Recently in Dubious Linguistics Category

Still Contemplating Tones & Genes


A few months ago, I was remarking that I was dubious about the purported connection between geneses and tone language. However a new overview from Scientific American may convince me...just a little more.

In this version of the hypothesis, the innovation is LOSS OF TONE, not the use of tone. That is, the new genes, which are in the European and Eurasian populations, are making more likely that tone will be lost. This actually makes a little more sense historically. Despite the fact that Westerners consider tone to be fairly "exotic", it's actually pretty common around the world EXCEPT for Europe/the Middle East. Also, there are some indications that certain types of pitch accent markings have been lost over the centuries in several Indo-European languages such as Sanskrit.

If tone had been the innovation the genes would have been spreading from Africa to Asia bypassing India and the Middle East. This pretty much disputes modern thinking on how populations spread through the Old World.

Having said that, I still wonder how this genetic difference really affects phonology. I suspect there are many speakers of Central Asian non-tonal languages who theoretically have the right tone genes.

I also wonder if non-tonality is just an areal (regional) feature. If tones are the norm, and non-tones are the exception, then what you really need to explain is the lack of tone in Europe/Middle East/North Africa!

Beyond tone, languages in Europe and the Middle East often share similar features including distinctions such as grammatical gender in nouns (i.e. masc/feminine nouns). A regional account might explain why Central Asian speakers from Mongolia and Manchuria are still toneless despite being next to China. Historically, the language family may have also been influenced by also being next door to the Middle East.

A regional account would predict that tones could spread from Southeast Asia into Northern India such as in the case of Punjabi.

I'm also concerned that the other "exotic" phonological features don't have genetic cues. They didn't even find a genetic marker for CLICKS which is pretty much only used in Southern Africa (although again by different language families).

Is the use and development of tone so different from other phonological features...or do we in the West only think it is?

Tones & Genes - Still Dubious


Note: This entry replaces an earlier entry which has been de-published.

The controversy over the article by Dediu and Ladd on a possible connection between genes and tones has been simmering the past few weeks.

However, I do compliment Robert Ladd for doing outreach to the linguistic community. He's posted his own commentary site and contributed a guest post to Language Log. It's clear that he understands some of the implications of what he's claiming for language theory.

He invited linguists to write comments and I took him up on it. Still I remain skeptical, if more neutral. Here are some of my comments to his hypothesis.

Differences in Linguistic Ability?

It's clear that individuals differ in their ability to speak and use language, but Ladd is proposing that there could be subtle differences across populations. On the other hand most linguists tend to believe that language ability is the same across populations. That is, an average child of Asian descent has the same ability as a child of European descent.

Where does the original assumption of equal language ability come from? It's rooted in the observation that an infant typically gains the ability to become a native speaker in whatever language he or she is exposed to. Immigrant children coming to the US gain native speaker proficiency in English. More crucially, European children living in China can gain native speaker like proficiency in tone languages. A famous example is American author Pearl Buck.

Another aspect is that children tend to acquire language at the same rate, and in the same order. In most cultures, it is expected that five year old children can speak with a certain level of proficiency. In order for this to be the case, the assumption has been that language has to fit within certain parameters. I would expect that tone languages would fit within those parameters. If it didn't, I think you would expect that European children would have significant delays learning language.

Ladd says he will be conducting experimental studies to determine if this is the case. However, I would object to characterizing the "equality of languages" as dogma. It's a reasonable assumption given what we know about language acquisition in general.

Neighboring Tonal and Non-Tonal Languages

Ladd claims that the genetic propensity for tonal languages would be subtle. How subtle?

There are many cases of East Asian speakers NOT having a tone language. Mongolian and Manchurian are non-tonal and are spoken by "Western desert barbarians" (the kind of people the Great Wall of China was trying to keep out). Yet both a Mongolian dynasty (Ghengis Khan) and a Manchu dynasty (Qing dynasty of the "Last Emperor") were established in China. We know Ghengis's DNA got into China as well Manchurian DNA. Yet the tonality of Chinese has never wavered in that time...nor have Mongolian and Manchurian appeared to have moved to being more tonal.

Yet Ladd is predicting that one of the populations should have drifted either towards tonality (if they have the tone genes) or away from it (if they're missing it). Instead the [±tonal] distinction has been in place in each language family for centuries if not millennia.

Another question is if a language with a Middle Eastern or European population could be tonal. Well Ladd's map identifies one tonal language in the vicinity of Iran, and that is well within Middle Eastern "non-tonal" territory.

The predictions, if any, would be extremely weak.

It would almost be like saying that because many people of African descent have the ability to excel in basketball, it should have been invented in Africa. The reality is that was invented by an Anglo Canadian (James Naismith) in the US, and that it is played by people of all ethnic backgrounds.

You might be able to predict the line-up of the NBA, but the sport itself is enjoyed by people of all populations.

P.S. Speaking of Ladd's map - what's the tonal language of Japan? Ladd's text excludes Japanese because it's "pitch accent", but it can't be Ainu either.

What Tonal Advantage is There?

It's not clear what advantage the tone genes would give you.

I'm not sure it's a perceptual advantage. The inventory of the English intonation system (as itendified by Pierrehumbert 1980) includes both simple tones (H,L) and complex tones (*LH, *HL, etc) - the difference is that they are distributed across a sentence or phrase...and that we do not use tone to distinguish different types of vocabulary items. But English speakers regularly use differences in pitch to determine speaker focus and mood. It's not a universal intonation system either - non-natives have to be trained in it (see Learn English intonation)

It would have to some sort of switch in the phonological system that says "we can do phonemic tone" but again how does this interact with the ability of Pearl Buck to acquire a tonal language?

Or does it have to do with tonogenesis (the creation of phonemic tone)?

Origin of Tones

There are cases where we can see tones coming into a language. One is a dialect of Kammu (Svantesson & House, 1996).

In this case, the non-tonal dialect dialect distinguishes voicing (level of vocal cord vibration) in the initial consonants, but the innovative tonal dialect has reinterpreted that as tone.

East vs. North Kammu
E. Kammu (No Tone)N. Kammu (Tone)Translation
klaaŋkláaŋ (H)'eagle'
glaaŋklàaŋ (L)'stone'
taaŋtáaŋ (H)'pack'
daaŋtàaŋ (L)'lizard'

As it turns out, this is pretty much how all documented cases of tonogenesis works. A population reinterprets a distinction in vocal cord vibration for voicing or aspiration as a difference in pitch. This type of sound change where a sound is reinterpreted is not that unusual. For instance French nasal vowels are reinterpretations of vowel+nasal.

An interesting question is if the tone genes affect how you hear voicing NOT tones per se. Can a Middle Eastern or Asian language go through the same tonogenesis phenomenon?

Apparently Punjabi (related to Hindi) is classfied as a tonal language. It's population is in Northwest India, so is probably mostly "Middle Eastern" (it's relatively close to where the Indo-European speakers probably came into India)

So the answer is...yes?

So my conclusion on Ladd's hypothesis is still - I just don't think so.

Language ≠ Thought


One of the great philosophical debates of cognition (and science fiction) is whether language is identical to higher thinking or just a way to encode thought. One version of the theory of language as thought is called the “Strong Sapir-Whorf Hypothesis” which essentially claims that the structure of your native language’s grammar and vocabulary exclusively controls how you “categorize” the world.

A classic “example” Whorf cited was Hopi which did not use “tenses” such as English has. In the strong Sapir-Whorf hypothesis, this would mean that these speakers might not be able to distinguish the time of events in the same way someone speaking English would.

However, you may be surprised to learn that many modern theoretical linguists do NOT equate language with “thought”. That is, many theoretical linguists feel that many key aspects of thought are outside of language. Some examples are listed by Stephen Pinker in his book The Language Instinct, but here are examples of what I have found.

In terms of the Hopi example, discussions with native speakers did show that they could distinguish past, present and future events (and even have some grammatical cues for them).

Grammar Bypassing Meaning

Many people assume that grammar represents "thought", but there are many counterexamples to this. Two include:
  • The word for "car" in French can be either masculine l'auto or feminine la voiture.
  • English lacks the hodiernal past tense (past events happening today), yet English speakers can distinguish past events happening today from those which happened before yesterday.

Visuals Bypassing Language

Some information can be stored in "picture form" only and may be hard to translate into words.
  • I can drive home almost automatically, but have to “think” about how to give written directions in words. My internal driving instructions are not stored as a set of linguistic procedures but in some other format (visual, kinesthetic?). And I remember very few road names around town, just at which building to turn.
  • I’ve been to enough shows in Impressionist art that if I see a new work by Monet, I’m often able to identify him as the artist before seeing a description. However, I absolutely cannot describe in words or even pinpoint which combination of colors/strokes I’m using as cues.

Other Senses Bypassing Language

Ever try to describe a new flavor to someone? It’s almost not worth the bother. Yet, you can probably “taste” your favorite flavor of ice-cream.

Here are two way's to describe "cumin". I wonder which works better?

  1. One of the spices found in taco mix
  2. Spicy and nutty and almost sweet

There are no Words for:

Intuition Bypassing Language

I think we’ve all had that sense that “something’s wrong” or “doesn’t sound right” but could not describe what made you think that.

Language does have some influence on thought, but even so, it is possible for thought to go beyond the confines of the target language.

An interesting case is that Arabic distinguishes “aunt” (mother’s sister) from “uncle’s wife” whereas English only has the one word “aunt.” Yet divorce has actually given English speakers access to the distinction that Arabic already defined even if you have “no words” to describe it. If your mother’s sister gets divorced from her husband, it’s clear that she remains your “aunt.”

But what happens if your uncle divorces his wife and remarries? Do you have two aunts, one aunt? Is length of marriage a factor. Does having a cousin from either marriage count? Remarriage after all is why we have the famous question - “Do I have to call you Mommy now?”

What impact does this have on education

1. I think most educational specialists a priori assume that what you teach is somehow stored as words. Thus, it is assumed that instruction contains words which are “supplemented” by images or simulations. Yet, is that always valid? When it is valid?

An interesting case is learning new software. Newer users may desperately want a step by step users manual, but more technically adept users may be just as comfortable clicking buttons and seeing what happens.

Do we need a manual to help users or is it too much of a “crutch”? I confess I still believe in the manual for new users, but also know that a manual cannot ever represent the full capabilities of a technology (nor can it be used all the time if you want to gain any proficiency). Somehow a user must transition from manual (verbal) to manual (non-verbal) usage.

2. We have seen many language campaigns where alternate words are suggested in order to “change” people’s thinking. But do the campaigns change attitudes or must attitudes/reality change before language shift truly takes hold? When I was a child, the notion of a “Congresswoman” was almost a joke among some older men. Now that we have many more women in Congress, no one thinks twice about the term, and it was certainly necessary to designate the title “Madam Speaker” for Nancy Pelosi.

You could argue that you might need a term like “Congresswoman” to imagine a different reality, but some queens of Egypt like Hatshepsut were perfectly able to “imagine” the concept of "independent female rule" - they just co-opted the preexisting male terms and went about their business of suppressing rival opposition and administering the government. Language is powerful, but maybe not as powerful as imagination.