Why do tone and simple syllable structure appear to be correlated?

Question

I happen to have been struggling to learn a bit of Mandarin Chinese lately, and it's been my first attempt to really deal with tones to any significant extent. I find distinguishing tones quite difficult -- in fact, I find that trying to get tones right has assumed a considerable amount of the effort that I'm applying to studying this language.

All this is just anecdotal background for a more theoretical, typological question which I now find myself thinking about.

Why should it be the case that variegated tone systems should appear almost always appear in languages that have a) an isolating grammatical system and b) a relatively simple syllable structure?

Firstly, are there any very clear exceptions to this correlation? There seem to be few, if any, examples of heavily agglutinating languages (along the lines of Turkish or Finnish, say) which also have a complex tone system.

Why should this be the case? Have there been studies about nature (or existence) of this correlation?

Answer 1

I think that Zhen Lin's historical explanation is essentially correct: in order for tone to arise in a non-tonal language, some other phonological contrast is reanalyzed as tone. Tone languages can both lose tone and regain a complex syllable structure over time, but in general tone languages are likely to be closer to their "starting point". (Another way to think of this is that tone is part of syllable/phonological structure. Under the communicative requirements of human language, which seem to be fairly universal, individual languages can allocate their complexity to tone and other aspects of phonology in different ways.)

As for the typological claims, it is simply not true that there are no/appreciably fewer agglutinating tonal languages. The WALS (World Atlas of Linguistic Structures) allows you to cross-tabulate different linguistic features. Here's tone and a measure of morphological isolationism. There are 8 languages with a "complex tone system" and "concatenative [=agglutinating or synthetic] morphology"; only 5 have complex tone and isolating morphology. The tabulation of tone and syllable structure illustrates the generalization that more tonality correlates with less syllable structure, but the intersection of "complex tone" and "complex syllable structure" is not empty.

Answer 2

For a few exceptions to the tone/syllable-structure correlation, look at the Chatino languages — members of the Zapotecan branch of the Otomanguean family. Several of them have complicated tone systems and a syllable structure that allows some seriously gnarly consonant clusters.

Here is a paper on San Juan Quiahije (SJQ) Chatino that goes into some detail on the consonant clusters. In this particular variety, the maximal syllable is NCCVN or NCCV7. In other words, you can have a nasal and up to two other consonants in a cluster at the start of a syllable, and a nasal or a glottal stop at the end of the syllable. So this allows syllables like nskwan or ngya7. Also, both consonants in the initial cluster can have various sorts of secondary articulation. (For example, jwjya7 is a single CCV7 syllable, with jw and jy corresponding to IPA /xʷ/ and /xʲ/.)

I'm afraid I can't find a good English paper on tones in SJQ Chatino that's available online. If you can read Spanish, try this one, which goes into a good bit of detail. Anyway, SJQ Chatino has ten distinct tones, and also some fairly complicated interactions between tones in adjacent syllables.

(Disclaimer: The author of the linked paper is a classmate of mine. I'm not an expert on Chatino, I just work down the hallway from a few people who are.)

Edit: Aaron says above that "in order for tone to arise in a non-tonal language, some other phonological contrast is reanalyzed as tone," and that this tends to lead to phonological simplification. This is probably true. But in the Chatino languages, tone didn't arise out of some other feature. As far as we can tell, they've always been tonal. The proto-language of the entire family is reconstructed with tone, and so there was no need for it to arise out of anything.

The current theory is that the Chatino languages are descended from a polysyllabic language with a complex tone system — but that some of them (like SJQ Chatino) lost all their unstressed vowels. Those ended up as monosyllabic languages; they still have the same complex tone system they always had, but they also have those big gnarly consonant clusters where the intervening vowels have disappeared.

This isn't meant as a criticism of Aaron's answer. I just wanted to elaborate a bit on the historical difference between Chatino and (for example) Mandarin, since I think it sort of clarifies the typological picture.

Answer 3

I think that the basic observation about agglutinative languages generally not having complex tonal systems is pretty well upheld. In languages where morphemes are generally segmentable, and words are long, tonal systems will be found, as in many languages of North and South America, also some Oceanic language of New Caledonia; but languages with the most complex tonal systems, in Southern Mexico, Equatorial Africa and SE Asia, tend to have short words. This is not to say that such languages do not have much morphology; it's just that the morphology tends to be non-concatentive.

My guess for why this is is that the historical processes leading to very complex tonal systems tend to require short words as their endpoint. In Africa, languages having the most complex tonal systems (with four or five contrasts of pitch level at the surface) have gotten to be this way because syllables get lost, but their tones stay behind, causing perturbations. The more lost syllables, the more complex the tonal system, and the shorter the words.

In SE Asian tone systems, the systems have gotten complex due to pitch perturbations caused by the combined effects of word-initial and word-final consonants, so getting to the most complex system requires starting out with short words. Effects of consonants on pitch have created or complicated tonal systems in languages already having relatively long words, like Cherokee or New Caledonia languages, but the systems don't seem to get to the point where four and five pitch-levels are contrasted.

Some references:

• Haudricourt, A.-G. (1954). De l’origine ds tons en vietnamien. Journal Asiatique, 242, 69-82.
• Hombert, J.-M., Ohala, J. J., & Ewan, W. G. (1979). Phonetic explanations for the development of tones. Language, 55 , 37-58.
• Hyman, L. “Tone: Is it different?” Submitted to The Handbook of Phonological Theory, 2nd Edition (John Goldsmith, Jason Riggle & Alan Yu, eds).
• Kingston, J. (to appear) Tonogenesis
• Rivierre, J.-C. (1993) Tonogenesis in New Caledonia. Oceanic Linguistics Special Publications, No. 24, Tonality in Austronesian Languages, pp. 155-173
• Uchihara, H. (2009) High Tone in Oklahoma Cherokee. International Journal of American Linguistics, 75.3: 317-336.
• K. Wedekind (1985) Thoughts when Drawing a Map of Tone Languages

Answer 4

Another example of the same combination of gnarly syllable structure and tone is Tlingit, where words like [tʃχánkʼ] ‘grandkid’ and [ɬʔùɬ.tʃí.nì] ‘vest’ occur. Tone provably arose in this language from breathy and glottalized vowels without any change in syllable codas based on the extinct Tongass Tlingit dialect’s non-tonal vowel system. So it’s not the case that tone must evolve with simple syllabic structure.

There may be something to the idea that contour tone cooccurrs with simple syllabic structure, but even then Tlingit provides a counterexample. The Southern Tlingit dialect has a falling tone, a contour tone going from high to low. It predictably occurs in two situations: *Vˀ > V̂ː and *VːR > V̂ːR where R is a sonorant ∈ {/n/, /j/, /w/, /ɰ/}. I think that there are also some Athabaskan languages with complex syllable structure and contour tone, perhaps Southern Tutchone or Tanacross.