I was struck recently while reading a paper by Daniel Dor and Eva Jablonka [PDF - READ THIS IT'S AWESOME] about the origins of language, and the similarities that paper had with the thought of Wittgenstein. The line from Wittgenstein in particular that I have in mind is this one: "Essence is expressed in grammar" (PI, §371). This is, in many ways, a very strange line, given that grammar is usually taken to be just a matter of organisational scaffolding, a mere formality foisted upon the 'good stuff' of semantics and meaning. Strange also for its invocation of 'essence', a loaded philosophical term usually avoided by Wittgenstein, who tended to view such words with great suspicion. Nonetheless, Wittgenstein's own interest in linking grammar to essence was due to his overwhelming interest in kinds - as he put it a few lines down: "Grammar tells us what kind of object anything is" (§372).

In a very concrete sense, the whole of the Philosophical Investigations can be read as a meditation on the importance of kinds of words, and the need to take kinds seriously. As early on as §6 of the PI does Witty already state the importance of kinds in his critique of Augustine: "Augustine does not speak of there being any difference between kinds of word."

So what does this have to do with the origin of language? Well, for Dor and Jablonka, grammar in language also reflects the types of concepts into which we classify words. For example, we speak about things differently from the way in which we speak about events, or else we speak about events being factual or their being imagined, or whether things are countable (like bottles and people) or not countable (like beer and fog). In all these cases, the kinds of things we speak about delimit the kinds of things we can say about them: I cannot ask 'how many fog are there?' (or better: 'how are many fog'?) without committing a grammatical error (or what what otherwise be understood to be a category mistake....).

Now, the crucial thing for Dor and Jablonka is this: while it is possible in principle to parse out grammatical categories in any way we like, for the most part, language only ever reflects a small subset of all the possible grammatical categories that we could use. For example, while differences between events and things are generally marked by grammar in most - if not all - languages, the difference between friend and foe is not. Nor are there grammatical differences between say 'interesting events' and 'boring events': we can speak about these latter things in the same way, without committing a grammatical or categorical error.

Implications

What is important about this, in turn, is that it means that language is not a general purpose communication tool. Language is better at communicating some kinds of things better than others. This makes intuitive sense, even though it is often not acknowledged - it is much easier for me to show you in a diagram how to tie a knot, than to describe it step-by-step in language. Not all languages are of course structued in the same way grammatically: while "there is a core set of categories that are identifiable in all languages... the way that they are indicated grammatically varies from language to language. In addition, different languages may structurally distinguish some categories that are not distinguished in others".

One further implication of this is that it allows language to be understood as a product of evolution: the fact that language is better at communicating some things than others would generally indicate that those things would have been selected for because of the advantage they provide in whatever social-environmental setting that any one particular language took hold. In particular, language is really, really good at describing specific things that have taken place: "It is a communication system structurally designed to communicate messages which are grounded in a specific and constrained categorical scheme. This categorical scheme is centered around a specific set of events and situations (not all types of events and situations), their participants, their time and place, their properties, and some of the properties of their participants" - and this because language has gone though what the authors call "epistemic selection" - selection in which "a specific subset of all possible categorical distinctions was isolated, highlighted and marked by grammars for the purposes of linguistic communication."

Back to Wittgenstein

This, in turn, helps us shed light on another one of Witty's otherwise enigmatic declarations: that "if a lion could speak, we could not understand him" (§223). In light of the above, the idea would be that a lion's epistemic concerns would be different from that of a humans. The kind of grammatical categories 'natural' to a lion would be - or would probably be (this is Wittgenstein's not-unfair-wager) - quite different from a humans. In fact, it's possible to argue that one wouldn't even have to be a lion for this 'non-understanding' to take place: speakers of languages from different roots (say, Chinese and Greek - my background!), would recognise too the different grammatical/categorical parsings that each respective language has, and the effort it takes to sometimes communicate certain things, because of that difference in grammar and classification.

One last consideration: to the degree that human languages mostly share the same 'core' set of grammatical categorisations (with a few significant variations here and there) can be to a large extent put down to our shared physiognomy: the fact that we are (mostly) upright, forward-facing, symmetrical and motile beings. Moreover, we occupy a certain and shared scale of space and time (not shared by a mountain, say, who, if could speak, we would definitely not understand), with similar sets of 'epistemic concerns'. This is a nice way to link our embodiment with language in a way that is I think often overlooked, and which was intuited by Wittgenstein in his remark on the lion.