Tones and the brain

« previous post | next post »

People are always trying to exoticize things Chinese.  Now comes this article with the sensationalistic and patently suspect headline:

"If you speak Mandarin, your brain is different" (2/24/15)

This is based on the following paper:

Jianqiao Ge, Gang Peng, Bingjiang Lyu, Yi Wang, Yan Zhuo, Zhendong Niu, Li Hai Tang, Alexander P. Leff, and, Jia-Hong Gao, "Cross-language differences in the brain network subserving intelligible speech", PNAS (1/22/15).


Language processing is generally left hemisphere dominant. However, whether the interactions among the typical left hemispheric language regions differ across different languages is largely unknown. An ideal method to address this question is modeling cortical interactions across language groups, but this is usually constrained by the model space with the prior hypothesis due to massive computation demands. With cloud-computing, we used functional MRI dynamic causal modeling analysis to compare more than 4,000 models of cortical dynamics among critical language regions in the temporal and frontal cortex, established the bias-free information flow maps that were shared or specific for processing intelligible speech in Chinese and English, and revealed the neural dynamics between the left and right hemispheres in Chinese speech comprehension.


How is language processed in the brain by native speakers of different languages? Is there one brain system for all languages or are different languages subserved by different brain systems? The first view emphasizes commonality, whereas the second emphasizes specificity. We investigated the cortical dynamics involved in processing two very diverse languages: a tonal language (Chinese) and a nontonal language (English). We used functional MRI and dynamic causal modeling analysis to compute and compare brain network models exhaustively with all possible connections among nodes of language regions in temporal and frontal cortex and found that the information flow from the posterior to anterior portions of the temporal cortex was commonly shared by Chinese and English speakers during speech comprehension, whereas the inferior frontal gyrus received neural signals from the left posterior portion of the temporal cortex in English speakers and from the bilateral anterior portion of the temporal cortex in Chinese speakers. Our results revealed that, although speech processing is largely carried out in the common left hemisphere classical language areas (Broca’s and Wernicke’s areas) and anterior temporal cortex, speech comprehension across different language groups depends on how these brain regions interact with each other. Moreover, the right anterior temporal cortex, which is crucial for tone processing, is equally important as its left homolog, the left anterior temporal cortex, in modulating the cortical dynamics in tone language comprehension. The current study pinpoints the importance of the bilateral anterior temporal cortex in language comprehension that is downplayed or even ignored by popular contemporary models of speech comprehension.

Before critiquing the claims put forward by the investigators, let's talk about how tones really work in daily life in Chinese languages.

Tones are not absolute, they are not sacrosanct, they are not immutable.  Chinese often argue among themselves about what tones certain words should be pronounced in.  Let's take 著, which in most usages, but not all, can also be written as 着.

著 has the following pronunciations:  zhuó, zháo, zhāo, zhe, zhù, chú.  着 can have the first four pronunciations but not the last two; the very last pronunciation is used for only one purpose, as the first half of an ancient astronomical term.

After character simplification, when 着 was designated as the simplified (i.e., official) form, except in a few special instances where 著 still had to be used, the pronunciations and related meanings of the two characters got shifted around and reassigned by the script reformers, so that there are now two different sets of usages for these two characters in Taiwan and China.

The six pronunciations listed above (zhuó, zháo, zhāo, zhe, zhù, chú) have a veritable blizzard of meanings and functions associated with them (they must represent a number of different morphemes that became attached to these two characters over the centuries; the fact that three separate Middle Sinitic [about 1,400 years ago] pronunciations of 著 are represented by [i.e., have collapsed into] the modern pronunciation of zhù is fair indication that the phonology of 著 and 着 was even greater a thousand and more years ago than it is now).

Definitions (from Wiktionary)

Associated with the pronunciations zhuó (which I consider to be more literary), zháo (which I consider to be more vernacular), and zhāo (which is restricted to a few specialized uses at the end of this section) are the following meanings:

to attach

to touch, to contact

to wear, to put on, to be dressed in; clothing

to put, to place (dialectal Mandarin, Jin, Wu)

to arrange

to blossom or bear fruit

something to depend on, something to fall back on; desired end; a sense of belonging

to catch fire, to burn, to combust, to be ignited

to turn on (a light) (Cantonese)

to love deeply, to cling to and be reluctant to leave

to make (someone to), to order

to cost

to live in a fixed place

name of an ancient drinking vessel.

to need to, to ought to, should (Min, Wu)

to be affected by, to be troubled with

to get, to receive (Hakka, Min)

to fall asleep (dialectal Mandarin, Wu)

to fall into a trap, to be trapped

particle denoting that an action was "appropriately done"; "not done in vain".

particle denoting the success or continuation of an action

particle used after verbs to denote the severity of the event (Beijing Mandarin)

one by one (Cantonese)

correct, right (Cantonese, Hakka, Min)

at, in (a place) (Min Dong)

it's now …'s turn (Min)

all right; OK; that's exactly what I was thinking (dialectal Mandarin, Xiang) (zhāo)

move (in go games), step; trick, device (zhāo 招)

Associated with the pronunciation zhù:

notable, remarkable, striking

famous, well-known

to show, to manifest

to praise, to speak approvingly of

to write, to compose

literary work, composition, book

to record, to document

to establish, to set up, to build up

achievements; attainments

aboriginal, native inhabitants

precedence, order

the space between the front gate and the screens / shields

(佇/伫) to remain (at a place), to be held up

(貯/贮) to store

(褚) to stuff a lined garment with cotton

Associated with the pronunciation zhe:

particle indicating the continuation of an action or a state; often used with 正在 or 正

particle used after some adjectives to denote comparison of levels

particle denoting a command, request or advice

Just within supposed Modern Standard Mandarin (MSM), I have heard literate, learned Chinese speakers argue over which of the pronunciations of 著 and 着 should be associated with which of the meanings listed above.  An example cited by Brendan O'Kane, who is gathering notes for a piece with the working title "A Student's Guide to the Worst Character" (which will delve into the historical background of the problem), offers this example of uncertainty concerning the pronunciation of 著 in one particular instance, namely,

…despite what the dictionary says (and what one would expect given the meaning of 著 read in the second tone), I've heard 著急 [VHM:  "worry; be anxious"] much more often as zhāojí than as zháojí. The latter reading would actually sound weird to me — or would at least sound as if the speaker were trying to put special emphasis on the word.

I agree with Brendan:  I hear zhāojí more often than zháojí.

Brendan also offers the following humorously illuminating example of uncertainty over how to pronounce 著:

On the topic of 著, the character that dare not speak its name, I recently remembered a scene in the (very funny, almost unknown) Zhang Yimou movie Keep Cool (有話好好說) in which a mispronunciation is played for laughs: a migrant worker, hired by the film's main character to stand outside the window of the girl he likes and serenade her, keeps shouting "我睡不著覺" (wo3 shui4 bu zhuo2 jiao4).

VHM:  Most people would say Wǒ shuì bùzháo jiào 我睡不着觉 ("I can't fall asleep") [Google Translate gives Wǒ shuì bùzháo jué, but that's a separate problem of distinguishing between the two MSM pronunciations of 觉, jué ("sense; feel; become aware/awakened") and jiào ("sleep")].

Moving beyond MSM, in Cantonese the first group of definitions are variously pronounced as zau1 zau2 zoek3 zoek6.

In various dialects of Hakka, we have chok8 tiok8 do3, chok8, chok8 chok7, cok8 zok7, cok8, zhok7 chok8, cok8 ziok7, cok8 tiok8 do3.

Teochow diêh4 / dioh4 (tieh / tioh) diêh8 / dioh8 (tiêh / tiôh)

Mindong ciŏk / diŏh

Minnan tio̍h / tio̍k / to̍h

Wu tsaq (T4)

In Cantonese, the second group of definitions are variously pronounced as zoek3 zoek6 zyu3.

In various dialects of Hakka, we have chok8 tiok8 chok7 diau2 zhu5 (do3), zok7 | zu5 | cok8, cok8 tiok8 zok7 diau2 zu5 (do3), zhu5 | chok7, zhu6, zhu5, zu5, cog6 dau3 do3 zog5 zu4, zu5

Teochow du3 (tù) diêh4 / dioh4 (tieh)

Mindong dé̤ṳ

Minnan tù

Wu tsr (T2)

And we have only begun to scratch the surface.  If we start to get into regional differences just within so-called Mandarin (e.g., Shandong and its subtopolects, Sichuan and its subtopolects, Gansu, Yunnan, Hunan…, and there are countless other topolects, the tonal variations among them are endless.

Quoting this comment from the following post, "The enigmatic language of the new Windows 8 ads" (5/14/13):

If you are a speaker of MSM (Modern Standard Mandarin) and you travel across the Mandarin-speaking areas of China, you will be astonished at the enormous variety of tonal configurations for words. My wife was from Shandong and grew up in Sichuan, so I got a liberal dose of the tonal patterns of both areas. I became so conditioned to them that, if I were upstairs and someone called who spoke Mandarin with Sichuan or Shandong tones, I could tell very quickly to which my wife downstairs was speaking, since — even though she spoke beautiful MSM — she would switch into the very different tone patterns of her interlocutor. I often said that Sichuanese tones were "upside down" in relation to MSM (though they are consistent within their own phonological system).

Other examples of tonal variation within Mandarin are given in posts such as these:

"Ma Ying-jeou and Xi Jinping: presidential language notes" (11/21/12)

"English and Mandarin juxtaposed " (9/6/13)

Tones can be distorted for many reasons:  emphasis, emotion, musical tunes, hip-hop and rap, and so on.  There are many other reasons that the designated tones of characters may be changed or ignored.  For example, Cantonese waiters will use the shorthand form 才 ("talent; natural abilities; a moment ago; just now; only; not until), normally pronounced coi4, for the more complex character 菜 ("vegetables; greens; food eaten with rice or alcoholic drinks; dish; course"), pronounced coi3.

To summarize the first part of this post, tones are not absolute and inalterable.  Even with supposed MSM as spoken in Taiwan and on the Mainland there are are stark differences, while tones in Shandong, Sichuan, the northwest — all supposedly Mandarin — are dissimilar, and this is not to mention the radically different tonal systems of the non-Mandarin Sinitic languages.

Turning to the question of tones and the brain, I asked several colleagues who specialize in the cognitive aspects of East Asian languages their opinion of the article and paper cited above.  Their replies follow below.

From Bill Hannas:

If the study did no more than the linked article states, it fails to support its conclusion.  The claim, reportedly, is that phonemic tone accounts for observed differences in brain network activation between English and Mandarin speakers.  While these differences, if valid, are interesting, why didn't the researchers back up their claim by testing additional groups of speakers, e.g., Vietnamese (south: 5 tones, north: 6) or, for that matter, native speakers of other Chinese languages where the number of tones, their contours, sandhi rules and, arguably, linguistic importance vary from the Mandarin "standard"?  Am I missing something here?

From Jim Unger:

There may be an area somewhere in the right hemisphere that controls suprasegmental pitch adjustments (tone) in real time, but so what?  It certainly doesn't mean that the brains of Mandarin speakers are unique in some transcendental way.  There are loads of other tone or pitch accent languages in the world, even if they are a minority of all languages.  It doesn't prove that phonemic tone is transcendentally special:  there are many other "unusual" phonemic contrasts that occur in only a minority of languages, any one of which might also be handled in some specific brain area or other.  It does not add significant evidence to support sweeping theories of hemispheric laterality, on which Kosslyn and others have lately poured a good deal of empirical cold water.  In fact, given that speech production involves the coordination of numerous articulatory gestures in real time, it is more likely that pitch adjustments, voice onset time for obstruents, etc., are under the control of several competing loci in the brain, not just one, even if that one seems always to be active during the time that gesture is occurring.

From a psycholinguist who specializes on Chinese:

This is of course pretty surfacy stuff. Our brains are all different, and if you were to hold everything else constant and allow to vary only the languages that we speak, you'd find very minor, insignificant differences. Mandarin lights up brain areas about the same as any other language when the same tasks are performed. Functionally speaking, tasks focusing on 'contrastive tone' and 'contrastive stress' would activate the same brain areas, and those areas would only minimally involve R-hemisphere 'musical tone'.

From a colleague who is a specialist on writing systems:

Usually when I read a story on "here's your brain on Mandarin vs. English" there's some sort of link to Hanzi, along with entirely unsupported claims of their special and useful difference. So it's refreshing to see something without that for a change. Still, the article doesn't say anything about just how much difference was observed. And I'm wondering a lot about what is called "intelligible speech." Is that Mandarin as it is normally spoken or with the exaggerated tones of sing-song speech so common when someone is asked to read aloud?

Is tonal vs. non-tonal really such an absolute distinction? Are there degrees? And as the write-up notes, "Tone matters in English, just not to the same extent as in Chinese." So what happens in the brains of English speakers if given tonal variations on the "Where have you been?" question?

Another thing I'd like to see is the same study run with native speakers of tonal languages with different tones than Mandarin: Cantonese, Taiwanese, and Thai, for example. Are there differences among them? And would PRC-based scientists want to risk a study that might point to possible differences (however small) in the brains of Mandarin and Cantonese speakers? Heh.

Tones are not sacrosanct, nor do they have the ability to modify a person's brain.

[h.t. Geoff Wade]


  1. Gene Anderson said,

    March 3, 2015 @ 12:50 pm

    The comments sort of overinterpret the article; it just says that attending to tones makes you do some processing of speech in the right half of the brain, where music is processed (music has also some left-brain stuff going on). I suppose the authors are implying that any tonal language would cause the brain to develop thus. Interesting to me is the way children learning to talk deal with this. My older kids were raised part of their lives in Cantonese households, and they learned easily to speak with tones; it was harder for them to learn NOT to attend to tones as phonemic when they spoke English. I wonder what their brains were like at that time.

  2. J. W. Brewer said,

    March 3, 2015 @ 2:26 pm

    It is a commonplace, I think, of language acquisition scholarship that as kids acquire their L1 they learn to distinguish between sounds that are phonemically distinct in that language and do not learn to distinguish between sounds that aren't, and then once they reach a certain age (although they were neurologically quite capable of infants of having learned an entirely different L1) the brain "freezes" to some extent. So if you grow up an Anglophone you can easily distinguish, say, /d/ from /t/ whereas if your L1 does not distinguish phonemically between voiced and unvoiced stops it can be quite hard for you to learn to hear the difference if you're learning English as an adult. (The same is true of Anglophones trying e.g. to learn a language where the difference between aspirated and unaspirated /p/ is phonemic/contrastive rather than allophonic/complementary.) Given sufficiently advanced technology that can turn brain scans into pretty multi-colored digital images it would not surprise me to learn that slightly different neurons are typically firing depending on whether you were conditioned as a child into recognizing versus ignoring the voiced/unvoiced contrast.

  3. Bruce Rusk said,

    March 3, 2015 @ 2:51 pm

    Rather than comparing Mandarin speakers to English speakers–too many confounding variables–the interesting test case would be between similar languages or dialects that do and don't make pitch contour distinctions (there are dialects of Korean that do, for example, while most don't). The finding would be more robust if it held in such cases.

  4. AB said,

    March 3, 2015 @ 3:53 pm

    A comparison between Swedes (who use tonal pitch accent) and Swedish-speaking Finns (who don't) might also be interesting.

  5. Vanya said,

    March 3, 2015 @ 4:05 pm

    @Bruce Rusk. Sure, or compare Stockholm Swedish (pitch accent) vs. Finnish Swedish (stress accent).

  6. J. W. Brewer said,

    March 3, 2015 @ 4:52 pm

    A different problem with the Bruce Rusk suggestion might arise if e.g. Swedish-speaking Finns don't use tones in such-and-such way in their own speech but have significant experience listening to Swedish-speaking Swedes who do (adapt as necessary to the different varieties of Korean). It sounds from the abstract like the claim about Mandarin v. English has to do with what's going on in the brain when listening/interpreting the speech of others, rather than producing speech.

  7. Eidolon said,

    March 3, 2015 @ 5:36 pm

    @Bruce, AB, Vanya, Brewer: an easier experiment to conduct is with multi-lingual speakers equally fluent in both tonal and atonal languages, though I suppose such a study fails to control for neural-developmental arguments. Immigrant communities that differ in tonality from their host countries are also a wealth of data for this sort of experiment.

  8. Victor Mair said,

    March 3, 2015 @ 6:58 pm

    From a specialist on East Asian languages:

    I saw that article the other day (it’s being widely circulated and talked about on the internet). As soon as I saw the headlines, all I could do was laugh at it as another example of “the Oriental brain is just different from yours and mine”. And I have to say, I immediately thought of much the same thing your last expert says and just be a bit surprised and relieved that it’s not just another example of the wonders wrought by Chinese characters.

    But good grief. The fact that these “findings” were the result of research by Chinese at Peking University makes me wonder if the Chinese are not trying to outdo the Japanese and all their Nihonjinron claims about the uniqueness of their people and culture.

  9. Nathaniel Mishkin said,

    March 3, 2015 @ 9:17 pm

    Recently I received an email response that ended with the sentence: "Thanks a lot". Other parts of the response and earlier messages in the thread led me to think about tones in English–and wonder in particular what tones I should imagine the sender would use if he spoke that sentence. Two different sets of tones yield two pretty opposite meanings of the phrase: one snarky ("You did not help me and I'm not thanking you") and one not ("I really appreciate the help you gave me").

    Coincidentally (or not), the email was sent by someone in China, and not a native English speaker. The latter most likely means he was not familiar with the potential for a snarky reading of the sentence, but who knows? It did make me think about how tones in English extend beyond the rising tone at the end of sentences though.

  10. DMT said,

    March 3, 2015 @ 11:44 pm

    It's worth distinguishing between the claims of the original PNAS article (that Mandarin speakers show right-brain activation during speech processing, while English speakers do not) with the click-bait headlines (that Chinese speakers have "different brains"). The former are disputable but worth discussing; the latter probably aren't worth wasting too much time on here. I would be very interested to see expert (neuro-)linguists' evaluations of the claims of the original paper.

  11. Shannon said,

    March 4, 2015 @ 2:13 am

    As mentioned in the comments, why does it seem like people really seem to love to do studies with speakers of Chinese languages and English-speakers and claim a difference is due to speaking a tonal language, when tons of confounds, many cultural, are possible?

    There are European tonal and non-tonal languages, as well as Asian tonal and non-tonal languages, and African tonal and non-tonal languages, yet you often only hear English pitted against some variant of Chinese as the non-tonal vs. tonal match-up, and the results are often even framed in an West vs. East context (eg. claims about how non-westerners think differently) which kind of a distraction if your goal is just about tone (and not claims about anything else).

    Is this just because English and Mandarin are so widely spoken that there are many people to find for these kinds of studies? With international collaboration, getting beyond a comparison of those two shouldn't be hard. So many African languages are famously rich in tones yet you often read articles associating tonality with the oriental "East", without mentioning that it is even found at all elsewhere.

    It's true tonal — African language, or tonal European speakers might be less plentiful than Chinese ones, but for these studies, surely you could get volunteers in a reasonably diverse city or college campus, right?

  12. Shannon said,

    March 4, 2015 @ 2:18 am

    For the last part of my comment, I meant to specify, even if you are not going internationally, you could probably find them in a reasonably diverse city or college campus in the United States (or even in other western countries where immigrant communities are quite diverse), for those who are doing the study there, right?

  13. Scasc said,

    March 4, 2015 @ 2:25 am

    @Nathaniel Mishkin,

    tone does convey a *lot* of meaning in any "non-tonal" language, just not on the lexeme/morpheme level, but rather at the segment level.

  14. Michael Watts said,

    March 4, 2015 @ 2:46 am

    J. W. Brewer:

    No, children don't learn to distinguish sounds that their local language distinguishes while failing to learn to distinguish sounds that the local language doesn't distinguish. They're born distinguishing all human phonemes, and they learn not to distinguish the ones that aren't distinguished in their local language(s).

  15. Bob Ladd said,

    March 4, 2015 @ 2:54 am

    I agree with DMT about distinguishing findings from headlines. Quite apart from any specifically East Asian angle on this, people LOVE to make "tone languages" mysterious. The problem is that "tone languages" are really hard to define, and (for example) there's a respectable school of thought (currently most notably championed by Bruce Morén-Duolljá) that the Scandinavian languages are not usefully characterised as tone languages at all. All that's clear is that pitch is used in different ways in different languages, not that the world's languages fall unambiguously into two types. So if there's genuinely a difference in right hemisphere activity in people listening to Chinese rather than English, that's useful information for understanding the typological variety.

  16. DMT said,

    March 4, 2015 @ 5:21 am

    The idea of comparing tonal and non-tonal dialects of Korean or Swedish sounds good in principle, but I would worry that tone isn't a sufficiently prominent aspect of these languages to see a good signal in the MRI scan. (IIRC, tone in these languages is really more like pitch accent, but the typological question here is a difficult one, as Bob Ladd points out.) If it were possible to demonstrate a difference in right-brain involvement in speech processing across these sorts of dialect differences, so much the better, but I can understand why people might be reluctant to carry out experiments that could easily end up demonstrating nothing at all.
    On the other hand, there are also some really obvious experiments that the authors of the original paper could easily have carried out without needing to recruit further experimental subjects: exposing English speakers to the Chinese phrases, exposing the Chinese speakers to the English phrases exposing subjects to Chinese words without tone (e.g. with all syllables in first tone), exposing subjects to English words pronounced with distinct pitch inflection, etc.
    Why weren't these tests included in the original experiment design? Limited access time for the MRI scanner, perhaps? But in that case, wouldn't it have been preferable to do a richer range of tests on 5-10 speakers of each language rather than repeating a very limited range of tests on 30 Chinese and 26 English speakers? (I don't know anything about the requirements for generating statistically sound fMRI data, however, so perhaps the larger number of subjects really was necessary.)

  17. Jeffrey Willson said,

    March 4, 2015 @ 9:07 am

    I think the notion that Swedish is a "tone-accent" language really doesn't get to the heart of the matter. The one thing that distinguishes it from other, non-tonal Germanic languages is a special kind of accent that puts the syllable after the stressed one at a high and falling pitch. Danish is supposed to have a homologous "prosodeme" that's expressed as a glottal stop at the end of a syllable. There might be a close parallel between this Swedish/Danish prosodeme and the origin of the Chinese rising tone.

    When I visited Iceland, I was struck by how similar the overall prosody of Icelandic is to Swedish, more so than Danish or many Norwegian dialects. My mother, who is near-native level in Swedish, independently noticed the same thing. And yet Icelandic does not have a tone accent at all.

  18. Eric said,

    March 4, 2015 @ 10:28 am

    Like others have noted, the article is not nearly as sensationalized as the headlines associated with it. I didn't read it to be saying the Chinese people are unique in the world, but rather, a bit more modestly, that the characteristics of a language shape the functioning of the brain. In it's most generic form, this claim shouldn't be news to anyone as the facts reviewed above about children learning the sounds of their language make it clear that, at some level, their brains are 'different' after learning language X instead of language Y. The claim in Ge et al.'s article is that the differences are bigger than we thought.

    In fact, the hemispheric differences between tonal/non-tonal speakers aren't the most notable result. Hemispheric differences have been in the literature going back to dichotic listening studies in the 1970s, and imaging results over the past decade have shed more specific light on what areas of the brain are involved (note: this research usually involves Thai, Mandarin, or Cantonese as the tonal language of choice). Instead, the unique contribution, if correct, is the modeling of the relationships of networks of neural areas that are involved in processing speech, specifically that the way tonal and non-tonal speakers rely on left-hemisphere areas differ. Semantic areas are more heavily involved with speech processing in Mandarin, while phonological areas are more heavily involved in English.

    As is usual, assuming they are reliable, the results are fairly clear—there are specific patterns of activation within and between neural areas, and these differ for Mandarin and English speakers—it's the interpretation that is open to challenge. Their main argument seems to be that current models of speech comprehension don't pay enough attention to differences between languages and their results show why this is an oversight. (I'm not expert enough to critique the modeling techniques used, or to evaluate how unique their contribution to models of speech comprehension really is.)

    The part I find most speculative is that the authors attribute the reliance on semantic areas in Mandarin speakers to the larger number of homophones in Chinese—which I'm sure won't be a popular interpretation here.

  19. Jeffrey Willson said,

    March 4, 2015 @ 10:36 am

    Most of what I know about degrees of tonal nature is from reading "The Prosody of Greek Speech" by Andrew Devine and L.D. Stephens. In this book, which attempts to describe the tone accent system of ancient Greek, the comparative chapters describe a continuum from pure tone languages (such as MSM), through restricted tone languages, tone accent languages and non-tonal languages. The more that sentence-level intonational patterns tend to override lexical distinctions, the further down a language is on the scale. On this scale, Swedish is clearly very close to being a non-tonal language, while I suspect that most Sinitic topolects would best be analyzed as restricted tone languages. Japanese is usually described as a tone accent language, and one of the closest parallels to ancient Greek, but Devine and Stephens note that it has also been analyzed as a restricted tone language. They felt that Chinese was too different in its nature to be a useful example in reconstructing ancient Greek prosody, but I'm not convinced. I think the main difference is in the grammatical structure. Ancient Greek distinguished four different tone categories in monosyllabic particles with long vowels (proclitic, enclitic, grave, and circumflex), just like Mandarin.

  20. Bob Ladd said,

    March 4, 2015 @ 2:04 pm

    @Jeffrey Willson: But "proclitic" and "enclitic" are not at all analogous to Chinese tonal patterns. The thing about tones in Chinese (or in any "real" tone language, whatever exactly that turns out to mean) is that the tone is a property of the syllable pretty much irrespective of its relation to other syllables. The four tones of modern standard Mandarin can be defined and identified in isolation. By contrast, being "enclitic" or "proclitic" in Classical Greek is about a prosodic relationship to other syllables in the string, not about inherent tonal features.

  21. Eric P Smith said,

    March 4, 2015 @ 3:59 pm

    @J W Brewer, @Michael Watts:

    I'm not sure that there is any substantive difference between your viewpoints. Children are born, it appears, with the potential to learn to distinguish between sounds that are phonemically distinct in any language, and they lose that ability at a certain age. I don't think the phrase "all human phonemes" describes a useful concept. Is the English phoneme /r/ and the French phoneme /r/ the same phoneme? The answer's a lemon.

  22. Jeffrey Willson said,

    March 6, 2015 @ 8:31 am

    @Bob Ladd

    Well, in Jin (Shanxi) dialects such as Pingyao, the 陰平 and 陽平 tones are not distinguished in isolation, but are distinguished in context, by tone sandhi.

    Also, the tone patterns of the Jin dialects depend strongly on the grammatical function of the phrases. That means the tones in Jin Chinese are "about a prosodic relationship to other syllables in the string", as you say about Classical Greek. Complex rules such as those of Jin Chinese are not found in Standard Mandarin, but I suspect they are not unusual in other topolects. What is exceptional about Jin Chinese is that it is necessary to describe these rules in order to document the tonal inventory of the language.

    In ancient Greek, the interrogative pronouns are written (exceptionally) with acute accent regardless of position (τί what), or circumflex if they have a long vowel (ποῦ). This almost certainly indicates an inherent high pitch. The indefinite forms of the same words (τι some; που) are described as enclitic. Now an enclitic normally appears in the low-pitch part of a phrase. It is my belief that this enclitic classification was simply a convenient way of describing the inherent low pitch of indefinites.

  23. Katie said,

    March 6, 2015 @ 2:28 pm

    @Eric–thanks for the article summary. I don't have access. Curious that the original blog post relied so heavily on this being a difference about tone if the researchers were really driving at semantics. I don't know enough about neuroscience to have any idea whether that is a sensible interpretation of the results.

    But I don't think it's actually too crazy to hypothesize that Mandarin would rely more heavily on semantic processing than English does. Not just because "they have so many homonyms"–which is almost certainly measurably true, at least on the morpheme level–but that in general Mandarin relies more heavily on broader context for interpretation than English does. For example–null subjects are permissible in Mandarin, not in English. English speakers are fairly attuned to things like articles, past tense and plural endings, etc. that generally would require somewhat broader context to interpret in Mandarin (whether it's a syllable instead of a phoneme, or a discourse context). Of course I'm speaking in broad strokes here. But I think we could probably make this more precise and it would still hold true.

    So really, the appropriate follow-up wouldn't be so much to investigate other tone languages but to see whether languages that are like Mandarin or like English in their discourse patterns or (lack of) morphology pattern like Mandarin or English in relying more heavily on semantic than phonological processing. Messy hypothesis, for sure, but with some careful work I would imagine one could start to quantify and tease things apart.

  24. Edward Cha said,

    March 10, 2015 @ 10:19 pm

    Interesting topic – another interesting comparison could be English vs. ASL – which would seem (at first consideration) to be a bigger "difference" than English vs. Mandarin; however, at least one study ( found "no difference in the brain."

  25. PRice said,

    March 15, 2015 @ 6:43 am

    We can also infer from the “Mother’s voice and heartbeat sounds elicit auditory plasticity in the human brain before full gestation” study that the language-shaping of people's brains begins during womb life.

RSS feed for comments on this post