Why has language only evolved once?

[koganbot]

I'm reading a book by Terrence W. Deacon called THE SYMBOLIC SPECIES: The Co-evolution of Language and the Brain that I pulled off the philosophy shelf at Barnes & Noble and then borrowed from the library, and I'm loving it. The author is unknown to me (which doesn't tell you anything about how well-regarded he might be; but a book that's on store shelves a decade after it was published is sure doing better than my book), I don't have the background to evaluate whether he's doing right by the information he's drawing on, and I don't know the linguistics discourse he's responding to so am only provisionally assuming he's right in his presentation of other people's ideas.

What I like is that he asks questions, and they seem like good questions, and it seems to make sense that he's asking them. Here's something that needs to be explained, he's saying, and he's saying that he doesn't like the explanations given so far, so he's going to come up with his own.

Deacon was talking to an elementary school class, and an eight-year-old asked him, "But don't other animals have their own languages?"

Deacon responded by explaining how human language is different from any other form of communication in its precision and its capacity to combine elements logically but with open-ended possibilities.

"Do animals just have SIMPLE languages?" the kid continued.

Deacon said that though animals have calls and such they have nothing like words, nouns, verbs, or sentences.

Another child asked, "Why not?"

And this stumped Deacon, who had no idea how to answer. So he poses a question in his preface: "Why are there no simple languages with simple forms of nouns, verbs, and sentences?"

This leads to another question: why has language only evolved once? If animal communication is a primitive or less-developed form of human language, the only difference being capacity, it would seem preposterous that over 650 million years no other creature similarly developed the use of words, nouns, verbs, and sentences, even in simple form, and that no current creatures have anything like sentences. (I'm not sure that this is altogether true, this last bit, but my guess is that it's true enough.) Human language is an anomaly that has to be explained. Deacon uses the analogy of fur and porcupine quills; it would be ludicrous to say that all fur was a primitive form of porcupine quills.

And here he puts forth a hypothesis that I wasn't expecting. He suggests that to acquire human language would conflict with some normally advantageous feature of animal communication; that to develop or (if you're a pet, say) learn a language would require something analogous to the unlearning of useful communication skills and their replacement by others. So it would be rare for evolution to select for the loss of adaptive communication skills, even though these skills interfere with the development of something potentially more adaptive: language.

So that's where Deacon starts. He goes on to try to explain what we've got that animals don't - basically, our way of relating words to other words - and what interferes with this, which is the ability to tie specific signs to specific other phenomena, an ability that animals do have. This section of the text is murky, but if I understand Deacon right, the "unlearning" involves a shift in attention from what we've actually experienced as associating with something else and replacing it with knowledge of what can associate with something else, whether we've seen it or not.

Deacon doesn't like the concept of Universal Grammar, that our grammatic ability is built into our brains, because he doesn't think it explains anything. He says that it tells us that we have the skill, but it doesn't explain how we got it. He tries to develop a more useful explanation. I don't have time to post about it now, but it involves both something like "unlearning" (more accurately, "not learning too soon") and the idea that languages evolve much faster than human brains do, and that the basic features of language must evolve so as to be learnable by toddlers, so in effect toddler brains are the instrument of natural selection for the basics of human language.

I'm about a third of the way through.

Comments

[martinskidmore.livejournal.com]

I don't like the universal grammar thing either - I talked a bit about this on a Slugs show on the radio a while back. What you go on to mention seems more likely, that brain evolution and language push and pull at each other. I could talk at some length about how the brain develops both in physical and mental terms, which I think fits far better with a Sapir-Whorf view than a Chomsky one.

The difference between humans and animals may not be clear cut. There were some interesting studies of primates who were taught a small vocabulary in sign language and who used abstract words in new combinations productively - I think the example I recall was one that learned 'dog' and 'cat' and after being taught 'young dog' for puppy, used 'young cat' for kitten without prompting. This seemed like the first evidence of combining words and using abstract concepts. That suggests that there is no 1-0 type difference.

Revisiting what I said I'd resist, when brains develop they develop lots of connections. In the early years, connections keep being added - but connections are deleted in large numbers too, as they are found to be not valuable. The huge quantitative difference in linguistic ability between humans and primates may have something to do with that - raising some apes with humans and humans with apes for many years might produce rather smaller differences. The physical evolution of the ability to produce such a wide range of sounds is part of this too, of course, and that is surely another push-pull case.

I am unclear how language takes away other communication skills, apart from clearly shifting the focus away from them. It doesn't stop you howling or pointing or waving arms, but maybe it makes you less quick to use such prominent approaches?

[koganbot]

I remember reading recently how researchers in the wild found some monkeys who were combining a couple of signs (say for "predator that comes from sky" and "hurry up") into something that meant some third thing, but unfortunately I've otherwise forgotten the story. It was in the NY Times.

There are definite examples that Deacon brings up of chimpanzees being successfully taught by humans some rudimentary syntax, such as "bring banana here" and "put water dish there," and learning that "water dish" can substitute for "banana" so as to produce "bring water dish here," and learning not to waste their time on "water dish put banana bring." But the teaching process is arduous and not all chimpanzees were able to get it.

I am unclear how language takes away other communication skills, apart from clearly shifting the focus away from them. It doesn't stop you howling or pointing or waving arms, but maybe it makes you less quick to use such prominent approaches?

Deacon's not as clear as he should be either, and maybe he's still working out his ideas, or more likely he'll add more later in the book. He's obviously not saying that we humans, because we have language, grammar, etc. and can relate words to other words in order to form sentences - he's not saying that because of this we fail to be able to make one-to-one connections between a sign and what it stands for in the way that animals can. If I understand him right it's a forest-tree type thing: if you think of the relationship between a word and a particular object or event as a "tree," but the grammatical relationships between words and other words as a "forest," then if you learn a lot of trees first, you're never going to learn the overall form of the forest unless you develop a way of forgetting or not getting distracted by the individual trees. Animals are very good at learning trees, or the equivalent of trees, in the wild (an animal doesn't have a word to designate "snake" but can know to be afraid of anything that looks like a snake, and that a rustle in the grass can be a sign of a snake, etc., so in this sense animals have abstractive ability). But there would seem to be an immediate adaptive disadvantage to dropping the tree ability - the ability to make such associations.

In effect human infants have a way of learning forests before they learn trees, and can then add tree knowledge later.*

Deacon doesn't think that to posit an innate "Universal Grammar" helps to explain how infants do this - "Universal Grammar" says that infants don't learn the form of a forest, but rather that they're born knowing it. And Deacon doesn't buy this; he thinks he has a better explanation, which I'm just getting to in my reading. He's saying that human infants have particular cognitive biases that make them more likely to notice a forest than trees, and that languages adapt themselves to these biases (that is, broad features of languages that can't be learned by infants don't become part of languages, because they wouldn't survive from one generation to the next; this is assuming that broad features aren't learned later in life, which is probably a good assumption). Deacon is aware that the word "bias" doesn't explain any better than the phrase "Universal Grammar" does, so what's to come in the book will be to be make precise what the biases are and how humans got them. Deacon thinks it'll be easier to explain how humans have a disposition and can adapt their language to the disposition than to explain how humans somehow got complex grammatical abilities hard-wired into their genes and brains.

*There's obviously more that needs to be said here. A chimp experiment that Deacon describes shows the interference going both ways; that is, chimps that had arduously been taught some symbol & sentence type skills were then able to deal with a very different symbol & sentence type situation, whereas an equally intelligent chimp with no previous symbol & sentence type experience not only couldn't use the new information, she actually got all confused and lost some sign-object associations she'd previously mastered. I don't have time now to type up Deacon's account of this, unfortunately.

[martinskidmore.livejournal.com]

The totally simple answer to that development is that if language is an evolutionary advantage, in being able to warn about more complex dangers than just 'snake!', in being able to cooperate and plan better, for instance, then natural selection favours those with brains that can handle it. From what I know about baby and infant brain development (quite a bit!), there are huge numbers of connections made, then they are whittled down or reinforced according to their value. A very simple linguistic example is that Japanese babies don't need the ability to distinguish the sound R from L, so lose it - nothing genetic there, entirely environmental. That particular example has been very well researched, but is too basic to help us much, though I would say it could be that there may be abilities potentially in the baby brain to handle a whole range of different types of grammar, and in the same way some are discarded and some strengthened as they proof useful or irrelevant. We don't by any means need to assume a natural ability in some platonic grammar. (This is an argument for the Sapir-Whorf hypothesis, that language shapes the way we can think.)

Nonetheless, assuming greater inborn talent for some structures than others doesn't seem implausible either, though I have no idea how we could study or prove this. Natural selection and the tendency of incremental development of language can reinforce each other as time goes on.

I am not expert enough on different languages to know how wide the diversity of grammars might be, but I have read bits that suggest some work in ways that I can't get my head around at all - this may suggest that there is no basic pattern that the brain naturally prefers, but I am sure there is a vastly greater range of possible grammars than I could ever remotely imagine. We need to meet a range of aliens, see how theirs work.

[koganbot]

Well, if language is simply an evolutionary advantage, then it should have evolved time and time again in the 650 million years since multi-cellular creatures appeared. So Deacon is asking what blocked (and blocks) other animals from evolving it, and what is the human difference that didn't/doesn't block it for us. He argues that "brain size" isn't an explanation either, since if the brain evolved in concert with language (the subtitle of the book is The Co-evolution of Language and the Brain) we'd expect that other animals could have evolved larger brains once they got language, for the exact same reason we did. (There are structural reasons why we can support large brains, but those structures evolved in us, and I'd expect they could have evolved in other creatures too.)

[martinskidmore.livejournal.com]

It's very hard to come up with an explanation for language that doesn't involve it being an evolutionary advantage - and it's easy to see lots of ways in which it is, so I am not really inclined to doubt that. We could say the same about making tools - a few animals use tools in some simple ways, but none make them. There are quite a few things that have only evolved once, so the fact that it hasn't happened more than once is no evidence for it not being advantageous. I don't find it hard to believe that a combination of a particular environment, vocal apparatus that had some flexibility, a certain kind of social structure, brains big and complex enough and maybe other factors (a wide range of dangers from many directions, say) combined to make the start towards advanced language a useful and possible thing. I don't think we have any precise knowledge of when it evolved (hard to see how that could be determined), so the above is admittedly vague speculation.

If we assume that some such combination of factors might be needed to make it both possible and useful, and believe that they are rare, which I suspect they are, we don't have to explain why it didn't happen with, say, other primates. Evolution has never worked by finding one perfect solution, it works by sort of random (well, within complexity and chaos theory!) chance and some things working - the advantage may have been small at first, or it may be that other creatures managed the same sort of communication and cooperation in other ways, so didn't need it. Those ways may have been more limited, but enough to mean our kind of language in its early stages couldn't provide them with significant extra advantages, so wouldn't take hold.

I am strongly suspecting some teleological tendencies in the arguments you are quoting. Maybe not, but it reminds me of a lecture I went to some weeks ago about what alien life would be like, which succumbed to the usual tendency to think that what we have as evidence (life on Earth) is pretty much the way it inevitably has to be, so alien life will therefore be pretty similar to ours. Anything leaning towards universal grammar makes me think of that weak argument.

[martinskidmore.livejournal.com]

Actually I am being a little unfair to the lecturer, in that he did try to justify his reasoning, by thinking through what seemed necessary and what didn't, but I thought he did a poor job of that. For instance, claiming a support structure something like a skeleton was necessary was unconvincing given the huge and very long success of cephalopods, for instance. Still, sensory apparatus and ways to process that, preferably near the apparatus to process with maximum speed, is a sensible basic requirement, so his arguments were far from worthless.

[koganbot]

Deacon is being the opposite of teleological. He's not arguing that language must have eventually developed or that events happened in the past so that it would develop, or that it was somehow waiting to happen. He's arguing, as you are, that the circumstances are obviously rare, that language is an anomaly, not a pre-ordained outcome. But he's giving a real interesting explanation as to why it's rare, one I'd never heard before.

I'm not sure where your line of thought is different from Deacon's, actually, except that your simple statement "language is an evolutionary advantage" doesn't explain why language is rare. It seems to me that we have at least two alternatives here: (1) proto-language, proto-grammar, is only a selective advantage in rare circumstances; and (2) proto-language is only possible in rare circumstances, and once those circumstances arise it's got an obvious selective advantage. Of course, there can be ways of mixing the two - e.g., language can be possible only in fairly rare circumstances and can be a selective advantage in only fairly rare circumstances, and the two obviously have to occur together. And I'll see where Deacon ends up, because he's trying to put together the outline of a story of how language and the brain evolved together.

[jauntyalan.livejournal.com]

-- Well, if language is simply an evolutionary advantage, then it should have evolved time and time again

consider the prerequisites for language to develop to be achievable, then your 'why has language only evolved once' could also becomes why is there only one species in the genus 'homo', and so on.

[jauntyalan.livejournal.com]

what's fun about this is the proliferation of theories because there isn't a fossil record to check with. it's a lot of clever ideas covering various fabulous grounds, some productive and suggestive, but with no decisive winner.

not only is there no record, but we are so far ahead in linguistic complexity to the 2nd-place species (whatever that might be considered to be) that it's everyone's guess what of the many differences contributed.

[koganbot]

...so what's to come in the book will be to be make precise what the biases are and how humans got them

That sentence seemed to interfere with itself as I was composing it. This is what I meant:

...so what's to come in the book should make precise what the biases are and how humans got them.

[dubdobdee.livejournal.com]

i'm drunk so i may not be in the best frame of mind to address this, but surely usefulness and rarity are simply not related, in evolutionary terms, positively or negatively

so for example: if language is a by-product of another evolutionary development -- less obvious but necessary for language, but this other development is in itself a product of quite rare conditions, then language never gets to broad chance to evolve, because the pre-condition is rare

usefulness is discovered after the fact: the anti-useful traits are bred against -- if a specific local usefulness arrives (agile ape tongue for sucking honey out of honeycombs in old tanzania), with a by-product trait that turns out to be useful generally ((eloquence of language), there's no reason for the by-product to turn up by any other route.. the lack of an equivalent for beast B old tanzanian honey in greece or timbuktu is what's blocking the evolutionary jump... language would be useful to them too, but the route to it has no reason to open

[koganbot]

if language is a by-product of another evolutionary development -- less obvious but necessary for language, but this other development is in itself a product of quite rare conditions, then language never gets to broad chance to evolve, because the pre-condition is rare

Well, yes, obviously language is rare, so the conditions that allowed it to develop must be rare; but since vocalization, social behavior, and communication are not rare, what might these other preconditions be? Deacon has an interesting idea about what these preconditions might be. Or another way of putting it might be, what preconditions did all other animals lack.

Obviously, a nonpoisonous butterfly might end up looking like a poisonous butterfly, whereas, even if it would be adaptive for us to look like poisonous bipeds, there aren't any available poisonous bipeds for us to evolve into resembling. So that's one adaptation that's not open to us. But it's not obvious that being able to use symbols and grammar should be all that specific and difficult. But since apparently it is, that rarity is one of the things that Deacon is trying to explain. And he's trying to explain it in a way that doesn't just go, "At some point we hardwired in grammar."

[jauntyalan.livejournal.com]

finally spotted that this book specifically is on old reading lists of mine from when i was going through the 'canon' of Cognitive Linguistics stuff. the general CL thesis that language is closely integrated with other mental faculties is largely inspired by evolutionary considerations, which is what yr man Deacon is all about, so it figures. not sure why i never got round to reading this one. i may well do so now :-)

[koganbot]

I'm baffled by the responses here. Is there something about your posts that I'm not getting? Something about my post that you're not getting? Are you pushing the question away, or just warming up?

I already know that language is rare. What is not obvious is why it's rare. To put it another way, it is not obvious why other creatures that are social, can vocalize, and can communicate don't use sentences. They can generalize - food versus nonfood, potential predator versus not a potential predator - and they can learn new associations: ringing bell means someone's going to bring me meat, not only does smoke mean fire, "Fire!" means fire, etc.

Asking why language is rare doesn't seem to me equivalent to asking "Why doesn't anything other than a kangaroo look just like a kangaroo?"

It'd be more like asking, "Why, since so many species developed gliding capacity, and others use appendages to extend their jumps, only one developed flying capacity?" Except that lots developed flying capacity.

I have no horse in this race as to whether Deacon's or someone's competing theory is right; but he is saying something that I hadn't heard elsewhere, which I wrote in my post (and then elaborated on in the comments): He suggests that to acquire human language would conflict with some normally advantageous feature of animal communication; that to develop or (if you're a pet, say) learn a language would require something analogous to the unlearning of useful communication skills and their replacement by others. So it would be rare for evolution to select for the loss of adaptive communication skills, even though these skills interfere with the development of something potentially more adaptive: language.

And where he's going with his argument seems to be, though I haven't gotten there yet, that the ability to tie specific signs to specific other phenomena, if it comes too early in an infant's development, would interfere with learning to relate symbols to other symbols. So that humans are born relatively immature would be one thing that makes language learning possible. (Obviously this isn't the whole story.)

If Deacon is right, he's explained a lot rather simply and elegantly, without having to go "Here are the 467 factors that came together for human beings"; and you don't have to go from species to species to wonder why for each one it didn't come up with 368, or 914, or 311 factors that would have worked.

[koganbot]

(Well, Alan seems to be looking to explore further. Not sure that I understand Martin's posts, or Mark's.)

[jauntyalan.livejournal.com]

a normal evolutionary view of the development of mental faculties would say that to acquire language would involve the mustering of (neural) resources that might have been used otherwise, or not developed and supported (nutritionally).

it's the convoluted path (simple to follow, but at some point the unobvious choice leads to the jackpot) vs the complex path (an obvious choice with a lot of ground to cover and problems to solve.

so it's an interesting alternative, but the complexity threshold or 'only one tallest tree' argument still seems valid. and cos hom sap is singular in so very many features, it's 'easy' to suggest than any of these, or combiantion thereof is also necessary.

again, so many lovely theories. what i like are all the confliciting theories about what was the primary 'driver'. My favourites tend to be about forming larger social groups for some reason. the resource-based rewards (nat selection gold!) for organised behaviours leading to larger groups, leading to strategies to cope with them, that lead to the mental faculties that form the basis of language. language as grooming/social-control.

<arthur c clarke>perhaps i like these because it suggests that in the super-connected age of t'internet developing new strategies for such social intereactions will lead us somewhere else again.</arthur c clarke>

[jauntyalan.livejournal.com]

My favourites, for some reason, tend to be about forming larger social groups for some reason.

[koganbot]

Your sentence works with both reasons: for some reason forming larger groups tends to be your favorite, and for some reason proto-humans formed larger groups.

Since Deacon is practically the only one I've read on the subject I can't really say he's doing something that no one else is doing, but by asking the question in the negative - why didn't other species develop language? - he's finding a different way into the question. That is, he's asking, "What inhibited other creatures from developing nouns and verbs and sentences and the like?" So he's not looking so much for a driver but an inhibitor. And this means he doesn't have to face questions like, "Why isn't forming large social groups a driver for a whole bunch of other species as well?" Of course, Deacon would have to answer the question, "Why is dropping the inhibitor so rare that only humans dropped it?" But there's at least a potentially logical answer to that one: "Because the inhibitor is really really really adaptive." So dropping the inhibitor is rare, but once it is dropped a species can then be open to various drivers.

I think one reason Deacon likes his own argument is that to some extent he's using it to kill two birds with one stone. A second question goes, how does a helpless, incompetent creature with a minuscule attention span and little impulse control - a human infant - master something so complex as a language at such a young age? I haven't really gotten to Deacon's answer, so I don't want to misrepresent the guy, but he seems as if he's going to intertwine his answer to this question and his answer to what opened up the species to evolve language.