8 Apr 2007

Mommy, where do tones come from?

I've been ranting too much about European linguistics. I've been shamelessly unfair to the other continents who have patiently been reading my rants and feeling unrepresented. This is a big world so let's get cracking on... [Glen swiftly spins a globe and then stops it suddenly with his finger]... ah, yes, Sino-Tibetan.

Don't know what Sino-Tibetan is? Then check out this explanation from Berkeley University in California. Learn it because this is where the most popular language in the world comes from. No, no, not English, silly. Mandarin!

When I was a wee lad, there was this nagging question in the back of my quirky mind: Why does Chinese have tones? No matter how educated one happens to be, most people honestly haven't the foggiest clue about how to answer that question. Afterall they don't teach language origins in high school for some crazy reason. I guess if they started that programme, people would feel more united and have one less reason to start wars with other countries. Surely global comradery would cause a collapse of civilization. (Okay, that was a terribly bitter thing to say, but I just finished watching George Orwell's 1984 on YouTube and it just makes ya think, y'know?)

Anyways, it was only in my twenties when I finally discovered the juicy answer. The keyword here is called tonogenesis. Tonogenesis is the development of tonal contrasts in a language that previously didn't have any. Apparently, this happens all the time. It probably happened in languages before the Ice Age. It's certainly happened into the modern day, and if we live long enough and haven't evolved into waddling fish-people, there will be new languages with tones, perhaps derived from English, in the future.

Asian languages aren't the only tongues with tones. Did you know that Swedish has two tones? See The Tone-bearing Unit in Swedish and Norwegian [pdf]. And have you ever heard of a Nilo-Saharan language called Nobiin? Where have you been? Read this article entitled A sketch of Nobiin tone [pdf] from the University of Leiden. And we can't forget our Native American brothers and sisters, some of whom speak Navajo.

Tonogenesis in Sino-Tibetan is pretty interesting. Surprisingly linguists have reconstructed Proto-Sino-Tibetan with tongue-twisting consonant clusters and no tones. Frankly, it sounds more like Russian than anything we'd recognize in China today. All the tonal contrasts that Chinese languages now use (four tones in Mandarin; six or more in Cantonese) came after Proto-Sino-Tibetan. I have to chuckle every time I think about that crazy thought. It's a mind-bender. The word "eight", for example, is pronounced ba with high tone in Mandarin and as baat in Cantonese with middle tone. The two words, including their tone are related, of course. Sometimes you can even guess what the tone will be in Cantonese if you know the Mandarin word but, trust me, be careful with that because I personally have miscalculated a few times and had egg on my face. Both Mandarin ba and Cantonese baat go back to Proto-Sino-Tibetan *bʀyat but this word is without any reconstructable tone. Yep, just like Russian. So where did these tones come from then? Thin air? How does that work?

It turns out that voiceless and voiced sounds that we make unconsciously when we speak a language have an inherent acoustic frequency or "tone". In the word *bʀyat the initial consonant cluster is voiced and therefore gives off a low frequency but the letter at the end, *t, is voiceless and has high frequency. In other words, the surrounding sounds of the vowel were the seeds for later tone and they eventually determined the tonal shape of the word, whether the tone was low or high, whether it rose to a high pitch or fell sharply, etc. I suppose we could say that the vowel assimilated the unconscious frequencies of consonants and this evolved into more conscious tones, now intrinsic to the meaning of that word. Different Chinese languages have different kinds and numbers of "tonemes", just as all spoken languages have different kinds and numbers of phonemes.

Similar tonogenesis happened in Navajo, a language which derives from Proto-Athabaskan. Proto-Athabaskan also doesn't appear to have had tonal contrasts. This article Vietnamese and Tonogenesis: revising the model is also interesting for those who want to really get their hands dirty.

11 comments:

  1. G'day Glen,

    From a fellow anthropologist (though my emphasis was in SW Archaeology, not linguistics) who has since converted to the dark side of geography and GIS, I would like your opinion on a certain matter: After a lengthy discussion with my wife-to-be (who happens to be San Carlos Apache) about linguistic similarities between Athabaskan and Sino-Tibetan language groups, my curiosity was sparked. You see, she heard from her aunt about a fellow named W.H. Horton who, in the late 1800's, supposedly witnessed first-hand the similarities between Apache and "Tartar" Chinese.

    Here's a link to the referenced material: http://southwest.library.arizona.edu/hav7/body.1_div.1.html

    Given my limited background in linguistics, and inability to find much in terms of reliable studies on this, I'm sort of at an impasse. Save for entries by the likes of historian Ethel G. Stewart and linguist Robert Shafer, there aren't many in-depth articles that deal with this subject (at least to my knowledge). What's worse is that the aforementioned articles are fairly dated.

    So, long story short, do you have an opinion or any insight regarding this little gem of a topic?

    Thanks, in advance!

    Ray
    zia_neomexicanus@yahoo.com

    ReplyDelete
  2. Hi, Ray.

    Your bride-to-be sounds like an inquisitive lady. Congratulations on your upcoming marriage. Sadly, this subject is a little more complicated and vague than people might like and academia so far doesn't like to talk about long-range linguistic connections like these, being the overly self-conscious conservatives that they often are. The following is my description complete with a festival of relevant embedded links to click through.

    The long and skinny is, Proto-Sino-Tibetan is dated to about 4000 BCE and afterwards expanded outward from within China. Most information on the languages and reconstruction of Proto-ST will be found in a university library so it's worth a trip.

    As for the Athabaskans, they were one of the later peoples to come to North America and probably by boat, island-hopping along the Pacific coast following the Ice Age, around 8000 BCE or so. Their languages are concentrated to the west of the continent and they are considered a branch of the larger Na-Dene family consisting also of Haida and Tlingit.

    Now for the juicy mystery. There are some interesting similarities between the family groups worth exploring, I think. However, considering the above information and dating, we would necessarily be talking about languages that existed during or just after the Ice Age. This is why there is little information. It dates so far back in time that it's difficult given our current knowledge to be certain about the validity of any linguistic connections.

    You may notice that both Sino-Tibetan and Na-Dene are rich in consonant clusters, the kind that will make the unadventurous tongue-tripper bawl as in my Sino-Tibetan example *bRyat. Neither Sino-Tibetan nor Na-Dene have reconstructable tone since as I said, the tones in Navajo and Chinese came later. It's possible that there are some common pronouns between the two such as Sino-Tibetan *nei for Athabaskan *nə- in the second person singular, for example.

    You may find Edward Sapir and his Sino-Dene Hypothesis worth investigating published from the 1920s. It lives again under the form of Dene-Caucasian, strangely enough on Wikipedia. I dislike linking to Wikipedia but in this case it may give you important leads. As always, reader beware. One note, the Dene-Caucasian description here tries to include Sumerian and Algic but this is really whimsical to me and may only serve to derail your quest. (In my view, Sumerian is much more likely to be Nostratic as already published by Allen Bomhard but nothing is conclusive. But Nostratic has nothing to do with your question.)

    Damn, why didn't I just write another blog ;)

    ReplyDelete
  3. Egad, I misspelled "Allan Bomhard" as "Allen". I know how irked I get when people try to needlessly add an extra "n" to my first name, so apologies to Mr Bomhard ;)

    ReplyDelete
  4. Some dialects of Panjabi have tones, so maybe a similar thing happened there. The letters that represent voiced aspirated (murmured) consonants in other Indo-Aryan languages are instead used to mark the tones.

    ReplyDelete
  5. Yes. You're apparently not as 'goofy' as your name suggests :)

    If you read this pdf of an article from the University of Hawaii by Mark Alves entitled The Vietnamese Linguistic Belt Buckle: An Example of Sprachbund in Southeast Asia (or if you click Google Books link here), you'll have your answer: "Consider the case of Panjabi. In this language, aspiration caused a tonal distinction not seen in other Indic languages (Hock, 199, pg. 98). An /h/ before the vowel caused a low tone on the following vowel. When the /h/ occurs after a vowel, it gains a high tone. This instance of a following aspirate causing a high tone (although not a rising tone) gives weight to the claim that Vietnamese could have developed the rising tone for similar reasons."

    ReplyDelete
  6. So what are your concluding thoughts on the claim that Proto-Sino-Tibetan (PST) was a non-tonal language?

    Do you fully agree with this assumption?

    Most, if not all, of its daughter languages tend to be monosyllabic and analytical. This is the reason why Chinese languages still use a logographic writing system as they have done for millennia; as opposed to developing an alphabetic system.

    The ancient Sumerian was also of this type, using a syllabic system known as "cuneiform".

    In order for an ancient Chinese person to distinguish between two similar sounding syllables, of which perhaps Old Chinese probably had many, then tones would do just fine in distinguishing them.

    Then again, Chinese language, at least in its Mandarin form, is still not free of homophones.

    Nevertheless, I doubt tonality is just a new, random invention that almost all S-T languages happened to acquire at some point in history. Is it plausible that a language family of 400+ languages such a phonetic shift in just 2000 years?

    Is it likely that PST could have started off with an agglutinating morphology with no need for tones, and then produce hundreds of analytical tonal languages?

    Are we to assume that the linguistic ancestors of almost all Sub-Saharan African languages underwent the same kind of evolution? Or could tonality be a very prehistoric aspect of human language that made it "Out of Africa", but got secluded to a small geographical setting as in the case of East/Southeast Asia?

    No, wait... Tonal languages are also found scattered throughout the Americas and are not limited to Na-Dene speakers. New Guinea and some other Pacific island regions also have a couple tonal languages. The only groups of languages where you will not find tonality are those of most Austronesian (excluding Cham, which has been influenced by surrounding tonal languages, and Iaai) and Australian Aboriginal languages.

    Sorry about the long rant, but this issue has been bothering me for a while. But it all boils down to the question of "Common ancestry, convergence, or coincidence?" Maybe in reality it's a mix of all those categories, but in my opinion, more vigorous research needs to be done.

    ReplyDelete
  7. Kalvin: "So what are your concluding thoughts on the claim that Proto-Sino-Tibetan (PST) was a non-tonal language?"

    The overall evidence shows that the tones in Modern Chinese languages like Mandarin and Cantonese are more recent. They can't be traced back coherently to Proto-Sino-Tibetan. So we can say rather conclusively that there's *no evidence* of a PST tonal system. Yet note that this can't rule out the possibility that PST had its own tonal system that was somehow lost in modern languages and replaced with new ones.

    However without evidence, this is all idle speculation. So the optimal and precise answer is that PST had no perceivable tonal distinctions that need be reconstructed.

    "Most, if not all, of its daughter languages tend to be monosyllabic and analytical."

    It's a grave mistake to assume that this was always the case and, in fact, Mandarin has pushed towards disyllabic words (eg. Cantonese meng versus Mandarin mingzi "name" with added noun formant -zi).

    In PST, it's pretty clear from the wealth of data that there were once consonant clusters and these clusters have been drastically reduced in Chinese while surviving into Classical Tibetan. This later consonant reduction is in itself strong enough motivation for the more recent tonal contrasts developing in order to avoid a tsunami of resulting homonyms.

    "Is it plausible that a language family of 400+ languages such a phonetic shift in just 2000 years?"

    Yes. The change is too historically accounted-for to ignore. And stranger things have happened.

    "Is it likely that PST could have started off with an agglutinating morphology with no need for tones, and then produce hundreds of analytical tonal languages?"

    Yes because in this case the later tonal systems cannot be rationally derived from PST. Comparison between Chinese and Tibetan demonstrate later innovation. If *you* can demonstrate their connection, by all means, but the evidence just says no and so then do Sinotibetanists.

    "Or could tonality be a very prehistoric aspect of human language that made it 'Out of Africa'"

    On this blog, I expect commenters to limit themselves to logic and Occam's Razor. Here you are speculating wildly about a timeframe that is completely non sequitur to ST and Chinese. One can never solve problems if one's imagination isn't tempered by a methodical and persistent preference for efficient solutions over random ones.

    "Maybe in reality it's a mix of all those categories, but in my opinion, more vigorous research needs to be done."

    Indeed, and not to pick on you, but yours especially. The topic of Chinese tones is terribly complex and I can't say that *I* know the whole picture about how Middle Chinese cultivated Mandarin and Cantonese tones. This is in itself a lifetime of study. While I have a moderate understanding, it's outside my main focus since I've been concentrating more heavily on Aegean linguistics. However I'm aware of some of these broader issues of Proto-Sino-Tibetan reconstruction and have learned to speak basic Mandarin and Cantonese. If one studies these two languages, one can see that their tonal systems are related at least.

    ReplyDelete
  8. By the way, here's an excellent overview of Sino-Tibetan languages for those who feel the itch to know.

    It says in more detail what I've explained: "For example, in Sinitic the tones developed out of consonant suffixes (*-s, *-ʔ) and loss of initial voicing (Baxter, 1992: 8.2), and in Lhasa Tibetan the tones developed independently, out of loss of initial voicing and the influence of final consonants."

    ReplyDelete
  9. Glen: "They can't be traced back coherently to Proto-Sino-Tibetan. So we can say rather conclusively that there's *no evidence* of a PST tonal system."

    Is there any full, more recent PST reconstruction in existence that all linguists would agree upon? Apart from Baxter's reconstruction? Given that many ST languages are poorly documented and that there is no concrete Old Chinese phonology available, how convincing can a single reconstruction be?


    "It's a grave mistake to assume that this was always the case and, in fact, Mandarin has pushed towards disyllabic words (eg. Cantonese meng versus Mandarin mingzi "name" with added noun formant -zi)."

    This tells me nothing. "Ming2 zi5" is a disyllable made up of two monosyllables. There is no reason to suggest that these syllabic combinations makes Mandarin polysynthetic any more than Finnish language is isolating. As you said "-zi5" (字) is a noun formant, but nevertheless carries independent meaning of its own (word) when said in isolation. In other words, it is not a bound morpheme such as the "-er" in the English word "worker".


    "In PST, it's pretty clear from the wealth of data that there were once consonant clusters and these clusters have been drastically reduced in Chinese while surviving into Classical Tibetan."

    And Old Chinese (which is known to be monosyllabic), and for sake of argument, the Tamang languages, predate Classical Tibetan. Old Chinese, however, apparently allowed consonant clusters, similar to the syllabic structure of Khmer.

    For example, the old Chinese reconstructed word for "moisten" is *trem (沾; "zhan1" in Mandarin, "zim1" in Cantonese). This word is monosyllabic and corresponds to a single logogram. Though a tone is obviously not reconstructed for the morpheme, it is possible that it could have been more of an "unconscious" tone. That is, before it developed into the established high tone in Mandarin and Cantonese.

    An other possibility is that there was no need for tone distinction; I myself could not find any obvious homonyms for it. Nevertheless, it "decays" into the Mandarin "zhan1" which has a homonym of different meaning.


    "On this blog, I expect commenters to limit themselves to logic and Occam's Razor. Here you are speculating wildly about a timeframe that is completely non sequitur to ST and Chinese."

    My apologies for having made myself misunderstood about the point I was trying to make. I was actually asking a rhetorical question of whether or not independent development of tones is an "all-or-nothing" in terms of final consonants (Can common inheritance from a distant ancestor be a better Occam's Razor than a tortured reconstruction for *no* reconstruction of tones?). I am by no means suggesting an "Out of Africa" origin of Sino-Tibetan tones to be the actually case.


    "Indeed, and not to pick on you, but yours especially."

    What do you mean by "yours especially"? If you mean my view on a tonal, monosyllabic structure for PST (or perhaps Old Chinese), I guess it's a matter of opinion. However, Old Chinese is known to be monosyllabic.


    "The topic of Chinese tones is terribly complex and I can't say that *I* know the whole picture about how Middle Chinese cultivated Mandarin and Cantonese tones. This is in itself a lifetime of study. While I have a moderate understanding, it's outside my main focus since I've been concentrating more heavily on Aegean linguistics."

    Indeed it is. I myself am not an expert in Sino-Tibetan linguistics; I like to think of myself as specializing more in Pacific linguistics.

    Whatever the real case may be for tones in PST, we will just have to wait for more conclusive evidence...that is, if we can find any.

    ReplyDelete
  10. Oh crap. To complicate matters I must keep reminding myself that míngzi is spelled 名字 and not *名子 (the last character is a different zi from the noun suffix I'm referring to) and a more apt example of it is 帽子 màozi 'hat'.

    ReplyDelete
  11. May as well mention one more thing.

    "Old Chinese, however, apparently allowed consonant clusters, similar to the syllabic structure of Khmer."

    Yes, but ST prefixes fall off in the Sinitic branch even by Old Chinese. Eg. Tibeto-Burman *g-sum (Matisoff) versus OC *srum (Baxter-Sagart).

    ReplyDelete