Geoffrey Sampson

A non-nativist account of language universals

Much of the impact which linguistics has had on philosophy in recent years stems from Noam Chomskys arguments from linguistic universals to a nativist or rationalist account of mind. Before Chomsky, linguists tended to believe that (as Martin Joos, 1957: 96, put it) languages [can] differ from each other without limit and in unpredictable ways. Ac cording to Chomsky this is wrong: there are certain respects in which all human languages are cut to a common pattern. The most plausible explanation for these common properties, Chomsky argues, is that our species inherits complex fixed psychological equipment which deter mines the nature of the ideas or mental structures that are accessible to us. If a language fits our inborn psychological machinery, we can learn it and use it; if it does not, we cannot. Chomsky thus uses linguistic universals to attack the view (traditional, at least in the English-speaking world) of mind as relatively plastic or adaptable; Chomsky urges us instead to assimilate our account of individuals mental development to our account of individual physiological development, as a process in which heredity plays a role as large as or larger than that played by environment. Clearly (and Chomsky is well aware of this, e.g. 1972: 42-3) the force of the argument from linguistic universals to nativism depends critically on the lack of plausible alternative explanations for the linguistic uni versals. Some have taken this as a ground for dismissing Chomskys argument altogether (e.g. Putnam, 1967: ?ii, (e)). That seems unreason able, since the logic of Chomskys argument is simply that of the deductive-nomological explanations which are standard in empirical science: all scientific theories are convincing only so long as the obser ved data which they predict are not also predicted by rival and more plausible theories. It may well be true (I believe it is true) that some of the relatively peripheral universals mentioned by Chomsky, such as those having to do with phonology, can be explained away without recourse to innate psychological machinery (see e.g. Liljencrants and Lindblom, 1972; Ohala, 1974); but the extant attempts to explain away the syntactic universals which have always played the central role in Chomskys nativist arguments seem quite unsuccessful. (I survey and

Linguistic Universals as Evidence for Empiricism.

I. Noam Chomsky turned the previously rather specialized discipline of linguistics into a subject of considerable general philosophical interest by his argument that the discovery of universal properties of natural language requires us to adopt a ‘nativist’ or ‘rationalist’ view of human mind – a view according to which ‘our systems of belief are those that the mind, as a biological structure, is designed to construct’ (Chomsky, 1976: 7). (I shall use the terms ‘nativism’ and ‘rationalism’ interchangeably in this article, since any difference we make between them is not important in the context of Chomskys work. The truth is that, as with many philosophical ‘isms’, the two words do duty for a range of many more than two closely related, partly overlapping theses.) When Chomsky began publishing, a widespread attitude to human language was that expressed by Martin Joos (1957: 96): ‘languages [can] differ from each other without limit and in unpredictable ways’. Chomsky claims that this is false: to quote one of his favourite examples, it is perfectly possible to imagine a language which forms yes/no questions simply by reversing the order of the words in the corresponding statements, yet in fact no natural language has a rule remotely like this (even though this rule seems rather simpler, in an absolute sense, than many of the rules which are found in natural languages). Human languages differ in some respects, but in other respects they are all cut to a common pattern. Much of Chomskys and his followers work consists of formulating and testing increasingly refined hypotheses about the precise limits within which natural languages may vary.

An empirical hypothesis about natural semantics

Chomsky has constructed an empirical theory about syntactic universals of natural language by defining a class of ‘possible languages’ which includes all natural languages (inter alia) as members, and claiming that all natural languages fall within a specified proper subset of that class. I extend Chomskys work to produce an empirical theory about natural-language semantic universals by showing that the semantc description of a language will incorporate a logical calculus, by defining a relatively wide class of ‘possible calculi’, and by specifying a proper subset of that class which, I hypothesize, includes the calculi needed for the semantic description of any natural language. I argue that the special status, with respect to natural languages, of this particular type of logical calculus is an empirical finding which does not follow from any independently-known principles, and I conclude that the question why the laws of human thought have the structure they do is a biological rather than a logical question.

POPPERIAN LANGUAGE-ACQUISITION UNDEFEATED

After exposure to a finite (though large) body of data exemplifying the language of their elders, children regularly succeed in learning the language, in the sense that they go on in due course to produce utterances distinct from any of those that they have heard but which are nevertheless well-formed in their elders language (and they refrain from uttering ill-formed word-sequences). This achievement is formally akin to that of an adult scientist who, from a finite data-set, infers a theory which permits him to make successful predictions about cases which he has not observed. Linguists commonly say that the child infers a grammar from his data, and that the childs grammar governs his subsequent linguistic behaviour. Linguists think of a grammar as the same kind of entity as a scientific theory, except that the possessor of a grammar is not (or at least is not necessarily) consciously aware of its content. There are large conceptual problems in the notion of a mans behaviour being governed by a theory, no aspect of which may have ever entered his conscious awareness and which he may be wholly incapable of stating to an enquirer, but I believe that these problems are not important for the matters discussed below. Noam Chomsky has argued in many of his writings that the invariable success of children faced with the task of first-language-acquisition can be explained only by supposing that humans inherit a language-faculty which builds much of the structure of a grammar into the individuals mind from the beginning, so that only limited details remain to be decided from the data of experience. He supports this by arguing that all individuals formulate grammars sharing certain common structural properties, although it is unreasonable to suppose that only grammars having those properties are logically compatible with the data-sets available to individual language-learners, and grammars possessing the properties in question have no advantage in terms of general methodological criteria (such as simplicity) over potential grammars which lack them. (The claim that different individuals grammars share these properties, which have to do with the hierarchicality of syntactic structure, is in turn supported by the claim that the different languages of the world can be observed to share them, an observation which is surprising inasmuch as we have no a priori reason to expect that efficient communication-systems would need to possess these properties.)

Fallible rationalism and machine translation

Approaches to MT have been heavily influenced by changing trends in the philosophy of language and mind. Because of the artificial hiatus which followed the publication of the ALPAC Report, MT research in the 1970s and early 1980s has had to catch up with major developments that have occurred in linguistic and philosophical thinking; currently, MT seems to be uncritically loyal to a paradigm of thought about language which is rapidly losing most of its adherents in departments of linguistics and philosophy. I argue, both in theoretical terms and by reference to empirical research on a particular translation problem, that the Popperian fallible rationalist view of mental processes which is winning acceptance as a more sophisticated alternative to Chomskyan deterministic rationalism should lead MT researchers to redefine their goals and to adopt certain currently-neglected techniques in trying to achieve those goals.

Comment on d'Agostino

D’Agostino presents linguistics as a social science whose practitioners have until recently taken the Methodological Individualist position for granted; he suggests that this unanimity is for the first time coming under challenge by Methodological Collectivists, and he sets out to defeat challenges. Let me say at the outset that on the philosophical issue as between MI and MC I side with d’Agostino; by instinctive inclination as well as by intellectual conviction I prefer individualist to collectivist accounts of human affairs, and I remain wholly unconvinced by the various arguments tha purport to show that certain social facts must be described in irreducibly collectivist vocabulary. Where d’Agostino’s paper seems to me more open to objection is with respect to his portrayal of the history of linguistics vis-d-vis the MIIMC issue, and that is the question with which these comments will be concerned. Certainly d’Agostino is right to describe the contemporary, Chomskyan school of linguistics as thoroughly Individualist in approach, cf. for instance Chomsky’s throwaway characterization of his subject as ’the particular branch of cognitive psychology known as linguistics’ (Chomsky, 1968, p. 1). The Chomskyans’ insistence that their rich system of theoretical principles will ultimately be interpretable exclusively in terms of individual psychology becomes all the

Discussions Notes: A DILEMMA DEFENDED

Summarising, the clock hypothesis goes beyond the Lorentz transformations. The clock hypothesis in the form of equation (3) is Lorentz covariant. It is in agreement with experiments on accelerating radioactive particles. In the special case when u is a constant, equation (3) reduces to the normal formula for time dilation. It is only the latter case which can be treated using the Lorentz transformations without making a further assumption about the behaviour of an accelerating clock. The clock hypothesis can now be looked upon as a law of nature, similar to the law of variation of inertial mass with velocity. Both laws can be developed using the theory of special relativity as a heuristic aid. Both laws have been confirmed experimentally.