Was the First Language Tonal?

I want to expand on last week’s post (about the FoxP2 gene) by considering another piece of research about genetics. A paper  published last May by Dan Dediu and Robert Ladd of the University of Edinburgh (author’s summary available on-line here, complete with links to the full article) argues that a population’s “genetic structure” can exert an influence on the features of the language spoken by that population. Specifically, they demonstrate that populations that speak tonal languages have one sort of genetic structure, while non-tonal populations have another.

Tonal languages are those like Chinese and many African languages that distinguish between words by altering their pitch. Non-tonal languages use consonants and vowels. Tonal languages use consonants, vowels, and pitch. A non-tonal language like English distinguishes between words by changing a vowel, a consonant, or both. Thus, we distinguish between do and shoe by keeping the vowel sound but altering the consonant. Grammatical distinctions can be made by the same changes. The Latin amas (you love) amat (he loves) depends on a change from s to t to make a grammatical distinction. In Swahili the change from k to v can distinguish between singular and plural forms: kitabu (book), vitabu (books). Tonal languages have three ways to change a word; thus, Chinese can keep the same consonant-vowel pattern do and do, and yet still distinguish between the two words by speaking one with a high pitch and the other with a low pitch. Tones can also make grammatical distinctions. The Masai language, for example, says the equivalent of saw he she, no matter who did the seeing and who was seen. Speakers use tone to indicate which word is the object and which the subject. Personally, I’ve always been glad I never had to learn a tonal language.

The Dediu and Ladd paper says there is genetic difference between populations that speak tonal languages and those that do not. For this blog, it raises the question of what kind of language was originally spoken, a tonal or non-tonal language. More generally, it suggests a new way of thinking about the process of speech origins.

Before going off into new waters, let’s remind ourselves of the basics that survive. The most important is that anybody can learn any language, or to be more precise: any normal human child will learn the language of the people where it is raised, even if that language is quite unlike the one spoken by its biological parents. This fact has been long known and serves as one of the pillar examples of the distinction between biological and cultural inheritances. Proof of the claim is going on today in a living experiment being performed in New York City’s borough of Queens, where immigrants speak almost 200 languages from around the globe while their children learn English.

Dediu and Ladd are claiming to have found something more subtle than a broad headline proclaiming a “gene for tonal languages.” They are talking about the “genetic structure” of a population. This term refers to the distribution of genetic variation in a group of people. For example, if we consider the eye color of the populations of Lagos, Nigeria and Oslo, Norway we will find the same eye-color genes distributed among both groups, but there are many more genes for blue-eyes in Oslo than in Lagos. That difference in the distribution pattern is what the concept of genetic structure draws attention to.

The genetic structure that is relevant to tonal languages concerns two genes implicated in brain growth and development: ASPM and Microcephalin. Both genes have “ancestral” and “derived” forms. Ancestral genes are older versions; derived ones are more recent variants on the ancestral gene. We can refer to the new versions as ASPM-D and Microcephalin-D. In both cases, the derived gene is relatively recent.

• ASPM-D is estimated to be only 5,800 years old, putting it within historic times. It has become common in central and western Asia, North Africa, and Europe, while remaining rare among the indigenous peoples of Sub-Saharan Africa, the Americas, and East Asia.
• Microcephalin-D is much older (37,000 years old), but still coming well after Homo sapiens began spreading beyond Africa. This version of the gene is much more widespread than ASPM-D and is common in Asia, Europe, and the Americas. It is also present to a lesser degree in North and East Africa, while remaining rare in Sub-Saharan Africa (apart from East Africa).

We still don’t know of any benefits for spreading these derived genes, although they probably were positively selected. They do not account for any variations in intelligence, brain size, social intelligence, or incidence of schizophrenia. Dediu and Ladd have found strong statistical evidence, however, that populations with a high presence of these derived genes in their genetic structures are much less likely to speak tonal languages. The authors examined a wide variety of other linguistic features as well and found no other correlations between language and genes, so the strong statistical correlation between genes and tonal languages was particularly striking.

Tonal languages are found where the ancestral gene predominates, suggesting that perhaps as recently as 40,000 years ago most languages were tonal. I say most instead of all because Australia provides a serious counterexample. Native Australians arrived on the island continent more than 40,000 years ago, but their languages are non-tonal. The authors write that Australia provides an important test case but that “obtaining reliable genetic samples seems very difficult.” So we don’t know if they have the ancestral forms of the genes.

Nor do we know why the changes in brain development put an end to tonality. Presumably, it was an accidental side-effect of some other change, but it is a surprising one, a little change in the neurological meat and, poof, a whole class of ways to distinguish between words evaporates. Officially, of course, the loss of tonality is not supposed to matter. Linguists insist that all languages are equal, but I confess to some suspicion of that doctrine and don’t believe any poet would hold it. Who would contend that the English of 1564 (the year of Shakespeare’s birth) was as good as the English of 1616 (the year of Shakespeare’s death)? The task of the poet is to find ways of saying things that previously wasn’t there. However, I’m not educated enough to know what cultural differences might arise from speaking tonal or non-tonal language differences. As a minimum I bet there is a whole extra dimension available for clever word play among the tonal speakers.

The most thought provoking element of this work is its attention to a population’s genetic structure instead of its history or an individual’s genes. It gives us a new way to think about linguistic change and speech’s evolutionary origins, but I will save those meditations for next week.

Comments

[Robin Shannon]

"Personally, I’ve always been glad I never had to learn a tonal language."

I think that you'll find that the language you speak IS tonal. Try saying the word "yeah" with HL, LH, HLH and LHL tones and you'll discover that the meaning of the different words you say are very different. Whats more a very large number of English's grammatical constructions are also differentiated by tone.
------------------------
BLOGGER: I thought about using the high and low tones of "present" to indicate different meanings, or maybe the rising tone to indicate a question; but these things are trivial compared to what goes on in a language classified as "tonal." Presumably linguists are talking about something when they make the distinction.

[Erasmussimo]

What's the relationship between tonality and syllable stress? Could prosodic structure be a watered-down version of tonality?

Are the various meters common to poetry in non-tonal languages in any way mirrored by tonal sequences in poetic forms in tonal languages?

[Robin Shannon]

They aren't trivial at all. In tonal languages, pitch is phonemically contrastive where as in "non-tonal languages" pitch (can be) morphemically/syntactically/semantically/pragmatically contrastive. It is still linguistically significant contrast just at a different level of the language. There is a view among many people who speak non-tonal languages that tonal languages are somehow significantally different from other languages, and it just isn't true. We all have a very finely adjusted linguistic understanding of pitch. Having said that, the distribution of FoxP2 and tonal languages is very interesting, but I think it is being over-sensationalised by lots and lots of people.

Note: tonal languages can also use pitch at levels higher than the phoneme.
---------------------------------
BLOGGER: Hmm. This whole discussion has sent me back to the text books (speaking figuratively; these days the texts books are on the web). I was quite startled to be told that I do speak a tonal language and have been looking around for some wisdom. A very useful site is the University of Minnesota at Mankato’s page titled “Introduction to Tonal Languages” (http://www.mnsu.edu/emuseum/cultural/language/tonal.html). Let me steal from it.

It is a mistake to say that a language is tonal because pitch differences occur in it. “All languages which have sounds have pitch differences. In tone languages, those pitch differences are used either to differentiate between word meanings or to convey grammatical distinctions.”

English is not a tonal language. “English is a stress language. Tonal languages differ from stress or non tonal languages like English where pitch doesn't have those same functions (Katamba 186). In a stress language, tone can be used to convey an attitude or change a statement into a question, but tone alone does not change the meaning of individual words. The function of tone is different in tonal languages. By using a different tone for one word, the meaning of that word can be dramatically changed.”

The ultimate test appears to be whether minimal pairs can be distinguished by pitch differences alone. “stress languages like English … do not have minimal pairs which can be distinguished by pitch differences alone”

Tone is one of many features for distinguishing between languages. Clicks are another striking feature of some. Word order is important in English, not so important in some others. All of these features are real, in the sense that they do distinguish between languages. Whether they are significant or not is another matter, which is why it is so surprising to find that what had seemed like just one more feature of a language turns out to have an unusual genetic underpinning.

[Robin Shannon]

Ok, I apoligise. I shouldn't have said that English is tonal, because tonal has a technical meaning to phonologists. What I meant was that pitch (not tone) is linguistically significant in English (and many other languages). English is a stress language and stress is manifested as pitch.

I haven't read (or heard of) Mankato, but I think you should take what he says with a grain of salt or two. "Tone alone does not change the meaning of individual words" in English OR in tonal languages. In tonal languages, tone does not change the meaning of one word, but rather tone individuates two words in just the same way as a different vowel or consonant individuates two different words.

Again, pitch (in English) does not differentiate between phonemes, but can and does differentiate between morhpemes. Your example of "present" and my example of "yeah" shows this clearly. These are in direct contridiction of Mankato’s claim that “stress languages like English … do not have minimal pairs which can be distinguished by pitch differences alone”. Yes amplitude differences can be involved but they need not be (and also amplitude can be involved in the manifestation of tones, so such an argument is irrelevent).

Anyway, my point here is that if one claims that ASPM is related to tonal languages, you have to be clear as to what you are saying. If you are saying that ASPM has something specific to do with orientation to pitch, then this seems unlikely given the very important role that pitch plays in non-tonal languages. If you are saying that ASPM has something to do with using pitch to as a way to differentiate between phonemes then firstly, this is a far weaker claim, and secondly, speech sounds in non-tonal languages differ from each other in terms of pitch. If speakers of non-tonal languages couldn't hear pitch differences in individual phonemes, they wouldn't be able to hear the differnece between /a/ and /i/.

Further the difference cannot be a length of time difference, since the pitch change on a single morpheme such as "yeah" is over the /ɛ/, ie. a single phoneme. In other words the only possible difference could be a specific relationship between pitch change over time and phonemes.

Yes, the corelation between tonal languages and ASPM is interesting, but it is not quite as amazing as it seems. Languages and genes both tend to spread with population movements. Now I don't know much at all about non-Australian historical linguistics, but if three seperate population (who happened to have the ASPM gene) spread through china, saharan africa and southern africa due to some technological/cultural advantage, then really it would be a much smaller coincidence and when taken with the tonal lanuguages in northern russia and southeast asia, and the ASPM genes found in Eastern russia and the pacific, it would look more like chance than strong evidence for a link between ASPM and tonal languages.

[yair shimron]

Having a hammer makes everything nails. The correlation between genes and tones is misleading not less than the sun running around earth. Genetics is a hammer not tuned for speech.

[István]

I found it interesting that tonality was "the only black swan" in the statistical correlational analysis - that made me think. why the only one?? Tonality seems to be very sensitive to neuroplasticity. A whole bunch of empirical work (e.g. Krishnan et al, 2005; Hsieh et al, 2001) suggest that speakers of tonal L's show different brain waves in the audit. cortex when processing tonal input. This also applies to adults learning tonality. I don't know of any other neuroscience research that says that first-language experience (the specific structure of a language) has such an effect on the neuronal connections (neuroplasticity!, of course, except for those studies on Chinese, Navajo and Hopi speakers, but these rather say that they use their brain differently - which is not neuroplasticity!?). What I'm suggesting is that the "missing link" btw. the brain size genes and tone bias might be explained by neuroplasticity, therefore the correlation.
But why the *tonal*-bias in the course of language evolution???
It may well be that the cognitive bias toward tonality was not triggered by chance (due to some changes in the brain) but rather people first started to exploit tones as a means of conveying different lexical meanings which became simpler with the advent of non-tonality (more words, though). It may have been a kind of trade-off. (I felt that Dediu and Ladd imply that the tone-bias could have as well been a non-tonal bias. This argumentation, however, does not leave room for the developmental position that language did evolve from refined vocalizations which turned out to be too hard after a while.)
It may be that tonal languages are more difficult (to process), and modern languages eased the work. Bichakjian (1999), for example, thinks that the protolanguage was a 'head-last' language which reguired right-hemispheric, global processing, but evolution made it simpler (hence, 'head-first' L's).

All in all, I believe in the hypothesis that the protolanguage might have been a tonal language - that sounds plausible.
NICE BLOG, thank you, keep up good work!!!