Thursday, January 22, 2009

Guest post: Burning down the house

This in, slipped under the door of a young colleague yesterday, signed only "√v".

As a minion of Mr. V, I do not like reading but sometimes I am forced to do it. I was pleasantly surprised to find that people are apparently working on the relationship between language and genes. Specifically, an article in Science Daily about language and genes was forwarded to me (reading email doesn't count). I had to read through the article a few times to fully understand it and have to go with what the article says about the research since the actual article hasn't hit PNAS yet …

The first part of the article makes perfect sense and I really can't imagine anyone who works on language to really disagree with the ideas. A few confusions need to be fixed though and some clarity added. The important summary sentence is:
By modelling the ways in which genes for language might have evolved alongside language itself, the study showed that genetic adaptation to language would be highly unlikely, as cultural conventions change much more rapidly than genes.
This research contains all of the important aspects of genes, language, culture and variation that need to be considered to say something enlightened about this question but unfortunately gets all kinds of things confused.

The main form of the argument as suggested by the SD article is that
(1) we know the time rate of genetic change
(2) we know the time rate of language change
(3) a similarity between the time rates of language change and genetic change would suggest that there could be a causal relationship between them
(4) we know that the time rates for language change and genetic change do not match (language changes too quickly)
(5) therefore, there can not be a causal relationship between language and genes
(6) therefore, culture must determine language
Steps (1)-(3) are rather straightforward and sensible. Its when we start getting to steps (4)-(6) that things begin to get a little strange …

We have to unpack (4) to ask deeper (possible philosophical) questions about language change. Specifically, what do we mean by language change here? More importantly, how tight is the connection between specific language features (e.g. words, rules, word order, etc.) and genes.

Many sensible people I know assume that the particular words of a language (e.g. gato, cat, kuching, chien, etc.) have no genetic basis what-so-ever and instead they are learned. If particular words were genetically encoded then we would expect much less flexibility in infants to learn particular languages when geographically displaced from genetic distribution areas. It would also be very difficult to explain how we continually learn words throughout our entire lifetime. Or, if we flip this around, the cost of genetically encoding specific words would be extraordinarily costly from an evolutionary point of view.

These same sensible people make similar assumptions about having to learn language specific rules, constraints, word orders, pragmatics, sociophonetics, etc. All aspects of language that vary crosslinguistically. This variation must be learned from experience and it appears that any homo sapien child that is cognitively normal can learn any language they are exposed to.

So the question is how to answer the question of whether 'language' has changed or not given a particular example of say Old English to Modern English.

Sure, language has changed in this example and we know this because a speaker of Modern English can not understand Old English. We can get more specific than this and show particular changes in the phonology, morphology and syntax of these two distinct languages. More importantly we can probably say that 'culture' has caused the change from OE to ME if we assume that culture is the accumulation of human behavior that other humans experience from birth. Put in a simpler manner, a baby born in the 10th century got input from a different language than a baby born in the 20th and thus it should not be a surprise that a different language is spoken by both.

The culture has changed and thus the language has changed but have the 'language genes' changed? This is where I think the SD article (and possibly the PNAS article) begins to go horribly wrong.

It would be an unimaginable state of hubris to even suggest that we know even the beginnings of how genes map to higher level behaviors such as specific languages. This is where the Royal Linguistic Society in London may have had it right in banning discussion about the origins of language. This is where the main flaw in the argument outlined by the SD article arrises. In order for the argument to go through, specific languages must be directly coded in the genes. In other words as Old English gradually changed into Modern English there must have been a similar and correlated change in the genetic make-up of speakers of Old English to Modern English. At this point in time, all sane people should rejoice that this hypothesis has been excluded by this line of research. Hooray.

The real question for people who want to investigate the relationship between language and genes is 'who actually believed the above hypothesis in the first place?' No progress in our understanding of language or genes or the mapping between the two is made by burning straw men. Progress is only made when truly informed theories are being compared. Unfortunately, most of the current work on genes and language is embarrassingly devoid of any knowledge of informed theories of language.

An informed theory of language will answer at least the following three questions:
(1) what is unique to each language
(2) what is universal to all languages
(3) how do children learn language
At the current time, we do not have anything beyond a beginning of a sketch of an answer to any of these questions. If these are good questions though, we can see where useful work in paralleling our knowledge about genes and language can occur.

What is universal to all languages should be universally encoded in our genes somehow. Looking at human behavior in an abstract enough manner may unveil the stability that would be required to allow a Baldwin type of effect to encode these abstract schema into the genotype of homo sapiens.

Having an answer to how children learn language is likely to improve our understanding as to what is encoded in the homo sapien genome specific to language. It appears that there is a surprising cross linguistic stability to how children learn language (both spoken and signed) which again suggests a general genotype.

Finally, we have to understand what is specific to each language because this will clearly demarcate what is not part of the homo sapien genotype. This will clearly highlight the role of culture in determining aspects of language.

The most beautiful aspects of these questions (all of them not just the last three) is that I have the strong suspicion that at one point in time most of the answers will be empirically based even if some of the answers suggest a rational basis for some aspects of human nature.

The most frustrating aspects of these questions is that most people are not patient enough to wait for our understanding of language to progress enough to be able to even ask good questions. The most likely cause of this impatitence is an expert's illusion in that since we use language everyday with little thought, it must be a very simple phenomena. Thus, who needs to really nail down the details … let's just fire up the gene sequencer and do some heavy math … that does sound cool though …


PS: I don't know who you are, √v, but you're welcome to join Team Verb.


Laura Staum said...

This is a very interesting post, and I'm inclined to agree with many of the points. One, however, struck me as worthy of reconsideration:

What is universal to all languages should be universally encoded in our genes somehow.

This seems sensible at first glance, but I think it's worth considering that there may be other universal aspects of the world than genes that could cause universalities in language. To use a silly example, suppose it were to turn out that every language in the world has a word for "sky". This would constitute a "universal," in some sense, but most people would tend to attribute that to the universality of the sky in all humans' experience - not to something genetic. Such explanations may be harder to find for structural aspects of language, but that does not mean that they don't exist - universal aspects of the physical world could well be to blame for many universal aspects of language.

Mr. Verb said...

My guess is that √v, who presumably reads the blog and will follow up on comments himself/herself/themselves, would regard 'universals' of the sort you mention as different beasts.

√v said...

I agree with Larua above that I oversimplified the 'universal = genes' statement. But we must recognize that given the Baldwin effect (and its apparent reality) that the 'sky' example given by Laura presents a perfect example of where environmental stability makes it an empirical question as to whether the (hypothetical but likely true) fact that 'all languages have a word for sky' is encoded in genes or not. On the one hand, Laura is absolutely correct that we could derive this uniformity in language from uniformity in environment. On the other hand, this exact type of long term uniformity is the exact condition that is necessary for the Baldwin Effect to operate.

Kudos for picking an abstract example to talk about though. There are examples of this type (kinship systems, color systems, pronouns, etc.) where all languages have particular instances of a given feature. Many of these features are also culturally universal. Thus, we find ourselves in the uncomfortable position of being able to recognize that we may have produced a question that we are currently unable to answer. I think understanding the nature of questions is much better than asking stupid questions.

Joe said...

Nice post. Just for the record: It is insane to pretend that we know anything real about the 'rate of linguistic change'.

Anonymous said...

Please post again, floydy!


Joe said...

I second that!