Language universals and pragmatics

by pieterseuren

A few days ago we had a well-attended public discussion in our institute about language universals. Peter Hagoort, who defends the reality of language universals, was pitched against Steve Levinson, who denies their reality. As was to be expected, no clear winner emerged and everybody went home feeling confirmed in their convictions. During the open discussion, I made one observation, of a methodological nature, saying that neither party could actually prove their point, since complete verification is out of the question. In the circumstances, the best one can do is pursue the right scientific methodology, which implies that one should search for the smallest number of the widest possible generalizations, while taking account of all relevant data, that being the essence of science. In the case at hand, this means that one should continue to do one’s best to find language universals and not reject them a priori—which is what Levinson does (see the article by Nic Evans and Steve Levinson ‘The myth of language universals: Language diversity and its importance for cognitive science’ in Behavioral and Brain Sciences 32 of 2009, pp. 429–448).

Out of this intervention arose an email exchange between Levinson and me, which led to a lucid overall statement on his part of his overall point of view. He wrote: “My point is: it’s much easier to find universals in pragmatics than in language structure, so language structure looks radically underdetermined by biological constraints.” My reply was: “Against your point: what you find in pragmatics, no matter how well formulated, does not reach the intimate features of language under close observation. The more and the better you observe, the more pragmatics recedes into the background as a possible explanans.” Here we are at the crux of the matter, something one rarely sees happen. Levinson and most other pragmaticists try to get the most mileage out of pragmatic principles of communicative interaction, based on the Gricean maxims, and claim that linguistic universals are not needed and, in fact, fail to do the job. Linguists like me see the justification for the pragmatic approach but maintain that pragmatics fades away as a possible source of univerals when you start looking at the facts of language more closely.

Those facts of language that have not been learned, either on the basis of frequency or on the basis of formal language teaching, are the most forceful in this respect. And there are a great many of them around. All it takes to spot them is a keen eye. Let me give a few examples. Levinson and other pragmaticists have tried to explain sentence-internal anaphora on the basis of what looks like a sophisticated theory of pragmatic principles. Yet the well-known and glaring fact that in a sentence like While he stood on the balcony, John’s room was being cleaned the pronoun he and the proper name John can effortlessly be interpreted as being co-referential, whereas this is not possible in He stood on the balcony while John’s room was being cleaned, is, in the pragmatic literature, systematically ignored or buried under such a mass of (quasi-)technical detail that the point itself gets lost. The point is, of course, that in non-reflexive sentence-internal anaphora the antecedent has to either precede the anaphor or, if it doesn’t, to occur in a higher clause than the anaphor. (This principle is commonly presented as “the antecedent has to either precede or C-command the anaphor”, but I find that rule too restricted, in view of sentences like the ones given above or While he stood on the balcony, it occurred to John that theoretical linguistics is in a mess, where the antecedent John does not C-command the anaphor he while the two can still be co-referential.) The reason for the pragmaticists’ reticence in this regard is clear: no pragmatic principle will be capable of accounting for this otherwise well-known fact. The explanation lies in a principle of grammatical structure. And this principle must be universal, as language learners are not taught it formally (it occurs in no foreign language textbook) nor are they in a position to abstract it from linguistic input on the basis of frequency. Moreover, speakers of a totally different language with only a little knowledge of English are immediately sensitive to this particular difference in anaphoric power.

Or take another example, of the same nature. No pragmatic theory or principle based on communicative strategies is capable of explaining why the following two sentences differ in meaning the way they do: I didn’t force myself. You did! versus I didn’t force me. You did! (both with heavy contrastive accent on I and you). The linguistic explanation is obvious. Let the first sentence be based on a semantic analysis in the terms of  ‘the x such that x forced x is not ME but YOU’ and the second on a semantic analysis like ‘the x such that x forced me is not ME but YOU’. In the former there is reflexivity in ‘x forced x’ (or: ‘x self-forced’), in the latter there is no reflexivity. The did in the first case stands for ‘you forced yourself’, whereas in the second case it stands for ‘you forced me’. Of course, this requires quite a bit of grammatical and semantic theory and analysis, but that is what we are linguists for. Pragmaticists, who see such examples coming, are therefore averse to any form of what they call ‘abstract’ grammatical theorizing, but there is no other way to account for such cases. Again, this difference is neither taught nor acquired by frequency. We have it in our minds as if it were an inbuilt linguistic a priori. If you want more examples of this nature, just look up my first blog, on Mickey Mouse linguistics, as the frequency fans are guilty of the same neglect of unwelcome essential facts.

So it’s the old story again. One starts off with a minimalist hypothesis, in this case the pragmaticists’ hypothesis, only to find that the facts force one to enrich the hypothesis, in this case in such a way that grammatical structure and inbuilt linguistic a prioris, i.e. language universals, must be taken into account. We had the same with behaviourism in psychology. One started off with the minimalist hypothesis that all behaviour is conditioned, one way or another, by external stimuli, only to find out that careful observation of behaviour shows that this hypothesis was insufficient and had to be enriched with the assumption of a mind capable of carrying out computations: the well-known paradigm shift from behaviourism to cognitive science. It seems about time we do the same with regard to the pragmaticist paradigm, which is too poor, and shift to a richer, more ‘abstract’, paradigm which takes linguistic formalisms and linguistic a prioris into account.