Linguistic commentary from a guy who takes things too literally

Fossil Phoneme Discovered in Living Language

Posted by Neal on January 14, 2011

If you (a) are not a linguist, and (b) have heard of “click” languages at all, it’s probably been in the movie The Gods Must Be Crazy. That was certainly my first awareness of this kind of sound. When I heard people in the movie talking, I found it hard to believe that the clicks were actually part of the speech, instead of a sound effect that had been added. A few years later, when I took a phonetics course in college, I was further surprised to learn that there was more than one kind of click; more like four or five actually. One of them, the dental click written as [|] in the IPA, even exists in English, but since we don’t have letters to represent clicks, we write it as “tsk”. (If you actually pronounce tsk, tsk as “tisk, tisk”, well, that’s not a click. You probably also pronounce ahem as “a-hem”, don’t you?)

Amanda Miller

In recent years, though, I’ve gone beyond surprised and into overwhelmed when I learned that five clicks is just scratching the surface. Thanks to the field research of Amanda Miller, one of my former fellow grad students, I’ve learned that there are on the order of 40 or 50 click consonants. It’s fascinating research, and she does it with technology that just didn’t exist a few years ago (because it was Amanda who developed it). This slideshow presents it well.

Cool though that all is, there’s more. I attended Amanda’s talk at the Linguistic Society of America conference last week, and as I listened to her, I was reminded of a famous story in the history of linguistics. In 1879, Ferdinand de Saussure published a paper about Proto-Indo-European, and hypothesized that it had had three sounds in its phonetic inventory that had morphed into other sounds in every known daughter language. Sally Thomason retells the story in this 2007 post on Language Log, and writes, “the idea of reconstructing unknown, unattested consonants did not appeal to traditionalists.” But several decades later, Saussure was vindicated with the decipherment of Hittite in the early 20th century, when two out of those three consonants were discovered in the Hittite texts. (That is, some words for which Saussure had proposed these sounds in PIE showed up in Hittite writing, with a mysterious character appearing where these consonants would have been.)

In her talk, Amanda talked about the African languages !Xung and Ju|’hoansi. In Ju|’hoansi, there is a pair of homophones, pronounced [gǃűű], which mean “water” and “belly”. ([!] is an alveolar click, a bit like [|], but sharper and louder.) Meanwhile, in the closely related !Xung, specifically the dialect spoken in an area known as the Mangetti Dune, the words aren’t homophones. “Belly” is still pronounced [gǃűű], but “water” is [gǁűű]. ([ǁ] is a lateral click, in which the tongue tip stays in contact with the alveolar ridge behind the upper front teeth, while the sides of the tongue move downward. According to the Wikipedia article, this sound is used by English speakers to call horses.)

The conclusion, then, is that in Proto-Ju, the ancestor language to M.D. !Xung and Jo|’hoansi, these words weren’t homophones. Suppose they had been. If the Proto-Ju word had been *[gǃűű], then we’d expect both “water” and “belly” to still be homophones in M.D. !Xung, pronounced as [gǃűű]. Likewise, if the Proto-Ju word had been *[gǁűű], then we’d expect both “water” and “belly” to still be pronounced as [gǁűű]. It would be highly irregular for the same sound in the similar phonetic environment (you can’t get environments more similar than in a pair of homophones!) to undergo a sound change for one word and not another.

So if “belly” and “water” in Proto-Ju weren’t homophones, how were they pronounced? “Belly” is easy: Since it appears as [g!űű] in both M.D. !Xung and in Ju|’hoansi, the most reasonable guess is that’s how it was in Proto-Ju, too. But what about “water”? We’ve already established that it most likely was not [g!űű] in Proto-Ju, since that would have made it a homophone with “belly”. So maybe it was *[gǁűű]. That’s where we’ll leave it for now.

Meanwhile, in 2003, Bonny Sands published a paper arguing that in Proto-Ju, there had been yet another click consonant, a retroflex click (in which the tongue tip curls backwards), which she wrote as [!!], which disappeared, gradually coming to be pronounced as [!] in Ju|’hoansi, and as [ǁ] in M.D. !Xung. Like Saussure’s reconstructed sounds for Proto-Indo-European, the sound [!!] was unattested in any known language.

However, Amanda has now found this sound, like a Coelacanth in the Indian Ocean, still present in a living (albeit endangered) language! With high-speed ultrasound technology, she has recorded this sound in the speech of a different variety of !Xung, spoken in the area known as Grootfontein. As in M.D. !Xung, “water” and “belly” are not homophones in this language. As we would expect, “belly” is once again pronounced [g!űű], but the word for “water” is [g!!űű], containing the heretofore unattested retroflex click! [UPDATE, Jan. 14, 2010: I should add that this kind of “minimal pair” data, in which a single difference in sound is all it takes to convey a different meaning, is the gold standard of evidence that two sounds are separate phonemes in a given language.]

So to sum up the parallel developments of the words for “belly” from Proto-Ju to Ju|’hoansi, M.D. !Xung, and Gfn !Xung:

  • Proto-Ju *[g!!űű] develops into Ju|’hoansi [g!űű], where the merger of [!!] and [!] creates the homophones for “water” and “belly” that exist today.
  • Proto-Ju *[g!!űű] develops into [gǁűű] in M.D. !Xung. It doesn’t create any homophones there.
  • Proto-Ju *[g!!űű] remains [g!!űű] in Gfn. !Xung.

To support this clasim, Amanda presented both acoustic evidence (waveforms, etc.) and articulatory evidence (the ultrasound data, plus palatograms and linguograms — results of a test involving painting the tongue or palate with a mixture of olive oil and charcoal dust, having the speaker make the sound, and then seeing where the oil/charcoal mixture has been rubbed off). Her diagnosis is that the merger of *[!!] and [ǁ] along the way to M.D. !Xung was motivated acoustically (i.e., the two sounded alike), while the merger of *[!!] and [!] along the way to Ju|’hoansi was motivated articulatorily (i.e., the two sounds are made in much the same way).

What I’ve summed up in this one post covers an incredible amount of travel, technical development, fieldwork, and lab analysis. An amazing piece of work!

8 Responses to “Fossil Phoneme Discovered in Living Language”

  1. Did you mean “Proto-Ju *[g!!űű] remains [g!!űű] in Gfn. !Xung.”?

    This is really cool stuff.

    • Neal said

      Oops. Thanks! I also took the opportunity to put in a comment on this kind of minimal pair that I forgot last night.

      • One of my professors made an interesting point in class about minimal pairs. (Well, actually, I think it was about allophones, but that shouldn’t matter too much.)

        When you say that a phoneme difference results in a meaning change, technically you are attaching phonology to semantics, which probably isn’t a good idea. He said when they were doing the tests (whoever “they” was), it was betting to ask “which one is different?”, rather than “which one has a different meaning?”, e.g. [pæt], [pʰæt], [bæt].

        The way I’m explaining it now doesn’t make it sound like there’s a difference, but when he explained it, it seemed to make logical sense that differentiation should be made….

  2. bearing said

    “([ǁ] is a lateral click, in which the tongue tip stays in contact with the alveolar ridge behind the upper front teeth, while the sides of the tongue move downward. According to the Wikipedia article, this sound is used by English speakers to call horses.)”

    Funny! I use it to get my children’s attention.

  3. Ellen K. said

    Is there any video or audio with the sound used by English speakers that’s represented by “tsk”? I’ve seen that plenty in writing, but I can’t connect it to anything at all that I’ve actually heard.

  4. cahillm said

    Amanda has done some pretty amazing research, all right. And it only adds to her presentation that she can actually pronounce all of them (though she did admit that she wasn’t too good – yet! – at the newly-discovered retroflex click). Good stuff.

  5. I first became aware of clicks in through Miriam Makeba’s hit recording of “Pata Pata“.

    As a matter of interest, “tsk-tsk” can also be spelled “tut-tut.”


    * * * * * * * * * * * * * * *
    … it isn’t a noise, it’s my language!
    — Miriam Makeba

  6. What I still want to see is clear, visual summary of Amanda’s results – e.g. a picture of what the IPA consonant tables might look like if informed by her research. The slideshow alludes to the possibility of drawing such a table but doesn’t show us, which I find frustrating.

    On this subject I was very disappointed in John Wells, who on his usually excellent phonetics blog has mentioned this topic only in the context of being, it appears, prejudicially dismissive of it.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: