My undergraduate subject was mathematics and physics (along with several other things, relevant to my formation as a scholar, but of doubtful relevance to the present discussion), and I was largely influenced by the lectures and the writings of the late Professor Richard Courant. This is responsible, I am certain, for any sense I may possess of the fact that one ought to be able to write perspicuously about formal-relational matters with something reasonably close to plain language; and of the difference between using the formal notation of mathematics for its own sake (often perceived, rightly or not, either as a way of seeming to be too deep for the use of ordinary language, and hence too deep to explain oneself to people not privy to the notation, or as a way of avoiding the problem of how to explain one’s ideas clearly in plain language) and using it for the purpose it uniquely serves best, explicit formal reasoning and the rigourous exploration of one’s ideas and intuitions, both for their coherence, consistency and completeness, and for their further consequences.
I learnt, in particular, that, applied to empirical observations at any rate, the significance of formal reasoning was exactly as good as the intuitions one may have acquired about one’s subject, and the skill with which one could interpret the objects of one’s formalisms empirically. Of necessity, or almost so, that interpretation has got to depend upon something other than the formal methods themselves, and much of the time this involves the skillful deployment of ordinary language— but also, of course, other expressive modes of modeling. It is precisely both the charm and the universal force of numbers and other mathematical objects in science that they do not comprise their own empirical meaning. In the case of statistical ‘inference,’ the burden of this lesson can be summed up in the proposition that a measure of ‘significance’ is important in theoretical work for the way it forces upon one the question of what it can or might plausibly ‘signify.’ It is a beginning of theoretical enquiry, and not the end of such enquiry except as the last-resort, or null, hypothesis that the correlation, or frequency-probability distribution, is a rock-bottom, or axiomatic fact of chance itself, as with the standard interpretation of the Heisenberg ‘uncertainty principle.’ In other than statistical formal constructions, of course, one has the unavoidable purely formal objects, without which consistent and productive computation seems unthinkable, the arbitrary constants, including such natural proportions as ‘pi.’ It is unnecessary, however to claim for them ‘explanatory force’ in themselves, since they arisewithin an ‘explanatory’ framework of empirically interpreted argumentation and proof.
What I have just said is that my style is admittedly, perhaps unavoidably, at once technical and somewhat old-fashioned, but, I hope, less than ordinarily plagued by trade jargon and the associated tendency to be so economical, telegraphically spare and supposedly uncomplicated in one’s prose (leaving out, or ‘saving,’ a supposedly unnecessary word is accounted a cardinal virtue of such writing) that all the connecting bits are left out and all clarity is lost — the reader being required to do much of the work that one might have expected the writer to have done beforehand.
Being a combination of linguist and cognitive scientist, it is, for better or for worse, probably inescapable that my prose style is to some extent influenced by the style that Chomsky has developed, and Jackendoff, I feel, has perfected, for the special task of using ordinary language (English) to explore, and build a theoretical interpretation of, the formal properties of language and thought. On the whole, I think some analytical philosophers of logic and language, e.g., Searle and Russell, have done a better job of it though, and when I see the opportunity, I try to employ the sometimes deceptively simple style of didactic language that such philosophers employ. The problem seems to be that, while the individual small steps in a line of reasoning are readily made more explicit with the use of short, easy sentences, those same short and simple sentences exact a toll, a certain loss of the complicated connectedness amongst the steps and ideas. It is here, surely, that my old-fashioned education nudges me inexorably, and no doubt often unfortunately, in the direction of complexity rather than simplicity of style (both as to construction and as to the plain length of my sentences). The reader of this paragraph will, however, understand, I think, that at least some of the time this choice is made with full awareness of the alternative possibilities and their relative advantages. If I have a tendency to complex sentences, I hope I have a balancing tendency to use a plain and simple word in place of a longer, more obscure, more learned word. I believe seriously that jargon and long, obscure words often serve to hide unclarity in one’s ideas — they constitute what the Canadians call ‘baffle-gab.’
The final influence upon the way I write (or try to write) about the analysis of language, thought, and human behaviour more generally is about as different, in one important respect, from that of Chomsky and the technical philosophers as can be imagined. I cannot think of anything more devoid than Chomsky’s writing of any trace of humour, or just a simple enjoyment of using language, and of taking account of the quirks and oddities of human thought and action. And whilst philosophers may now and again crack a learned joke, the writing remains remote and ‘lofty . Yet, for simple, self-consciously joyful (light-hearted yet not light-headed), but deceptively often quite profound analysis human thought and behaviour, what better examples do we have in English than, let us say, the writings of the late C. Northcote Parkinson?
His original paper on ‘Parkinson’s Law,’ first published in the Economist, was eventually reproduced, along with a lot of trivial nonsense, as mere humour. Nevertheless, it was the fruit of his perfectly genuine technical research as a naval historian and a teacher of administrative economics (towards the end of the British colonial period in Malaya — in the same tradition of using of English at its best for the expression of technical ideas that started me off in another corner of the Empire in South East Asia, a generation earlier). The work seems to me actually to embody perfectly sensible, and sometimes reasonably novel, even deep, findings about human thought and behaviour and the odd consequences of the way systems of ideas drive behaviour one might otherwise expect to be governed mainly by utilitarian adaptive motives. Indeed, it carries much the same message about the developmental tendencies of bureaucratic organisations as does the infinitely more ‘learned’ (and certainly much more ponderous) work of the great German social historian and philosopher, Max Weber. It happens also to say these things at once more pointedly and with considerably more explanatory force.
You can generally make a point (one, single, concentrated point, mind you), if you really have one, in quite a short piece of writing, and comparing Parkinson (the English lightweight) with Weber (quintessential German heavyweight) makes it clear to me that you can have fun with the use of language, and with what you take notice of, while writing about your subject, and, I hope, while reading it.
This brings me to the point that, though I work in large part as an anthropologist and my degree is half in that subject, I can think of no anthropologist, indeed no social scientist of any kind, whose style of writing I would let influence mine. I am not thinking of the standard complaint against social science writing, that it is excessively jargon ridden, though I find that as off-putting as others have found it. I have in mind, rather, the fact that anthropological writing at least is almost uniformly, often self-consciously, devoid of the means inherently available in discursive language for exploring and conveying a logico-empirical argument.
Arguments typically have a rather complicated formal, or logical structure. The force of argument depends upon much more than assertions and a naïve cause-and-effect, or at any rate ‘if-then’ style of reasoning. This is so even when the assertions and ‘conclusions’ drawn from observation on the basis of simplistic assumptions are buttressed by, or arguably even induced from, the simple piling up of observations, however much those observations are tabulated and ‘processed’ with statistics used as a supposed analytical ‘discovery procedure’ (on this basic failing of the [Logical] Positivist programme for theoretical work in science, see especially Karl Popper’s The Logic of Scientific Investigation , N. Chomsky, on ‘discovery procedures,’ in his (1957) Syntactic Structures, and also, passim, in my Lectures on Anthropological Theory).
Indeed, it has been my experience (and I think the pretty uniform experience of generative/cognitive linguists for over forty years now) that you get down to really empirical work, as opposed to simply the self-proclaimed empiricism of most anthropology (at any rate most anthropology that is not self-consciously Postmodern, Deconstructionist, Interpretivist and Reflexively Critical — see my Lectures on Anthropological Theory #1) to-day, and most linguistics before Chomsky, just when a sufficiently rich formal argument forces you to examine minutiæ of fact that you had never been forced to take notice of before, even though you may have known them, observed them and even recorded them somewhere. This has also been the pervasive experience of the hard sciences, especially physics. There, the logico-empirical force of ‘prediction’ in theoretical work is not so much the often spectacular, and always highly gratifying discovery of things never before seen (and often the development of the very means of detailed and exact observation itself, for the purpose of testing theories) as the way interpreting a piece of formalism empirically forces one to take notice of factual details that have had no articulated framework previously to call attention to them.
Finally, let me explain something about the way I have put together the present series of essays, and my reasons for choosing to do it this way. It is not my intention in these notes to present any kind of textbook-complete argument for the importance of a radically intentionalist and computational basis for a proper general theory of cognition, or the important contribution that what I may call the ethnographic method (as against an experimental method grounded still in the canons of Learning Theoretic Behaviourist Psychology, whose very axioms deny the reality of cognition, or at least suppose it reducible to the behavioural ‘response’ of, for instance, possibly subliminal speech) and the kinds of facts it deals with especially well. That is, I do not find it useful at present to survey systematically the whole field, the bulk of the literature, or in particular the literature of arguments for a different view of cognitive theory. At the present stage of the game of cognitive theory, it seems to me, it is more useful, in fact arguably necessary, to try and put together a fairly systematic and, at the same time, pretty wide-ranging argument for the kind of cognitive theory I have in mind; wide-ranging because the real force of my position must depend upon my being able to show how many of the different sorts of problems in cognitive science, factual and formal, both, it is capable of dealing with perspicuously, and even, I hope, resolving. Nor do I intend (I am quite aware this is partly out of laziness) to try and summarise every reference I make to the literature on the various matters I take up in these essays. What I want is to point out to the reader (and I trust there are going to be interested readers) what I believe the most salient pieces of this literature have been for the development, (and for my intellectual formation, no doubt) especially for forcing me to turn my attention to various facts, problems and arguments; in order both to show that what I am doing indeed has proper connexions with the literature and with, in particular, a fair part of the spectrum of interesting and seriously respectable work in cognitive science and the several disciplines that bear on it, namely, cognitve psychology, linguistics, philosophical logic, cognitive anthropology, mathematics, especially discrete mathematics, and so on. Having, I trust done that sort of thing, I want the reader to go and look at the works I cite, rather than having to rely upon any, necessarily partisan and highly selective, summary I might be able to provide. If the reader does this to any reasonable extent, he or she will be bound to connect up with just about all the available and imaginable literature of, and bearing upon cognitive science; after all, each piece I cite is likely to contain a thorough list of references, each of these likewise, and so on recursively.
I have not so far taken the trouble in these essays of trying to argue for this ethnographic method very explicitly, or of making an explicit argument justifying it as serious scientific method, which I am clear in my own mind can readily be done. The interested reader can easily tell that the employment of this method is grounded in the view of scientific method set forth by Karl Popper in the book aforementioned, that is, in the view of science that is opposed to the formal claims of Positivism and its reliance upon a supposed methodology of algorithmic data-processing according to the canons of a supposed assumption-free array of object languages and metalanguages as proposed in the work of the Logical Positivists as a discovery procedure for explanatory truth. My particular views about why the latter views are additionally to be rejected are, in any case, set forth explcitly enough in five places already; my Lectures on Anthropological Theory, my paper ‘Falsification and Science Again,’ Journal of the Steward Anthropological Society 8,1: 53-66 (1976), my (1985) paper, ‘Cognition and Computation,’ in Directions in Cognitive Anthropology, edited by Janet W. D. Dougherty (Urbana: University of Illinois Press, in Janet D. Keller and F. K. Lehman (1991) ‘Complex Concepts,’ Cognitive Science 15,2: 271-292 (revised an expanded, pp. 74-92 in P. Boyer, ed., Cogitive Aspects of Religious Symbolism. Cambridge University Press), and in F. K. Lehman and G. Bennardo ‘A Computational Approach to the Cognition of Space and its Linguistic Expression,’ (2003) much expanded in the e-journal MACT (Mathematical Anthropology and Cultural Theory). . In the final analysis I trust that the present series of essays will be able to stand as a fair illustration of the ethnographic method, and of its advantages and its justification as rigourous method. If there is anything that especially distinguishes it and can be summed up quite economically, it is this:
The proper way to get at what is in people’s heads cannot and must not be limited to just eliciting responses (spoken and other behaviours). If one assumes that mental facts are real and cannot be reduced to behavioural ‘responses’ such as possibly subliminal speech, if, that is, one does not suppose that what people ‘say’ is all there is to what people have in mind, and so long as one takes the position that the relationship between conceptual thought, say in the form of rules of well-formedness, is that of Rule Government, as set forward in these essays and in my other work just cited, then what was once considered, by the Behaviourists in psychology and the rest of social science, as the vice of putting ‘leading questions’ to one’s informants must become the method of choice, though it must be self-consciously monitored and properly controlled. That is, one observes behaviour, including the things people say independently about their actions and their answers to initial enquiry (including prominently carefully crafted and statistically well analysed elicitation experiments of the standard kind), and then one formulates, as explicitly as one can, coherent hypotheses about what one has observed. One draws from the latter, as rigrously as possible, empirical predictions about what is really going on from these hypotheses, and, on the bases of this one probes the informants’ minds by carefully making it clear to them evidence that just possibly what they said may not have been exactly what they had in mind. In this sense, one talks with, reasons with one’s informants (‘with’ in the interactive sense of making them to some limited extent participants in one’s own analytical work); after all, they may know something that they did not think to tell you before (and that you yourself did not notice, or know about at all, or that they did not realise was even relevant), or had not been consciously aware of before, and it is your job as a scientific investigator to try, however you can, to get at as much of the relevant evidence as you can, understanding full well that you can hardly know in advance what and where that evidence might be, which is what scientific ingenuity in empirical work is all about. On the basis of all this, you reformulate hypotheses (surely they start out as pre theoretical hunches — see Lectures on Anthropological Theory (especially Lecture #1), and especially my aforementioned ‘Cognition and Computation,’ for the argument, against the extreme claims of the theory of the Sociology of Knowledge, that the social, cultural and political background out of which such pretheoretical hunches inevitably have to arise is hardly the exhaustive basis for a critique or a judgment of the validity of the explicit theories that build on them; that it is the very business of formal reasoning coupled with controlled and nevertheless imaginative experiment to try, as far as is possible, to ensure a plausible measure of objectivity and to make explicit and refine one’s analyses, and the assumptions underlying them, in order that one’s theories may rise above the parochial origins of the pretheoretical hunches from which they may have been developed.
