Just as there is no single thing ‘literary criticism’ which produces a ‘critic’ who can be identified by reference to his methods and beliefs, so linguistics as a homogeneous, evangelical, operative-producing subject does not exist. We cannot altogether predict a linguist’s attitude to his analysis of a text. There is no one linguistic method with easily characterizable modes of operation and endproducts. Certain fundamentals are common to all who call themselves linguists: the beliefs that language changes, is patterned, is only conventionally connected with the outside world, and has an analysable form, for example. But when we look closer we find fewer agreements on details. The history of linguistics reveals tentative effort, revisions of opinion, lack of universal continuous progress. There have been, as in the development of any new discipline, independent schools of linguists: following Bloomfield, Chomsky, Hjelmslev, for example. There is diversification within the field of language studies independent of adherence to general theories: a linguist to the world may be, to his colleagues, a phonetician, dialectologist, grammarian, lexicographer. There is disagreement on the relationship between linguistics and other subjects: is it a social science, a physical science, a humanity? is it most closely related to literary criticism, sociology, psychology, philosophy? We cannot, it seems, be sure what is meant by ‘linguistics’ in a title like that of the present chapter, nor what the author intends to produce when he places linguistics vis-a-vis literature. A linguistic analysis of a literary text could, judging by some precedents, be exclusively grammatical, or metrical, or lexical, or phonetic, and, whichever of these emphases governs it, it could use the language of any one of a number of existing forms of linguistics. In effect, most linguistic analyses of a text of whatever kind will be linguistic by virtue of certain fundamental tenets common to all linguists since de Saussure. Variation of detail in the techniques and terminology will depend on selection of a particular school of linguistics to be followed.

Today, perhaps more than at any other time in the twentieth century, there may be great uncertainty in our decision as to what linguistics is. We are conscious of great differences between our ways of thinking and those of our predecessors in the formative years of ‘classic’ linguistics before the Second World War. We are, it is hoped, less likely to champion one school exclusively. Almost everyone must be familiar with the basic texts of the major schools of thought: not only Bloomfield, Sapir and de Saussure, but Harris, Chomsky, Firth, Hjemslev and Jakobson. Linguists of different persuasions are in close contact through reciprocal visits between England and America. The post-war years have seen developments outside as well as within our field which could not have failed to influence. Acoustic phonetics has increased enormously the precision with which the sounds of utterances can be described, and has necessitated a new look at the way we interpret vocal noises as meaningful activity. Communication theory, methods of quantifying information, and statistical procedures have provided new ways of analysing and talking about language. Machine translation has made demands on our ingenuity in the compilation of grammars and vocabularies for this specialized use. The war itself, bringing a sudden need for intensive language-teaching activity, provided an impetus to pure and applied linguistics. Yet more changes in the field of linguistics are moving us in new directions. In this country, the expansion of linguistics as a university subject, and improved facilities for communication between linguists (principally the formation of the Linguistics Association) have led to the corporate building up of a new linguistics much indebted to the late Professor Firth. In America, a mode of analysis which is certainly new and which has been hailed as revolutionary (transformation-generative grammar) has grown up in less than ten years.

So, in 1965, linguistics is a huge, diverse cluster of subjects of complex historical origin and evidently in a period of great change: the conception of it as a set, arid, insensitive and limited group of stereotyped techniques is a fiction. In my title is implied no question of an established and unprogressive linguistics moving in on a field outside its competence; rather, it is a developing discipline hoping to learn about language by turning its attention to those texts called ‘literary’, and in the process exploring its relation with literary criticism, with which it has in common a basic concern with the uses of language.

The current desire to investigate literary uses of language should be seen in a historical setting: as the end of a series of alternations in the interest of language students in texts of different character. One of the early influences on the rise of twentieth-century linguistics was reaction against the past, specifically against lack of interest in spoken language and against preoccupation with highly valued examples of written language. So (to oversimplify somewhat) we find nineteenth-century philologists interested predominantly in written texts, despite the great advances in linguistic methods made in that century, and on the other hand the writers of prescriptive grammars persisting in basing their comments on the language on the forms of written documents, advocating the imitation of ‘classic’ literary works. Against this (and additionally because of their concern with the unwritten languages of the indigenous peoples of America) American linguists erected the principle that speech is the primary form of language. Written language, it was argued, is a derived system, based on speech and coming after speech in the development of both civilizations and individuals. We find countless examples of definitions of language framed deliberately to exclude written language, of grammars which claim to be based entirely on spoken forms. This attitude leads to an implied denigration of written language, to a view of peculiarly literary forms as modifications of ‘normal usage’. One type of normative grammar was replaced by its converse. It is not until the ‘fifties that we find evidence of a swing away from this position. American linguists/anthropologists had long interested themselves in the oral literatures of non-literate peoples—often in the correlation between cultural factors and the selection of deviant, characteristically literary, features. Circumstances in the ‘fifties were particularly favourable to making the step from oral literature to written texts, involving as it did giving less prominence to the ‘primacy of speech’ principle and recognizing that, in fact, much past analysis of so-called colloquial texts had treated material which often manifested grammatical forms not particularly associated with spoken language. In the last few years, some discussions of literary texts by linguists attracted attention to the applicability of linguistic techniques to literature, and there have also been a number of general or theoretical essays.¹ At the same time some literary critics came to acknowledge the necessity for the close study of the language of literature.² In these recent years, also, developments in linguistic theory have directed attention to problems of great interest in literary studies. For example, in Language for 1952 Z. S. Harris gave an outline of a new form of linguistic description, ‘discourse analysis’, one of the cardinal points of which is its interest in those distributions of linguistic elements which link sentences within a text. Previously, linguists had taken the sentence, a relatively small unit, as their upper limit of magnitude for description; now, methods of analysis were promised which dealt with stretches of language longer than the sentence. Now perhaps linguistics could move to a consideration of formal devices, including patterns in literature, which tended to unify or structure continuous texts. Again, we find in the writings of Chomsky in the late ‘fifties a concern with the question of grammaticalness—with the status of utterances like colorless green ideas sleep furiously in relation to the grammar of English. As such utterances are more likely within literature than outside it, Chomsky was in fact approaching some facets of literary expression. It appears (e.g. from the discussion in Sebeok’s Style in Language) that the Chomskyan notion of ‘degrees of grammaticalness’ is of no slight importance in the study of literary language. Finally we should mention the very recent development, chiefly by British linguists, of ways of discussing lexis (vocabulary), especially the study of combinations of lexical items (collocations); a development which takes some of the load off grammar in the description of forms of expression. All three linguistic interests are symptomatic of the linguist’s growing desire to increase his range.

I mentioned above changes in the attitudes of linguists to the relation of written to spoken language. One view was that literature is characterized by formal features different from the conventions of spoken or colloquial usage. This argument was used in the past to justify concentration on spoken language (‘normal’ or ‘common’ usage) alone. But it is also one which deserves consideration on its own account, as one of the several possible views of the relation between literary and non-literary forms of discourse. We should now look at some of these views, implying as they do the definition of literature.

Several relationships, and sets of terms, are involved in this basic question: written language and spoken language, literature and non-literature, common usage and literature, poetry and literature. The simplest opposition is between written and spoken realizations of language, and the chief difference is the obvious one: that speech and writing employ different media for the transmission of messages. One works through noises in the air, the other through marks on paper or other material. These means of transmission are termed substance by linguists.³ Substance is only the surface of language, and I introduce it only as a linguistic category with which we are not really concerned. It is the province of phonetics, or calligraphy, or typography. The linguist and literary critic need to go deeper than this. There are in fact other characteristic differences between speech and writing. They are not translatable into each other simply by changing their substance. This chapter cannot be turned into speech merely by reading it aloud: if this is done, it will be found to exhibit formal characteristics (i.e. linguistic features at a deeper level than substance) which would not be found in a sample of speech which had not been ‘translated’ from written language. Conversely, a written translation of an informal conversation will not fulfil one’s expectations of the forms of prose. That there are formal as well as substantial differences is attributable partly to the circumstances attendant upon the selection of either phonic or graphic substance: the greater speed and spontaneity of speech; the consideration of permanence of the written word; in the case of writing, the prescriptive training and the consultation of normative models in our education and reading—only rarely is specific training given in speech. But more influential in making speech and writing differ formally is what Mr Leech (Chapter 8) calls the register scale. Differences of form between speech and writing may, it is true, often be accounted for by factors just linked to the difference in substance; but more powerful in determining formal features—irrespective of realization in substance—is the place an utterance occupies on the register scale. Spoken utterances differ formally from written because they are, typically, used in different communication situations and for different purposes: for calling a dog, comforting a child, ordering a meal, but rarely for annotating a text, giving a recipe, making a will. I shall return later to the register scale, the scale of situations and functions which governs the selection of forms, and treat it as a selective principle overriding the distinction between speech and writing; but this distinction does occur at a definite point in the scale: some groups of situations select written substance and characteristic forms, while others tend to select phonic substance and another set of forms.

Poems have the characteristic that while they are most usually originally realized in written form, their reading is often a spoken one. We need not consider the question of whether a poem is designed as a spoken utterance and committed to writing only as a guarantee of permanence (but some poems, for example some by George Herbert or by Dylan Thomas, have visual form independent of their phonology). Professor Firth considered that all written texts have ‘implication of utterance’, and this is certainly a necessary assumption in the study of metrics. We need note only that poetic texts rather than any others may exist in both types of substantial realization—phonic and graphic. The problem is to determine what aspects of phonic substance are part of the poem, and what are to be relegated to the reading only. As several readings of the poem may offer many phonetic variations, the question is a vital one: in many phonetically (substantially) different readings, in what sense is the poem repeated? what features of the readings are part of the poem, and what are merely the embellishments of the recitation? Four roughly corresponding pairs of terms are available to aid our thinking over this. General linguistic theory provides langue vs. parole (de Saussure’s distinction between the language as an abstract system and speech as an individual act of articulation), and form vs. substance (current British usage making a similar distinction, form being an abstraction made from substance). C. S. Peirce’s distinction of type from token has sometimes been adopted to distinguish between a linguistic unit (e.g. ‘poem’) and its realizations. Finally, some of the contributors to Sebeok’s Style in Language spoke of message vs. performance. Message is an unfortunate term, carrying its more usual connotations of ‘information carried by an utterance’. Performance is a useful word in connexion with recitation. All these pairs of terms serve to distinguish a poem ‘which constitutes a replicable, invariant structure’ from ‘its acoustic implementation in the concrete performance’ (Edward Stankiewicz, Sebeok, p. 75). Stankiewicz goes on:

I shall summarize my view of this distinction, using three of the available terms. A poem (like any utterance) has form, which is invariant and repeatable; form is the proper object of study for stylistics, as for linguistics; it has describable meaning on various sublevels. It can be realized in two ways, both of which are substance: as a written record (e.g. the poet’s written record, or the transcription of a poem of oral origin) or as a spoken recitation. As poetry, of all literary utterances, demands recitation, it is useful to signify this by the use of a special term, performance. A performance, although most usually stimulated by a reading of written substance, is not to ...