The Semantics of Science

Citation

Bade, D. (2006), "The Semantics of Science", Journal of Documentation, Vol. 62 No. 1, pp. 145-153. https://doi.org/10.1108/00220410610642093

The use of the literature of linguistics in information science was the topic of a dissertation by Amy Warner (Warner, 1987) in which the author described her findings that there was no significant use of linguistics literature in the literature of information science. An occasional uncritical nod to the “standard authorities” was found, but little more. While the lack of influence of linguistics may be deserved and perhaps even a blessing, nevertheless there are some linguists developing ideas which should be not only of great interest to librarians and information scientists, but which ought to provoke a serious critical examination of some of the most cherished assumptions on which our theories and practices are based.

In the literature of information science the references to linguistics which appear most frequently are to Chomsky and his followers working in what was originally (but not so originally) called The Standard Theory, a name which was modified with a following publication becoming The Extended Standard Theory and then shortly thereafter The Revised Extended Standard Theory. Numerous changes in the theory since the 1970s have had different names, leading many linguists now to write simply of The Standard Theory meaning by that “whatever Chomsky believes today”. Chomsky's genious lay not so much in his understanding of language as in calling his theory The Standard Theory, an appellation which at the same time established all competing theories as non‐standard.

In contrast to Chomsky's political writings, the Standard Theory has been described by the author of the book under review here as “a fascist concept of languages if ever there was one” (Harris, 1983). A similar judgement was put forward by George Steiner in a recent interview:

It could be that, by an irony of history, the future may view Chomsky not as the anarchist and extreme leftist radical but rather as the proponent of a vision of grand linguistic capital, of the monopoly of power of a single language. A world of Chomskian universality would be the world of an anglo‐american patois (Quoted in Duch, 2004, p. 109).

While this Anglo‐Americentric Chomskian vision of the universe fits well with linguistic attitudes commonly held by non‐linguistis (including many librarians and information scientists) such non‐standard views as those represented by Harris and Steiner are inimical to and probably for that very reason rarely cited in the literature of information science, the only exception that I know of being Julian Warner's use of Harris' work, including his review of two of Harris' earlier books in this journal (Warner, 1997).

Among linguists Roy Harris is well known but not always appreciated and almost as rarely cited in linguistics as he is in the literature of information science. He has been described as a linguistic anarchist, an accusation no one could ever lay on Chomsky qua linguist. In the words of one of the contributors to a festschrift for Harris:

The criticism of Roy Harris most frequently voiced by other linguists is that his work destroys without rebuilding. Nowhere in his writings does he articulate the kind of practical program on which the linguistics industry could continue to survive, let alone expand. As if his relentlessly trenchant judgements on other theoreticians of language from the beginning of history to the present day didn't do damage enough, he casts them in a lucid and elegant prose style that can actually be read by people outside the field. Hence he is a threat to linguistics not only from within – where he can be safely ignored – but from without, where his scepticism over the possibility of any scientific linguistics puts the general reputation of the discipline into peril.The failure to establish a practical programme is indeed a problem that Harris should confront… (Joseph, 1997, p. 9).

There are good reasons for this cool reception. To follow Harris entails accepting severe criticisms of the current profession of linguistics and all existing theories of language, an insistence upon situating any theory of language within a wider theory of communication, the admission that linguistics should be prescriptive and not merely descriptive, and even questioning the existence of “languages”. For most linguists trained under Chomsky (and his predecessors and his followers), this is simply too much to accept, even conditionally. Yet in spite of my being among that crowd (University of Illinois, MA, 1977), my experiences as a librarian prepared me to greet Harris' books and their heretical ideas with excitement and great expectation. And Harris' latest book is a book that I think many other librarians ought to read as well, at least all those librarians whose minds are engaged in thinking about information technologies, information retrieval, subject searching, natural vs. controlled subject vocabularies and multilingual or cross‐language information seeking.

With the 1980 publication of his second book The Language Makers Harris initiated what has come to be known as integrational linguistics, a theory of language which makes a radical break with all previous western theories of language. In the years which followed The language makers Harris published a number of monographs on the history of linguistics, tracing the legacy of myths about language from the Biblical Tower of Babel and Plato to Saussure, Wittgenstein, Chomsky and Pinker, demonstrating how certain ideas about language have influenced the way we think, speak and write about everything else. He also published a number of books on the signs of writing and semiology in general, argueing for a wider theory of communication as essential to an adequate understanding of the varied expressions of language. By insisting on the communicative nature of language, Harris put himself at odds with the Chomskians, for whom language is simply an innate mental device which was perhaps accidentally put into the service of communication. In the aforementioned festschrift, two of Chomsky's faithful (Borsley and Newmeyer) insist that communication is irrelevant to linguistics and The Standard Theory makes no claims nor attempts to explain linguistic communication:

Chomsky has missed few opportunities to stress that language is only incidentally an instrument of communication; its “design features” do not manifest any signs of communicative ends … The following passage, from Chomsky (1979), is typical:There is no reason to believe – to repeat myself once again – that language “essentially” serves communicative ends, or that the “essential purpose” of language is “communication”, as is often said, at least if we mean by “communication” something like transmitting information or inducing belief (p. 87).Lately, Chomsky has taken to claiming that the organization of grammar makes it ill‐designed for communication and that only a series of “computational tricks” allow it to be used to these ends at all (see for example Chomsky 1991).The goal of generative grammar is not now and never has been to explain linguistic communication (Borsley and Newmeyer, 1997, pp. 46‐47).

Perhaps in response to the complaints of Joseph and other linguists noted above, during these first years of the new millenium Harris has been taking his analysis abroad, investigating the role that language plays in the creation and maintenance of a number of cultural and academic “supercategories”: first art (The Necessity of Artspeak (Harris, 2003a)), then history (History, Science and the Limits of Language (Harris, 2003b); The Linguistics of History (Harris, 2004)), and now with the volume under review, science. In each of these books his topic is not the supercategory itself, but the ideas about language which ground and fashion the speaking and writing of the practitioners of these cultural activities: artists, art critics, and lovers of art; historians and all those who speak of the past; scientists, philosophers of science and all those working in, as he puts it, “laboratories and libraries”. This larger project of investigating language is not, of course, what Joseph and other linguists had in mind; what they really wanted was a new manner of writing grammars so that the linguistics industry could renew itself with an improved model and once again redescribe 6,000+ languages with all the opportunities for employment, publication and fame that this would provide.

The Semantics of Science is not about the language of science as that topic is ordinarily conceived; it is about the ideas and theories (lay or professional) about language that are presupposed by those people who write about science, whether scientists, philosophers, journalists or laymen. When people speak and write of the linguistic character of scientific discourse, of the language of science, what do they think that language is? What does language do and how does it do what it does? When a physicist speaks of mesons or gravity or space‐time, does she assume that there is “really” something called a meson or a “thing” called gravity or space‐time which can be discovered, described and known once and for all? May they not also end up like ether and polywater? And if people writing in other languages use different words, are they referring to the same things? Harris introduces the problem thus:

I think that for most of its history science has subscribed to an erroneous theory of language, originally propagated in antiquity, which I call the “language myth”. It still flourishes today, not only in Laboratories and Libraries. According to this myth, language “works” as follows. Words are items belonging to a conventionally agreed linguistic code, shared by all members of a linguistic community. This code allegedly functions as a system enabling one member of the community to exchange thoughts with any other member who understands the code. Thanks to this, A can know what B thinks (provided B has used the code correctly to express those thoughts) (The Semantics of Science, pp. 2‐3).

This process of thought transference differs from telepathy in that it requires the mediation of public signs (which the code provides). Harris calls this telementation. Harris claims that this theory of language and its assumptions:

… underlie the whole enterprise of Western science and scientific education, including mathematics. They provide the basis for believing that there is such a body as “the scientific community”, whose members, although divided into various subcommunities with their own technical dialects, nevertheless have access to a common language of science (The Semantics of Science, p. 3).

If these assumptions are not credible, Harris suggests, “the impressive edifice of scientific thinking is itself based on linguistic foundations of sand” (The Semantics of Science, p. 3).

Why should librarians and information scientists read such a book, especially since the author is an advocate of a non‐standard theory of language? It is precisely Harris' insistence on situating the study of language within a larger more comprehensive theory of communication that makes his entire project (and not simply this latest volume) of interest to everyone in LIS. We all know the model of communication set forth by Shannon and Weaver so many years ago: sender‐message‐receiver. This is the same model which Saussure proposed nearly 100 years ago and which remains the foundation for linguistics as well as information science. At the heart of this model two principles are established: Harris calls these telementation and a fixed code.

Communication must not be confused (although it often is) with the successful use of a shared system of signs. And one of the prime reasons for the neglect of communication in semantics is nothing other than that confusion. This point has important implications for the study of science … The traditional philosophy of language, shared by humanists and scientists alike, has at its core a “fixed‐code” semantics. Languages are regarded as providing their users with a vocabulary in which words have – or should have – fixed forms and fixed meanings. For this is the linguistic property par excellence which, according to the Western language myth, enables speaker and hearer, writer and reader, to understand each other, and allows truth to be established. In the eyes of those who accept this position, a language in which words had no fixed meanings would be as absurd as a currency in which the coins had no determinate values. Under such conditions, consistent and reliable verbal communication would be impossible. hence science would be impossible, unless it could be carry on without reliance on verbal communication at all (The Semantics of Science, p. 109).

This model Harris completely rejects, and in its place he offers the integrational model:

Integrationism is a philosophy of language which rejects fixed‐code semantics lock, stock and barrel. Instead it adopts a different approach to communication altogether. In this approach, meaning is treated as being radically indeterminate, whether expressed by words or by non‐verbal signs. But integrationism is not just a philosophy of language. The indeterminacy of meaning is, for integrationists, one of the basic features of the human condition, and is intrinsic not only to language but to the development of all human institutions, social and political … .

The integrationist alternative to fixed codes construes communication as a continuum of creative activities in which the participants strive to integrate their own actions and objectives with those of others, as best they may, in particular circumstances. The communicational continuum is open‐ended and that is why there is no determinacy of meaning. Nor is there any guarantee in advance that a satisfactory integration is possible. In integrational semiology, signs are not prerequisites of communication, but its products (The Semantics of Science, pp. 109‐110.)

The implications of this alternative theory of language are far reaching for any project involving automated language processing, the use of restricted indexing language, subject analysis, international shared databases and cross language information retrieval; in short, for librarians and information scientists.

The structure of The Semantics of Science is straightforward and loosely historical. The first three chapters (“Language and the Aristotelian scientist”, “Before and after Aristotle” and ‘Semantics and the royal society”) discuss the origins of reocentric theories of language in Aristotle, as well as all anachronistic modern attempts to trace “science” back to the Greeks (or even beyond) and most importantly the development of a language of science, this last project being ultimately the center of Harris' interest. “How do words relate to the real world?” was the matter solved by Aristotle with a reocentric theory of language and much later by Wilkins (1668) in his Essay towards a Real Character and a Philosophical Language, a book which should be familiar to any student of the history of classification and nomenclature in the sciences. Harris describes the reocentric ideal thus:

The model offered is one in which Nature's stall is already laid out with her genera and species neatly arranged upon it. All that scientists have to do is come along and, by means of careful observation and experiment, affix the right verbal labels to the right items. Getting it right is the foundation of good science (The Semantics of Science, p. 14).

The problem encountered by Wilkins (and others) in setting up a language of science is that such a vocabulary presuppose the results of scientific knowledge:

In the first place, drawing up a definitive universal language on this basis of a comprehensive classificatory scheme such as Wilkins proposed presupposes that a complete knowledge of the world of Nature is already available  … . In the second place, if the proposed language is to allow in due course for the acquisition of new knowledge, it must be so constructed as to admit additions. But these additions must not be such as to upset the basic classification: otherwise the whole language has to be restructured every time such an upheaval occurs (Wilkins, 1668, p. 61).

To see this is to see that there is no way out of Wilkins' reocentric dilemma. He cannot put the cart before the horse: a classification system that purports to reflect the “real” organization of the world of Nature cannot be proposed in advance of the experimental investigations necessary to determine that organization. That would be to risk prematurely adopting an “incompleat” language. And an “incompleat” language might be a positive hindrance to “real Knowledge” rather than a means to attaining it … (Wilkins, 1668, p. 62).

He notes that in the matter of biological classification, biologists are even now:

…forced to fall back on quite un‐Aristotelian “polytypic” terms and classifications. Why? Because they now recognize that in Nature “things are not so clear‐cut”. In brief, when it comes to establishing a comprehensive isomorphic correlation between words and the possible classifications of living organisms, that scientific basis is lacking, or has not yet been discovered (Wilkins, 1668, p. 68).

There are some who have claimed that these perennial problems are no longer problems in the age of the internet. The possibility of full‐text searching eliminates the need for any classification, order, and even any description at all. Yet this assumption is also founded upon the reocentric notion that words have fixed meanings and have no necessary relation to their use or context (temporal, geographical, disciplinary, etc.). Most importantly, the entire dimension of time is eliminated and impossible to incorporate into the simultaneity which the internet requires. Harris notes that one of the basic tenets of integrational linguistics is “the ‘principle of cotemporality’, which treats it as axiomatic that all signs are time‐bound. Communication is a matter of integrating past experience with present experience and anticipated future experience.” (The Semantics of Science, p. 115) Many readers of this journal will be familiar with the similar claim made by Hjørland that the provision of subject description (of existing materials) must be made (by us now) with potential future users in mind, this being one of the many points of intersection between Harris' integrational linguistics and the practices of documentation and librarianship.

Chapter 4 “Science in the kitchen” takes up the problem of the relationship between the language of science and the language of the community, a matter reflected in the LIS literature in the debate between the use of a restricted thesaurus of subject terms and the use of “natural language”. He looks at a number of controversies in the modern history of science and philosophy of science concerning this relation. The question was asked at the end of the Introduction:

Is it possible for science to construct a semantics on scientific principles that is independent of the non‐scientific language that most of us speak and write for the purpose of conducting our everyday affairs? (The Semantics of Science, p. 4).

In Chapter 4 the question is rephrased:

Wherever and whenever specialists embark on developing a language of their own, the question will inevitably arise at some posing: “How does the specialized language relate to the general language of the community?” (The Semantics of Science, p. 71).

The problem lies, according to Harris, in the preponderance of reocentric approaches to definition:

If and when the ghost of “real definitions” is finally exorcised, where does that leave the discourse of science? Perhaps in a position where many scientists would not like it to be: namely, as a form of discourse like any other. And therefore language‐dependent, subject to all the defects and fallibilities that words are heir to (The Semantics of Science, pp. 68‐69).

At the conclusion of this chapter Harris points to a problem which is exactly the problem of documentation in all its forms. All of the controversies discussed in the chapter, Harris claims:

… are not about the validation of what the scientist says. Nor are they about problems of scientific method. They are controversies, rather, about whether and how what the scientist says can be linguisticaly integrated into other (non‐scientific) forms of discourse. This integration is the ultimate locus of dispute (The Semantics of Science, p. 81).

In academic libraries, one of the principle loci of that integrational activity is the library catalog, and bibliographical description, subject analysis and indexing are all founded on one or another solution to disputes of exactly this nature.

Chapter 5 “The rhetoric of linguistic science” discusses the history of linguistics, and among the many revealing and relevant comments he makes, his remarks on some nineteenth century attitudes towards comparative philology are sure to catch the attention of readers of this journal:

The corollary of Müller's strategy is to treat the viewpoint of the language user as totally irrelevant to the scientific investigation of language … The scientific method that Müller espouses requires the linguist to abstract from the speakers altogether, to ignore the communicational activity of the language community and to treat “the language” simply as a set of forms and combinations of forms. Only then is there any basis for the operation of a scientific method which equates the collection, classification and comparison of words with the collection, classification and comparison of rocks and plants.Thus Müller's claim for the scientific status of comparative philology was ultimately based on a misleading metaphor. The metaphor removes language from its natural embedding in the communicational practices of a living community and reduces it to a static inventory of discrete, collectable items. These items, recorded one by one in the notebooks of grammarians and lexicographers, are available for inspection, analysis and classification on whatever principle the linguist may decide (The Semantics of Science, p. 88).

In chapter 6 “Mathematics and the language of science” Harris provides a radically original interpretation of the meaning of numbers and a criticism of the notion of the “language of number”.

If reocentrism [is] to yield fixity of meanings, it can only do so on the supposition that this world of external things, from which words derive their meanings, remains constant. Otherwise, meanings would be changing all the time. In the case of numerals, however, reocentric semantics encounters a problem; namely, that there are no obvious “real world” entities which stand as their meanings in the way that butter can be regarded as designating a certain fatty substance, or melts as designating a certain kind of process (The Semantics of Science, p. 120).

The integrational approach to number does not tie the meanings of numbers to any specific abstract meanings, timeless eternities or Platonic ideal forms. The search for a language of science which more and more closely approaches the mathematical ideal presumes that such a language would be more and more perfect, “an ideal fixed‐code in which misunderstanding is impossible … somehow able to override all linguistic diversity, because its definitions are based on universal truths that brook no denial.” (The Semantics of Science, p. 111). Instead, Harris claims, numbers are like all other words:

Their value depends on the context. When 554256 functions as a telephone number, it has no arithmetic value at all. But when integrated into certain types of operation it does. The basic types of operation involved are calculation and measurement (to be considered in Chapter 7); and the reason why numerical signs then acquire specific contextual values is that this is required in order that the integration of activities can proceed and be successfully completed. The social utility of these activities is thus the ultimate reason for the apparent semiological stability of this type of sign and its incorporation into pedagogic programmes of instruction and other traditional practices (The Semantics of Science, pp. 127‐128).

The seventh chapter “Science and common sense” discusses measurement and operationalism in science, and, as one would expect, Harris focuses on how scientists write about the results of science which run counter to all common sense, i.e. the “communication barrier” in science. How does one translate mathematical statements into English (or Navaho, Silozi, Buginese) prose? The theory and mathematics behind Galileo's famous “Eppur si muove” was incomprehensible to his audience, but what he meant was understood in spite of that; had his comments been restricted to a mathematical formula, he would have had no trial in the first place.

Kuhn and Carnap get their due in chapter 8 “Supercategory semantics” where the issue is “What is science?” At the heart of this chapter is an investigation into how it is that sciences get to be science. Some, like library science and information science, simply appropriate the term into the name of their discipline in order to win the battle before it starts: given library science, one asks only “In what does it consist?” not “Really? Since when?”. Harris himself is not worried about whether or not linguistics or library science or any other science actually is or is not a science. Nor does he think that this should concern scientists any more than the possibility or impossibility of a perfect language of science.

Those who regard themselves as scientists need not be unduly worried about this. unless, perchance, they attach as much – or more – importance to being called “scientists” than to the work they actually do. If there be any such, I do not think it will hinder their work, or make it less valuable, to rid themselves of the illusion that the language of science has a more scientific basis than the language of the home or the street; or to concede that it is subject to the same semantic indeterminacy and the same context‐dependence as all forms of human communication that have so far been devised. That realization might even be a scientific step forward? (The Semantics of Science, p. 175).

In the concluding chapter “Integrating science” Harris states clearly the basis of the problem for any unified semantics of science, the reason that there can be none:

Why not? Because human experience does not reduce to the quantifiable, and human experience as a whole is what underwrites language (The Semantics of Science, p. 180).

This understanding alone puts Harris in a class almost by himself, at least among linguists. It is this understanding which allows Harris to ask questions where nearly everyone else rests comfortably in myths, whether those myths come from the Bible, Plato or Chomsky. And the difference which Harris brings to linguistics is what makes his work, virtually alone in the literature of linguistics, not only worthwhile for everyone outside the discipline of linguistic, but absolutely exciting for anyone in LIS.

In his 1997 review mentioned above, Warner warned that “Neither work is easy, nor fully comprehended here” and I would have to add that Harris' books are perhaps more difficult books for linguists to read than for others precisely because they require the reader to rethink some of the linguists' “deeply ingrained notions” (Warner, 1997, p. 194). Yet stylistically all of Harris' writings are consistently as clear as any academic writing one could ever hope to read. Having read some thirty volumes of the writings of Harris and other linguists working from the perspective of integrational linguistics, I am convinced that this theory presents a departure from previous linguistic theories as radical as Copernicus' astronomical theory differed from the cosmology of the ancients Greeks. Neither a short review nor a much longer review would suffice to argue the case for integrational linguistics (that argument has been carried out in a few dozen books and many articles), let alone prove its relevance to librarianship and information science.

Throughout this review I have quoted extensively and limited my own comments; Harris' own words are clearer and more concise than any summary that I could provide. The quotations should suggest –to some readers at least – the relevance of Harris' book – in fact many of his books – to the principle problems confronting those working in the various areas of documentation and information retrieval, from catalogers in libraries and software designers working on web search engines to researchers involved in cross‐language information retrieval and automatic translation and indexing.