Cartesian Linguistics
Page 4
II.2 The Chomsky difference: naturalizing language and mind
While his basic assumptions about the mind and the strategy to use to study it are much the same as those of his RR predecessors, and while his predecessors – many of them the scientists of their days – would very likely have welcomed his efforts and their result, Chomsky, unlike the others, has tried throughout his career to turn the scientific study of language in the mind (and of mind in general) into a natural science, ultimately a branch of biology and other natural sciences. (After his earliest efforts, of course, he had the help of many others working within his assumptions about how to proceed.) This project is implicit in RR assumptions about the mind. If much of the mental machinery needed to develop concepts and their combinatory principles is innate and one is going to try to explain how it comes to be in the mind at birth, it won’t do to say that God put it there (Descartes) or to construct myths of reincarnation (Plato). The only course open to us is to look to biology and those other natural sciences that can say what an infant human begins with at birth and how what s/he is born with develops. And taking that tack also makes it possible to at least begin to speak to the question of how human beings came to have apparently unique machinery in the first place – to address the issue of evolution.
Placing the study of concepts and their combinatory principles in natural science also imposes specific and well-known methodological constraints on the task of the scientist of the mind. Adopting nativism amounts to beginning a research effort that assumes that concepts and language are somehow implicit is some kind of natural ‘mechanism’ of the human body-mind, under (partial) control of the genome and the course of development it controls. Constructing a theory of concepts and language is, then, constructing a theory of the relevant mechanisms, and of how they develop. Trying to do that places one’s efforts firmly within the natural sciences, and demands that one meet the criteria of successful science that other natural sciences aim to meet. Of course, one’s theories are theories of an internal system and how it develops; but ‘going inside’ makes no difference. One’s theory must meet the standards of success that any natural science must meet, regardless of subject matter: the linguist or lexical concept-theorist must construct theories that satisfy the same demands that the theories of the physicist aim to meet. Would-be internalist mental scientists must produce theories that are descriptively adequate in that they fully and accurately describe all the elements and properties of the ‘things’ the theory deals with – in the case of language, words, sentences, and how words come to be put together to make sentences. Further, a theory must be explanatorily adequate; in the case of language, it must say why a child has just these elements in his or her language at a specific stage of development, and – once one has explained this – one must deal with further explanatory issues, such as how language came to be introduced in the species. Further, a theory must provide an explicit, formal specification of everything that the theory aims to describe and explain – it must be formalized with, if relevant, quantification according to an appropriate measurement scheme of the elements and their ‘powers’ or features. In a connected vein, a naturalized form of mental science must aim towards simplicity: the theory must offer a compact yet full account of its subject matter. With language, for example, this can amount to constructing theories with as few principles or rules of combination of words as possible, and stating these rules as economically as possible. Further, a theory must aim towards objectivity. As the history of science indicates, that requires abandoning the anthropocentrically oriented concepts of common sense (the ones that we are born with, including the common-sense concept LANGUAGE) and inventing concepts that can adequately describe and explain – for language – any language whatsoever, plus say how and why a particular language developed in a specific individual. And a theory must aim towards accommodating the mental science in question to another science – in the case of language, certainly biology – among other natural sciences. Finally, a condition that reflects everything said so far. The internalist theory must make progress – successive or at least temporally nearby theories of the relevant domain (theories of language, vision, facial configuration. . .) that show improvement in one or more of the dimensions just mentioned. Progress is measured by the standards of the methodology itself.
Chomsky’s efforts at constructing sciences of language have improved in all these dimensions – that is, his theories have made considerable progress. I point to some of the signs of progress later. That there has been progress suggests that his efforts and those of the many who, like him, adopt an RR view of the mind, are on the right track and somehow seem to cut nature – in his case, the language organ – at the right joints. The progress in turn makes it tempting to hold that the target of these theories – the language organ – is ‘real’. And it makes it tempting to adopt a policy of saying that what the best theory (by the relevant standards) says about the organ is true, and that it does describe and explain ‘how things are’. The temptation should not be resisted. We rely on both science and common sense to get an understanding of ourselves and of the world. Each approach has its merits. Common-sense concepts notoriously fail in science, while scientific concepts such as MU-MESON are hopeless in solving the kinds of practical problems that common sense deals with. Each kind of capacity to solve problems has its place, and neither deals with all problems. Where the aim is objectivity and the precision of explicit and formal statement, however, there is no choice: the methods of the natural sciences yield the only answers we are likely to be able to get. And if at a time a theory is the best available, it tells us what language – or vision, etc. – is.
Certainly Chomsky’s efforts have progressed well beyond those of his RR predecessors. Where his predecessors countenanced unexplained powers and had to be satisfied with pointing in a direction, or where they had little idea of how to make sense of innateness, Chomsky cannot and does not. For example, Descartes and the Port-Royal grammarians in the sixteenth and seventeenth centuries in the final analysis attributed the articulation and productivity of language not to an organic internal system, but to an unexplained power called “reason” or “thought.” But thought, for example, is best understood as constituted, in part, by language: we use language to think. So we should try to say what language is without appealing to thought. Chomsky clearly distinguishes language and its study from the ways in which language can be used, including thought and reasoning. To do that, he constructs a theory of a biophysically based organic system. And to make sense of innateness, his predecessors might have pointed to divinity (it’s a gift from God) or they might hold that it has something to do with Nature, but say little about how God managed to get concepts and combinatory principles into our heads. His RR predecessors did, though, make at least some progress. In contrast, empiricists seem to have added little to Locke. And, like Locke’s efforts, theirs generally fail to meet the conditions on adequacy of a naturalistic theory.
Many empiricists would not be bothered to be told that their efforts are far removed from naturalistic theory-construction. That is because many current empiricists, at least, think of language as a complex form of social practice learned from others, with practices varying from language to language and environment to environment, and of concepts learned as a part of acquiring linguistic and other practices. They might, as Wittgenstein did, think of concepts as epistemically governed ‘roles’ in linguistic practices or ‘language games’. For them, languages are social institutions and artifacts, not states of a biologically based mental ‘organ’.15 Still, they want to call themselves scientists. To make this seem credible, they might argue that they are offering a far simpler ‘hypothesis’ than Chomsky’s (once) rather daunting-looking theories of Universal Grammar (UG). Their hypothesis is that language and concepts are learned by some kind of generalized learning procedure. They take their generalized learning procedure to be the simplest hypothesis concerning concepts and combinatory principle
s because it is the least committed to the existence of any kind of domain-specific mental machinery – anything devoted to language, for example, for that would make language innate. So if they want to speak to what is in the head at all, they stipulate that what is in the head must be some kind of neural net that has the property that it can be modified by ‘experience’ and correction of its outputs. Perhaps there are innate organic mental modules for vision, audition, and the like. And no doubt these have some quite remarkable innate properties. But there must be a lot of the mind/brain that is plastic and can be modified by experience and training procedures, for that is where language must go. To show the merits of their ‘hypothesis’ (which is rarely, if ever, explicitly stated), they might introduce computer models in the form of ‘neural nets’ of what they believe are the plastic regions of the mind/brain and subject these computer models to input that is supposed to simulate their views of human linguistic experience and the data of language learning. They consider their efforts successful if the computer model ‘learns’ to perform the ‘task’ set it to the satisfaction of the ‘experimenter’.
This story makes language acquisition a miracle: standard naturalistic explanations routinely employed elsewhere are simply rejected, and for human language alone. The empiricist is happy enough to say that a child’s pet songbird acquires the song it does for naturalistic reasons; input may be required for development to proceed, but development, and the song patterns that develop, are largely under genetic control. But human language? It must be explained some other way. Refusing to see languages and concepts as natural objects for which one needs naturalistic theories is a form of what Chomsky calls “methodological dualism”: when it comes to crucial features of the mind, the empiricists abandon not just internalism and nativism, but the methods of the natural sciences. They lodge languages outside the head and might (e.g. Prinz 2002) treat the everyday concepts we have, such as CURRY, as built out of primitive mental items like sensory features, but built under exogenous control, not the control of internal developmental machinery that demands many non-sensory features (such as ABSTRACT). Or they might (Sellars and many others) speak instead of the ‘token’ “curry” in terms of learned games and social practices, construing its meaning in terms of its contributions to socially sanctioned truth or correctness conditions for the ‘token’ “curry.” However glossed, the character and certainly the (supposed) referential powers of the concept CURRY depend on how a community of speakers use the word curry – the contexts, what get counted as the right or correct or true or appropriate uses, and so on. The only internal and presumably ‘natural’ machinery relied upon are some sensory capacities and some form of generalized learning procedure, cashed out perhaps in a version of associationism and/or behaviorism embedded in a plastic neural net. The procedure might consist in some kind of statistical sampling procedure lodged in what starts out as an undifferentiated neural net, a modern empiricist’s version of Locke’s blank slate.
Not all empiricists are methodological dualists. But that is not because they think that languages and concepts are somehow natural objects. Some, such as Herder and Foucault (who count as empiricists by virtue of being anti-nativist and externalist16) were unable to understand what a naturalistic scientific methodology is, or appreciate the very different objective view of the world it yields when compared to the anthropocentric view got from common sense.17 Only those empiricists with an appreciation of the differences between science and common sense (“folk physics,” “folk psychology,” and the like) are apt to adopt methodological dualism. The twentieth-century American philosopher Sellars and his intellectual progeny (e.g. Churchland, Brandom), and the philosophers Quine and Putnam and theirs, might have anything from a reasonably clear to a very good (e.g. Putnam) idea of what naturalistic scientific methodology is,18 but they clearly refuse to hold that language and concepts could be investigated using this methodology, rather than some version of an empiricist one. At this stage in the development of the science of mind, their refusal is to the RR strategist puzzling, at best. With concepts, perhaps there is some room for sympathy with the empiricist strategist. While there is no doubt that the poverty and creativity observations make a RR strategy reasonable, naturalistic theories of concepts are still in their infancy. Among other things, even those sympathetic to the strategy, such as Fodor (1982, 1998), are drawn to views (such as that concepts by themselves denote) that – as noted before – make a straightforward internalist and nativist approach impossible. But, RR strategists such as Chomsky must feel, surely it is unreasonable to adopt a non-naturalistic methodology for the study of language. Here we find articulated theories, and very considerable progress on all fronts. Quine’s view that in the study of language, “behaviorism is mandatory” appears to be nothing but dogmatic refusal to face the obvious facts.
But is not what empiricists propose naturalistic inquiry after all, especially in the form that some of the connectionists have devised, where we find ingenious efforts to get “simple recurrent networks”19 and the like to (say) “recognize” classes of things, or after massive training produce outputs that seem to indicate that the network’s connection weights have somehow embodied a ‘rule’ – that is, yielded a pairing of inputs to outputs that suit the experimenter’s criteria for correct behavior? For Chomsky the answer is flatly “No.”20 However ingenious the techniques used, and no matter how much technology is thrown at prosecuting them, empiricism is wedded to a picture of the mind and of how it gains and uses language and ‘content’ that does not and, unless modified so that it becomes largely indistinguishable from a rationalist approach, cannot address readily observable facts about human language acquisition and use. Even the few celebrated ‘successes’ turn out to be failures when what the connectionist offers is required to speak to questions that must be taken seriously and addressed. Quoting from one of Chomsky’s comments on the manuscript for this introduction:
. . . No matter how much computer power and statistics . . . [connectionists] throw at the task [of language acquisition], it always comes out . . . wrong. Take Elman’s famous paper – the most quoted in [cognitive science,] I’ve been told – on learning nested dependencies. Two problems: (1) the method works just as well on crossing dependencies, so doesn’t bear on why language near universally has nested but not crossing dependencies. (2) His program works up to depth two, but fails totally on depth three. So it’s about as interesting as a theory of arithmetical knowledge that handles the ability to add 2+2 but has to be completely revised for 2+3 (and so on indefinitely). Such approaches could do far better trying to duplicate bee communication or for that matter what’s happening outside the window (where they would do vastly better than physicists). Why don’t they do it? Because it would be ludicrous: no scientist is interested in some way of matching data. [Clearly, insisting on this with language is] . . . just more methodological dualism.
It is an irrational insistence on taking human language and concepts outside of the domain of naturalistic research.
Perhaps there is reason to take empiricism seriously with some aspects of the study of cognition – perhaps with analogical reasoning, perhaps even with some aspects of the acquisition/learning of scientific theories and their concepts. But with common-sense concepts, and especially language, there is no reason to take empiricist speculations at all seriously. No one finds children subjected to the training procedures for concepts or language explored by connectionists, for example. That is only one of many reasons why Chomsky believes – not unreasonably, given the facts and the progress made with language – that dogma, not reason, drives empiricist research strategy. The apparent aim is not to explain the facts of human language and concepts and their growth, but to show that what are claimed to be plausible models of neural networks can be made to simulate (to an extent) some human cognitive behavior or another, to the satisfaction of the experimenter’s view of the job that the behavior is supposed to do. There might be some minor success in
that effort: acquisition of irregular verbs is claimed to be a success. But this and even weaker candidates have nothing to do with the conditions under which children acquire concepts or language, nor with what they have in their minds when they acquire them – tools, apparently, that can do all sorts of jobs, not just display a favored kind of behavior to the satisfaction of the experimenter. Getting a ‘neural net’21 that meets empiricist conditions after massive training to display some behaviors (classifying, inferring. . .) in performing some task or another to some degree of satisfaction might be useful for some purposes. It might offer a clumsy and time-consuming and probably unreliable way to get some machine to ‘learn’ how to do something without just programming it to do what you want it to (assuming you know how to ritualize that). It might also charm those impressed by what one can accomplish with minimal tools – and lots of time and sufficient funds. But unless one can explain what anyone can readily observe in children – what the poverty and creativity observations point out – empiricist efforts like these make no contribution to sciences of the mind.
This view of Chomsky’s should not, incidentally, be understood as a blanket condemnation of computer modeling of various aspects of concepts and language, and perhaps of aspects of their use. Some of that can be very useful – among other things, one finds interesting work being done on the lexicon and lexical features that is useful, perhaps even important, to an RR strategist. The objection is to insisting that the facts must be otherwise than what they are with the strategy to use in the study of language and concepts.