Noam Chomsky - The Function of Language

good morning professor Thomson I'm a biologist the nation's the question I have is a ravallo up to this one when a biologist looks at something as complex as the FLN the recursion device ostensibly is the first question to questions you ask is what is it for and what is it from that is what it evolved from and you suggested that maybe it didn't evolve or at least that's not the right way to look at it it evolved but the way it's with hoppitty molecules getting into the syllable so it's properties in other words on that

view are more contingent on its function it's it's contemporary function rather than its evolutionary origin even that's misleading because there's a factual question about what the function is under the usual assumption about language has been that its function is to facilitate communication right this I've never believed that and this point of view takes a very different approach it says the function if you want to use the word of language is to link interface conditions and when you look here it's just an empirical question and when you look closely at the structure of language I could

mention some cases I think what you find is that the computational system is optimal is apparently optimal with regard to linking the interface but it's very non optimal with regard to communication I see so but I actually have a specific question about its origin so it there is a case to be made that animal Minds evaluate contingent circumstances and control their most purposeful motion in response to those that those systems are in fact hierarchically nested in combinatorial there's actually an argument to be made that that's true that looks remarkably like the FLN that looks remarkably

like the recursion device and if that's true in turn then it means that the properties of the recursion device perhaps including some that looks suboptimal for its function are in fact like the bones in a bird's wing they are historical accidents rather than functional properties of the system could you perhaps explore that just for a moment whether you like or dislike that because the difference between hierarchical structure and recursion there's all kinds of the texts of the birds and you know the bones and I got five fingers like a randomized three you know and so

on that's hierarchical structure but it's not recursion there's no recursive process that gives you an arbitrary number of fingers each stage of which has particular significance remember this recursion is not like you know walking I mean you take one step and then in two steps and then three stripes every stage in the process has particular significance determine significance at two interfaces and that's something completely different than hierarchical structure yeah I was actually referring to things like Biederman's view of perception for example that it's that it's inherently combinatorial in hierarchically nested do you like yeah that's

fine I mean in fact maybe we could have gone into this like I mention it that the could be that the recursive systems and not only maybe they're connected with insect whatever's involved in insect navigation but they could also be involved with what a lot of perceptual psychologists call the rules of vision you know the rules that make you see some object some presentation as a three-dimensional object in motion there's a very pretty well understood by now computational procedure that determines that from very few presentations like three estes copic presentations they get you to see

a rigid object in motion okay there's some kind of principles involved in that and it wouldn't be outlandish to assume that those were all rules of seeing you know are similar in some way to whatever is going on here yeah I think this didn't come from nowhere you know the question is where it came from and in my view and exciting and not in clause achill conclusion is that it just comes from the way nature works with computational systems and therefore you going to find it all over the place yes if there's there's one more

question I'm really against as a biologist struck by why linguists are I'm not sure the word is hostile but apparently indifferent to this question of what language is for a biologists look at that and say that look at language look at its sophistication it's it's Paley's and watch on a Heath it's a complex thing that can only ultimately be produced as a result of its adaptive consequences the only apparent adaptive consequences are the ones we think that leap immediately to mind that is exchange of information cooperative change that's what leaps to mind it's just a

mistake I mean just look at the I mean what if you want to well it leaps to the mind of almost every biologist in the world about Saturday and when the motion of function is a very you know loose and vague notion we all know that so what's the function of the skeleton I mean just to keep you from falling into a puddle on the ground there is it to store calcium or it's not narrow what to ask what the function of a bird wing is is a well focused question maybe but you know if

you actually look at the history of it's not so well for the evolution of it it's quite complicated a lot of exaptation and so on and so forth but the notion of what a function is is a very vague notion however what is usually assumed plausibly is you can get some guide to what the function is by looking at the characteristic use okay let's take the characteristic use of language characteristic use of languages for the thought not for communication almost all language use close to a hundred percent is internal my understanding was that that's controversial

that is in particular that most thought is in fact not linked oh that's another question yes there could be plenty of thought that's not linguistic but I'm asking a different question what's the use of language statistically speaking it's almost all internal if you want to check that just introspect for a couple hours well it's but that's bait but that's based on that's based on introspection I can make my own observation but introspection is a kind of observation I mean if we had an outside way of looking at introspection we'd find the same but an equivalently

plausible interpretation of that is that that appears to be true subjectively because the purpose of language is to prepare our thoughts for the ultimate purpose of social communication control phase and say this if you want but that's now we're off way and I know those are equal district competing models should be set them in desk all we know is all we know is the following that the overwhelming use of language close to a hundred percent is internal and when you think about what that internal stuff is it has very little to do with communication like

you can spend half your life obsessing because somebody criticized you 10 years ago you don't expect to communicate it but that's kind of thing it was on your mind all the time oh you can even think about ostensibly you were preparing from an for an invert into meeting with that person maybe maybe it's an argument I need I you know I mean it's just not you know it's just factually on that correct it's it straight is arguable it's true that look everything's arguable but the overwhelming evidence is that internal thought is playing some function for

us yeah you know planning agonizing whatever we do it right and some tiny piece of it ends up as being communication incidentally plenty of other things are involved in communication too now it could you can always make up a story that's one of the things about evolutionary fairytale manufacturing about all of science there's notes there about stories yes they're not testable sorry so Alliance just so stories are particularly prominent in this domain but if you want to use plausible arguments but like characteristic youth tells you something about function then I think you'd have to say

that the function of language in this loose sense is a forethought some of it ends up being communication very small parts in fact even the part that is externalized is communication I mean a very odd sense I mean like if you meet somebody it's very hard for people to be next to each other and not to talk to each other yeah it's kind of you know it's like dogs looking at another's eye you just have to talk to each other so if you meet some you're standing with somebody a bus stuff it's just uncomfortable if

you're not talking to them so you talk to them about the weather or the baseball game or that's not communication that's just it's sometimes called Phatak communication it's just a way of establishing human relations social the human but that's not communication in the sense of transmitting information or anything if by communication you just mean interacting with other people yeah okay then it would turn out if you do it descriptively I think it would turn out like this overwhelmingly language is internal I mean it's literally close to a hundred percent some of it is externalized of

the part that's externalized a lot of it is just fatik communication and we want to be part of a group or we don't want to feel hostile to each other like right you don't look at it personally you don't this like gaze is always off to the side or you stay at a certain distance from somebody that's rightly a lot of it just needs to be like that and some very small part of it is communication and some independent sense of the word communication say well you can make up a story saying that it's that

very small part that drives the system but we have no evidence for it furthermore if you look at the mechanics of the system now we're getting into more technical aspects I think you can show that a lot of the mechanics of the system are badly designed for communication and well-designed for linking the interfaces actually one important asset which is if you if you if any of you are writing parsing programs you know programs where you're trying to do automated parsing there's one overwhelming problem that you always face and that is a word comes along like

a WH word you know what or something like that and you know because of the way language works if there's going to be it's going to be linked to some position in the sentence but that position has to be have not heard to get so you have the problem of taking word who and figuring out what unexpressed there's no space there's no break there's no nothing but what point in the utterance is it connected with just think how easy it would be if you repeated it so if you didn't say what did you see but

what did you see what okay then the problem is trivial however that's computationally more difficult if you look at the way our beliefs about the computational systems if they are recursive is in the way we think that would involve more computation for linking the interfaces so here's a pretty striking case and they're quite more where the system seems to be mechanically designed so that it reduces computation and increases the problem of perception and I think there are a lot of cases like this thank you thank you