've found examples of it across different dialectal regions. Like S-retraction, it seems to be a change that has developed multiple times in different places.

From New York City - musicians as distinct as Latin pop singer Ricky Martin and j-pop singer Utada Hikaru have d-affrication. Utada's example is especially song. (I recommend headphones to be able to hear the frication more clearly compared to the music.)

Her lips are dzevil red

The dzaily things

Billy Joel also has t-affrication at the end of a word:

If that's what it's all abouts, mama if that's movin' up, then I'm movin' outs... mmm I'm movin' outs.

Over in California, Disney Channel child actress Emily Osment (from Hannah Montana) has some dramatic d-affrication:

Dzoesn't matter what you say, I'm knockin' you dzown, dzown, dzown

And in Texas, fellow Disney Channel alum and singer Demi Lovato has some d-affrication, though not consistently (audio slowed down):

Are you kidding me? I'm so not a dziva

In the Midwest, the lead singer of the rock band Fallout Boy, Patrick Stump, frequently uses affrication in his music. Sometimes it's from lengthening the consonant. Compare the 'k' in 'mistakhhh' to the simply released 'k' in 'take':

And just one mistakhh is all it will take

Stump also seems to display a curious case of p-affrication where 'proof' sounds like 'pfroof' (at least, the official lyrics are 'proof'). This has led multiple fans and lyrics sites to mistakenly hear the lyric as "frozen fruit."

And here's the frozen pfroof

Some of these examples seem to result from attempting to 'extend' a stop consonant (Stump's "mistakkhe"). Some might result due to being near a front vowel, like Demi Lovato's "dziva." This sort of affrication before front vowels also happens in Canadian French (audio slowed down):

La fondation Celine Dzion

A similar thing happens in Russian. In Russian, oral stops that are palatalized (pronounced with the tongue raised, as if prepared for a 'yuh' [j] sound) also have some frication (source.

Affrication can progress even further to become lenition, so that the 'stop' component disappears completely. In Liverpool English, this can result in 'khh' simple sounding like a sharp 'hhh' [x]. The following clip was originally found via Dialect Blog and shows the [x] appearing in Liverpool footballer Steven Gerrard's speech.

Every player is loohing (looking) forward

I have [dz] as an allophone of /d/ in my speech. I especially have it at the beginning of a syllable, not at the end. Do you have any stop-affrication? What about stop-lenition? Are there any other examples you are familiar with, cross-dialectally or cross-linguistically?


  1. I can't watch the Ricky Martin video, I didn't bother with Céline Dion because I know the Canadian French affrication already, and I didn't hear Demi Lovato's the first time because it's not assibilation: the fricative is [ç] or thereabouts.

    But the other cases of d-assibilation just floor me. They're strikingly unambiguous, and I've never come across the phenomenon of unconditional d-affrication before, except when reading about one oasis in Algeria where */t d/ have become /ts dz/ across the board (and new /t d/ from other sources exist as well).

    In Liverp[ɸ]ool, on the other hand, they're simply replicating the High German consonant shift: all aspirates become affricates – /p t k/ come out as [pɸ ts kx~xkx~xː]. /t/ as [ts] is very widespread in England nowadays, there are even Conservative MPs who talk like that. This is also known from completely different languages: Burushaski reportedly has free variation [pʰ~pf~f] and [qʰ~qχ~χ] (while /kʰ/ is apparently going nowhere; neither are /tʰ/ and /ʈʰ/, but that may be because /tsʰ/, /tɕʰ/ and /ʈʂʰ/ plus their unaspirated conterparts are already in the system). d-affrication does not occur in such shifts.

    1. I'll make sure to add audio clips soon! It's been on my to-do list, but it takes quite a bit of time to extract and process audio (1 hour to process 12 clips the most recent time I did it!) and I haven't had the opportunity to do so. Also haven't been able to respond for same reason (though I don't need a whole hour to respond :) ).

    2. All clips uploaded now!

      How come you refer to it as assibilation? I do think there's still a [d] in there. Or do you think the result doesn't qualify as an affricate?

      My suspicion, which is not justified by any sort of research into the matter (and I'm not even aware of any research into the topic of it happening in contemporary American English) is that it started allophonically before high vowels, similar to French and (historically) Russian. Another hearer may have generalized this to a rule affecting all initial /d/, resulting in [dz]. This is based on nothing other than it feeling 'easier' for me to say 'dzi' than 'dzo', though I have 'dz' before any vowel.

      First time I heard anyone mention this was on an obscure linguistics community in 2008, referring to a song by a Black American singer from 2005 which I unfortunately can no longer find. At least the Ricky Martin clip is from 1999, which suggests it's not as crazy new as the other clips make it seem.

      Re: High German Consonant Shift: it's fascinating to see these things replicate themselves all the time, over time. I used to ask myself how these changes could happen without someone commenting on them, since they seemed so noticeable. And now I realize that they're happening right under our noses, and people won't comment on them until they're solidly established. :)

      The case of s-retraction is another one that's happened historically and seems to be appearing in English now, too.

    3. How come you refer to it as assibilation? I do think there's still a [d] in there. Or do you think the result doesn't qualify as an affricate?

      Oh, oops... an assibilate is an affricate in which the fricative part is a sibilant. I was just being more precise than necessary.

      I used to ask myself how these changes could happen without someone commenting on them

      For the HG consonant shift that's easy to answer: German and Longobardic weren't written in the 6th & 7th centuries*, and they weren't written about, so if someone commented, that's not documented.

      * Apart from a few Longobardic legal terms in laws otherwise written in ("barbarous") Latin. My favorites, both shifted, are the crimes of marhworf (throwing someone off their horse) and crapworf ("throwing" someone out of their grave).

  2. *facepalm* Tocharian.

    Tocharian completely lost the voice and aspiration distinctions of Proto-Indo-European, ending up with a simple plosive system of p t k. Yet, *d did not participate in the collapse (at least not all the time); instead, it stayed distinct as ts despite the absence of identifiable conditioning factors like front vowels.

    This has mystified generations of researchers. Well, apparently it's a thing that's known to happen...!

    1. It remains mystifying, because something like half the time *d did actually give t – it's not even clear what the default regular outcome was!