At a summer school on Historical Sociolinguistics I attended a couple of years ago, on the lovely island of Lesbos, Dennis Preston mentioned that he asks his first-year linguistics undergrads every year to ‘point at your language!’ Some of them, we were told, point vaguely in the air, some point at their tongues, some (predictably) at their genitals; others look confused. The correct response, as some of you might have guessed, is to point at your head. Dennis’s point is that language exists as a grammar, or a population of grammars, where a grammar is the information an individual has in their head for producing and interpreting utterances. For most linguists this is rather unremarkable (though many would nitpick a little—see below), but for most non-linguists, I suspect, this is rather less intuitive. That, after all, is the point of the exercise: that he doesn’t expect them to be aware of this already, but that learning to see language in this way is a part of becoming a linguist.*
I did say some linguists would nitpick, however. William Croft certainly would; to him, language is a population of utterances (where an utterance is a particular case of linguistic behaviour, in its context). Croft is a little unusual in this, however, and I suspect that most linguists would find themselves somewhere in the middle, holding that language can be looked at either as a collection of utterances, or as a collection of grammars, or perhaps both: that ‘my language’ is constituted as much by the utterances I produce as by the information stored in my head that allows me to produce them. That said, insofar as there exists a well-defined entity with some degree of constancy that I can call ‘my language’, then that entity is the grammar in my head.
But obviously there’s something missing here: what about those larger scale linguistic entities that stretch over multiple speakers? Surely ‘my language’ is English (or Welsh, or whatever language it is I happen to speak). Well yes, that could also be called ‘my language’; nor do linguists shy away from referring to these entities as languages. The issue here is really about level of analysis, and this is why Dennis Preston’s question is interesting. There seems to be a tendency among human beings (itself worthy of study) to think of a language like English as being like some Platonic entity existing beyond its speakers. It doesn’t, of course (except insofar as their utterances leave a trace in the environment). A language like English is to its speakers as a crowd is to the people in it. Now, a crowd has interesting properties that make it worthy of study in its own right; Rudi Keller calls it a ‘phenomenon of the third kind’: a large-scale phenomenon that emerges from small-scale intentional action, without actually being the goal of the intentional action. A traffic jam is another example: individuals don’t (usually) try to make a traffic jam; the jam emerges from lots of small-scale actions performed by drivers. Language, Keller argues convincingly, is similar. The small-scale linguistic behaviour of lots of speakers produces a large-scale phenomenon like English. The trick is to learn not to reify this, to not take it for more than it is. The English language is really a statistical clustering—of utterances or grammars, depending on our focus.
That is not to imply, of course, that the similarity between grammars in this cluster is primarily coincidental. How could it be, when speakers acquire their grammars by listening to each others’ utterances? What they don’t do is acquire it direct from some platonic entity existing beyond the speakers; and yet I think that this is how a lot of people, perhaps without realising it, see language: that English speakers are derived from English rather than the other way round. This is probably where certain prescriptivist attitudes come from: the idea that people who say ‘I don’t know nothing’ are abusing English, or the idea common in the Arab world that the various colloquial Arabics are debased forms of proper Arabic. In the latter case, of course, religion has played a part in strengthening the view that the language of the Qur’an is some sort of Platonic ideal form against which the speech of modern Arabs falls short. The development of writing has no doubt played the same role, to a slightly lesser extent, for a great number of languages throughout the world.