It seems that a central purpose of this book is to dispose of a popular error concerning the existence of a universal language, with the three horsemen of this particular apocalypse being Messrs. Chomsky, Pinker and Fodor.
I have learned that back in the mists of linguistic time, in the fifties of the last century, Chomsky came up with the idea that humans are born all wired up to do a universal language. All they need to do when they are born is to stick on a front end to provide local vocabulary and such like, rather like you might add a new driver to your computer if you want an odd keyboard, perhaps because you are French. This idea had plenty of traction for plenty of years.
A handy corollary of this idea is that since all languages are essentially the same, all being based on the one and only universal language, one might as well just study English and not bother with all the other variations on the theme.
A related idea was that all this hard wiring came into being rather abruptly and is quite unlike anything a mere animal might have.
I myself was signed up to this to the extent that I used to think that the way to do machine translation was to come up with a universal language to translate to and from, which was attractive from a theoretical or design point of view and which also served to down the number of permutations that one had to bother with. See reference 2.
Mr. Evans is concerned that lots of people who should know better are still, well and truly signed up.
His argument is developed using evidence from the animal kingdom and from our own evolution. Lots of animals do things which might be thought of as heading towards language. Some of them can, for example, almost point at things. Or lie. While humans have evolved a range of capabilities which enabled them to go much further, not least the complicated machinery needed in the mouth & throat for the production of articulate speech, far in advance of anything that a mere monkey can manage, even one which has been especially reared among us humans.
Young humans learn to use these capabilities, in large part by copying what is going on around them. This learning involves doing a lot of wiring up - which wiring is not hard at all.
So there is no blueprint for language lurking somewhere in the interior, just a collection of tools.
By way of an example, and borrowing from Hurford, I offer the sounds we use. We have the equipment to make sounds within a certain range. What any language needs to do is to pick out (say) a dozen vowels and a couple of dozen consonants in that range with which to build the words of a language. They all need to be sounds that we can make. They need to be sufficiently different from each other so that we can recognise them. But apart from that there is a lot of freedom. There is nothing much to say that it should be done this way rather than that way and languages do, in practice, vary a good deal in this matter. There is no master plan. Young humans learn to identify and then copy the sounds that they hear around them and are apt to have difficulty with others, certainly in later life. I know I do.
But so what? Modern humans are clearly born with various language relevant capabilities, some physical some mental. We can probably enumerate various properties that a language has to have in order to support normal human life in communicating groups, certainly if we do not insist on too much precision. It then seems clear that there are lots of solutions to the problem: that is to say instances of languages which work within these constraints. I remember a number of the order of 10,000, although a lot lies in the counting. Lapsing into the jargon of mathematics, the theory has lots of models. As we learn, as we get wiser, we might add to the list of properties that a good language would have. We might decide that English was the best language available for the conduct of modern life as we know it. But it seems unlikely that it would ever be the only such language. There would always be other ways to meet the requirement. Lapsing into the jargon of structured design, one should focus on the requirement and should certainly not build favourite solutions into that requirement.
But does arguing about whether these language relevant capabilities amount to a universal language capability add very much? Would a negative answer make the search for a mechanical universal language with hard rules (as opposed to one which took a more statistical approach) any less interesting? I shall report further in due course.
I am reminded of earlier disputes about nature and nurture. It seems clear to me that the answer here is somewhere in between and I don't get all that excited by trying to draw an exact line between them; indeed, I suspect that no such drawing is possible. It all depends on the person concerned and his or her circumstances.
Apocalyptic image being a painting by John Martin. See http://pumpkinstrokemarrow.blogspot.co.uk/search?q=Huge,+queer+and+tawdry.
Reference 2: http://psmv2.blogspot.co.uk/2014/09/och.html.
No comments:
Post a Comment