Whorf's Language, Thought and Reality of 1956, which, on the basis of a study of the Hopi language, propounds something quite close to what I have called 'the guidance and constraint hypothesis', has proved enormously, if indirectly, influential both on philosophy, through Quine, and on anthropology, in the relativism and rationality debate, where it interacts with the complementary sources highlighted by Geoffrey Lloyd in his Demystifying Mentalities. Whorf was himself deeply interested in the work of LévyBruhl, and in fact believed that the viability of the mentalities scheme 'is only one of the great psychological world-questions that fall into the domain of linguistics and await the impersonal and positive type of answer that linguistic research can give'.

What is infuriating about reading Whorf – and this, paradoxically, may be in no small measure responsible for the breadth of his influence – is the constant difficulty of gauging the exact import of his principle of linguistic relativity. Clearly, Whorf is wedded to the idea that linguistic structure influences thought and action; but just what he intends by 'structure', the manner and extent of its influence on human perception of and reaction to reality, and the nature and strength of the evidence for relativity, all remain far from clear, for Whorf never produced a definitive statement of his position. Indeed,

there are grounds for a suspicion that at least the implications of some of his assertions might not be compatible. Thus, although he most frequently cites purported lexical data in support of his claims, he accords vocabulary less weight than 'structure and grammar', whose nature he apparently assumes to be self-evident. He claims, for example, that 'these abstractions [i.e. for which our language lacks adequate terms] are definitely given either explicitly in words – psychological or metaphysical terms – in the Hopi language, or, even more, are implicit in the very structure and grammar of that language, as well as being observable in Hopi culture and behavior'. Whorf explicitly denies that all thinking is linguistic, and once, disconcertingly, rejects even the relatively modest thesis that there is any determinate correlation between language and culture, let alone a causal one. But, on the other hand, he often permits himself (quasi-) metaphorical expressions which imply strong determination of thought by language; claims that 'this study [of language] shows that the forms of a person's thoughts are controlled by inexorable laws of pattern of which he is unconscious'; and goes so far as to attribute a partial linguistic genesis to Newtonian concepts: 'Newtonian space, time, and matter are no intuitions. They are recepts from culture and language. That is where Newton got them.' If this oscillation in Whorf's commitments never ceases, it was the memorable turns of phrase tilting towards fairly hard 'guidance and constraint' which nevertheless lodged themselves in the public mind and established his fame, such as his most celebrated pronouncement, in which he embraces pure 'guidance and constraint' without the slightest inhibition: 'thinking also follows a network of tracks laid down in the given language, an organisation which may concentrate systematically upon certain phases of reality, certain aspects of intelligence, and may systematically discard others featured by other languages. The individual is utterly unaware of this organisation and is constrained completely within its unbreakable bonds.'

As so often happens in cases of interdisciplinary cross-fertilisation, Whorf is no longer a name to conjure with in linguistics proper. One reason for his eclipse was the new ascendancy of transformational grammar, whose leading exponents assumed (perhaps too quickly) that their revivified universalism exploded the pretensions of Whorfian linguistic relativity. Thus, Jerrold Katz has claimed that the 'well-known doctrine of linguistic relativity, which states that cultural differences produce incommensurate conceptual frameworks, derives neither from the discovery of exceptional facts about exotic languages by linguists like Whorf nor from important breakthroughs in the study of methodology by philosophers like Quine. Rather, the doctrine derives from the empiricism common to these linguists and philosophers.' Katz supposes that linguistic relativity jeopardises the universal intertranslatability he assumes to be immediately entailed by rationalist transformationalism; but aspects of this pioneering defence of universal grammar against the threat of the then popular relativism suggest that he is perhaps mistaken:

it is commonly held that modern linguistic and anthropological investigations have conclusively refuted the doctrines of classical universal grammar, but this claim seems to me very much exaggerated. Modern work has, indeed, shown a great diversity in the surface structures of languages. However, since the study of deep structure has not been its concern, it has not attempted to show a corresponding diversity of underlying structures, and, in fact, the evidence that has been accumulated in modern study of language does not appear to suggest anything of this sort.

Evidently there is nothing to preclude the peaceful co-existence of transformational grammar and some version of Whorfian relativity, so long as 'guidance and constraint' is confined to surface rather than deep structure. It all depends on just how incommensurate 'incommensurate conceptual frameworks' must be.