Chapter 7 The meaning of cohesion 7.1 Text In Chapter I we discussed what was meant by TEXT, and introduced the concept of cohesion to refer to the linguistic means whereby texture is achieved. In this chapter we resume the discussion in the light of the account that has been given of the various types of cohesion in English. A text, we have suggested. is not just a string of sentences. In other words it is not simply a large grammatical unit, something of the same kind as a sentence but differing from it in size - a sort of supersenteru:e. A text is best thought of not as a grammatical wllt at all, but rather as a unit of a different kind: a semantic unit. The wllty that it has is. a unity of meaning in context, a texture that expresses the fact that it relates as a whole to the environment in which it is placed. Being a semantic unit, a text is REAL! ZED in the form of sentences, and this is how the rdation of text to sentence can best be interpreted. A set of relared sentences. with a single sentence as the linUt:ing case. is the embodiment or realization of a text. So the expression of the semantic unity of the text lies in the cohesion among the sentences of which it is composed. Typically • in any text, every sentence except the first exhibits some form of cohesion with a preceding sentence, usually with the one immediately preceding. In other words, every sentence contains at least one anaphoric tie connecting it with what has gone before. Some seruences may also con-- tain a cataphoric tie, connecting up with what follows; but these are very much rarer, and are not necessary to the creation of text. Any piece of language that is operational, functioning as a unity in some context of situation, constitutes a text. It may be spoken or written., in any style or genre, and involving any number of active participants.. It will usually display a form of conslstency that is defined by the concept of
294 register: a consistency in the meaning styles or types of semantic con- figuration which embody its relation to lts environment. In other words, a text is usuaUy reasorubly homogeneous, at least in those linguistic aspects which most dosely reflect and express its functional relationship to its setting. 7.1.1 Length of text Text may be of any length. Since it is not a unit of the grammatical rank scale. and does not consist of sentences, it is not tied to the sentence .as its lower limit. Many familiar textS in fact come out as less than one rentence in the grammatical structure. Warnings, titles, announcements. inscriptions. and advertising slogans often comist of a verbal. nominal. adverbial or prepositional group only, for example r7: I} a. No smoking h. Site of early chapel c. For sale d. National Westminster Bank e. Do not feed Equally, there is no upper limit on the l~ngth of the text. An entire book may, and in many genres such as fiction typically does, comprise a single text; this is what is implied in the term 'a novel'. The same is true of a play, a sermon, a lecture. or a committee meeting. The type of presupposition that provides texture in the text, in other words what we are calling cohesion. can extend over very long sequences. We ftnd in evetyday conversation elements turning up which presuppose earlier passages from which they are separated by many minutes and even hours of speaking time; and writers exploit this potential by making cohesive ties across vety long stretches of text. It is dear that the awareness of text that we develop as part of the learning of the mother tongue is rather free from constraints of time, and depends much more on con- textual relevance and. integration of the language with the environment. 7.1.2 Definitiveness of the cr:mcept of text It would be misleading to suggest that the concept of a text is fully determinate, or that we can always make dear decisions about what con- nituces a single text and what does not. We can often say for certain that the whole of a given ~ssage constitutes one text; and equally we can often say for certain that in another instance we have to deal with not one text
7.! TEXT 295 hut two, or more. But there are very many intermediate cases. instances of doubt where we are not .at all sure whether we want to consider all the pans of a passage as filling within the 'Same text or not. Usually for practical purposes this does not matter very much. We are all intuitively aware of the validity of the general concept of a text; we know that there is such a thing, whether or not every instance can be unambiguously identified. What we react to, as speakers and listeners, readers and writers, in forming judgments about- texture, is precisely the sort of cohesive structure the details of which we have been exploring in the preceding chapters. Since the speaker or writer uses cohesion to signal texture, and the listener or reader reacts to it in his interpretation of texture, it is reasonable for us to make me of cohesion as a criterion for the recognition of the boundaries of a text. For most purposes, we can consider that a new text begins where a sentence shows no cohesion with those that have pre- ceded. Of cour5e, we shall often find isolated sentences or other structural units which do not cohere with those around them. even though they form part of a connected passage. But usually if a sentence shows no cohesion v..-ith what has gone before, this does indicate 2 tr.msition of some kind; for example, a transition between different stages in a complex transaction, or between narration and description in a passage of prose fiction, We might choose to regard such instances as disconrinuities, signalling the beginning of a new text. Sometimes then the new text will turn out to be an interpolation, as in {r:8] and (1:9] in Chapter I, after which the original text is once again resumed. So although the concept of a text is exact enough, and can be adequately and explicitly defined, the definition will not by itself provide us with automatic criteria for recognizing in all instances what is a text and what is not. In all kinds of linguistic contexts, from the most formal to the most informal, we constantly have to do with forms of interaction which lie on the borderline between textual continuity and discontinuity. But the existence of indeterminate instances of this kind does not invalidate or destroy the usefulness of the general notion of text as the basic semantic unit of linguistic interaction. 7.1.3 Tight and loose texture The frequent shift between narrative and verse in Ali« provides an excellent illustration of the kind of transition that takes place between
296 subtexts within a text. The verses are often quite outside the context of the narrative, and function as independent texts in thei.r own right; they display no cohesion with what has preceded them. An example is The Queen of Hearts in tbc final chapter of Alice in Womlnl.md. At the same time, the verses. are often ;mticipated by some reference to poetry or rong, or to the poem or song in question. so that the verse text .as a whole is placed in an environment not unlike that of quoted speech. Here is an example: [7:2] 'The piece I am going to repe~t,' [Humpty Dumpty] went on without noticing her remark, 'was written entirely for your amusement.• Alice felt that in that case she rea11y ought to listen to it, so she sat down, and said, • Thank you • rather sadly. •m winter when the fields are white, I sing this song for your delight - . , . '. Here there is lexical cohesion : song ties wlth piece in the first sentence and this in turn with poetry occurring a short while earlie[". This gives a fair indication of something that is a general feature of texts of a11 kinds. Textuality is not a matter of all or nothing, of dense dusters of cohesive ties or else none .at alL Characteristically we find variation in texture, so that textua1ity is a matter of more or less. In some instances there will in fact be dense dusters of cohesive ties, giving a very close texture which serves to signal that the meanings of the parts arc strongly interdependent and that the whole forms a single unity. In other instances, ho"h-ever, the texture will be much ]ooser. There will be fewer cohesive ri«, perhaps just one Q[" two. In Alia this altemation between tight and loose texture gives a very defmite flavour to the whole. At one level. the whole of Alice is very much a single text. But when we shift our focus of attention we find that it contains portions that ar:e less closely knit with the remainder, particularly the songs md the verses. And this is signalled by the relative cohesive independence of these from the surrounding passages- mually. however, a partial not a rotal indepen- dence. Such a thing is typical of texts of many kinds. Some writers in particu- lar seem to achieve a sort of periodic rhythm in which there is a regular alternation between tight and loose texture. In this connection we see the importance of the paragraph. The paragraph is a device introduced into the written language to suggest this kind of periodicity. In principle, we shall expect to find a greater degree of cohesion within a paragraph
7.1 TEXT 297 than between paragraphs; and in a great deal of written English this is exactly what we do find. In other writing, however? and perhaps as a characteristic of certain authors, the rhythm is contrapuntal: the writer extends a dense duster of cohesive ties across the paragraph boundary and leaves the texture within the paragraph relatively loose. And this itself is an instance of a process that is very characteristic oflanguage altogether. a process in which two associated variables come to he dissociated from each othcr with a very definite semantic a.nd rhetorical effa::t. Here the two variables in question are the paragraph sttucture and the cohesive structure. The paragraph evolves first of all as the written symbol or representation of a periodic pattern that we might represent in the following way: more -~ fewer The vertica1 lines represent the paragraph boundaries and the wavy line represents the density of cohesive ties. Subsequently however the paragraph comes to function as a pattern maker (as distinct from being merely a pattem marker) in its own right, :md something like the following picture emerges: more .Jl fewer This represents the sort of writing in which the paragraph structure is played off against the cohesion, giving a complex texture in which the rhythm of the eye (and associated bodily rhythms of reading) is balanced against the rhythm that is engendered by the alternation between tight- ness and looseness of cohesive patterning. 7.1.4 Imaginary texture Finally we may mention the type of cohesion which imposes an imaginary texture. by setting up expe-ctations in the reader or listener which. since they are expectations of the- past, by theic nature. can not be satisfied. Alice will again serve as an example.
298 TH£ MEANING- OF COHESION The very first sentence of Through the Looking Glass i.s: [ 7: 3] One thing was certain. that the white kitten had had nothing to do with it;- it was the black kitten's fault entirely. This sentence is dearly marked as cohesive, by the oceutrence of the reference item it. In other words, the narrative begins as if one was already in the mid-dle of i.t; it appears to presuppose a great deal that has. gone before, but in fact nothing has gone before so "We have to supply it for ourselves. Our interest is immediately engaged, since we inevitably start searching fOr some interpretation of the it. In this instance the reference, as often, is resolved cataphoricaiiy; we Ieam two paragraphs later that it refers to unwinding and entangling a ball of wooJ. This -device is commonly exploited in the opening of short stories, where it sets the tone for a genre whose meaning as a genre depends on the implications that what is in the text is not the whole story. It is also used in other contexts; the example was quoted in Chapter I (z.I..z) of the radio comedian who began his patter with the words so we pushed him under the ether one. This type of false or unresolved cohesion creates an effect of solidarity with the hearer or reader. It puts him on the inside, as one who is assumed to have shared a common experience with the speaker or writer. In its use in written fiction it is perhaps akin to the typical beginning of an oral folk narrative, which assumes prior knowledge of the matter of the tale on the part of the audience and makes allusion to the characters. the events or the circumstantial backgroWid in a form which often looks anaphoric, although there has been no previous mention, at least on the occasion in question, Similar properties are found in the oral narratives of young children, which presuppose .a sharing of experience with the listener. The line between real a.nd imaginary anaphora is not, after aU, very dearcut; a great deal of news reporting depends for its interpretation on the assump- tion that the previous day's newspaper was part of the same text. And what is one text for one participant in a situation may not always be so for .another, as appears when a person who has been day-dreaming suddenly voices one ofllls thoughts aloud. 7.2 The general meaning of cohesion The general meaning of cohesion is embodied in the concept of text. By its role in providing 'texture', cohesion helps to create text. Cohesion is a necessary though not <1 sufficient condition for the crea-
7.2 THE GENERAL MEA!'ING OF COHESION 299 tion of text. What creates text is the TEXTUAL, or text-forming, compo- nent of the linguistic sysre:m, of which cohesion is one pan. The textual component as a whole is the set of resources in a language whose semantic function is that of expressing rdationship to the environment. It is the meaning derived from this component which characterizes a teA1: ~ which characterizes language that is operational in some context, as distinct from language that is not operational but citationa.l, such as an index or other form of verbal inventory. The textual component. and the place of cohesion within Jt, was discussed briefly in Chapter I. The concept of a textual or text-forming fl.lllction in the semantic system provides the most general answer to the question of what cohesion means. The textual component creates text. as opposed to non-text, and therein lies its meaning. Within the textual component, cohesion plays a special role in the creation of text. Cohesion expresses the continuity that exists between one part of the text and another. It is important to stress that continuity is not the whole of texture. The organization of each segment of a discourse in terms of its information structure, thematic patterns and the like is also part of its texture (see 7,4.I below), no less important than the continmty from one segment to another. But the continuity adds a further element that must be present in order for the diM:ourse to come to life as text. The continuity that is provided by cohesion consists~ in the most general terms, in expressing at each stage in the discourse the points of contact with what has gone before. The significance of this lies in the simple fact that there are such points of contact: that some entity or some cir- cumstance~ some relevant feature or some thread of argument persists from one moment to another- in the semantic process. as the meanings unfold. But it has another more fimdamenta] significance, which lies in the interpretation of the discourse. It is the continuity provided by cohesion that enables the .reader or listener to supply all the missing pieces, .all the components of the picture which are not present in the text but are necessary to its interpretation. One of the major problems in unders.tanding linguistic imcracrion - it is actually a problem in the understanding of ALL text processes, whether those of dialogue or others, though it is usually posed in the context of dialogue- is that of knowing how the listener fills in the missing informa- tion. The listener assigns meanings and interprets what is said to him; but in doing so he is himself supplying a great deal of the interpretation. The sentences and clauses and words that he hears, however perfectly formed lex:icogrammaticaUy (and, contrary to a popular belief. in ffiOS[
,300 speech contexts they are very well formed indeed), are semantica11y full ofholes. Or rather. this is the wrong metaphor. The situation is: sometimes represented as if there were omissions which the hstenet" had to supple- ment; as if the semantics of discourse was like a jfgsaw puzzle with missing pieces in it. It would he more appropriate to describe it in terms of focus. What the lexicogrammar of t-he text presents is more like a picture that is complete but out of focus. with the outlines blurred and the details imperceptibk And if we take one further step and postulate that the picture to start with was not a photographic likeness but a symbolic representation, then we sha11 get some idea of the nature of the decoding process -for that is what it is -that the listener goes through. What makes it possible for him to go through the process is the fact that what he hears is systematically rdated to its environment - it has 'textual meaning', as we have expressed it; and an essential component in this relationship is its continuity with what has preceded. The continuity is not merely an interesting feature that is associated v.-ith text; it is a necessary element in the interpretation of text. There has to be cohesion if meanings ar:e to be exchanged at all. This is so easy to illustrate that it is often forgotten. Con$ider the exam- ples that have been cited throughout this book. The vast majority of them have been either drawn from Alice in Wonderland or made up. Why? This is the only way to ensure that attention would he focused. on the point at issue: either to use a text that is so familiar that the reader will not pause over .its interpretation, or to construct examples that are so arti- ficial that they avoid the problem. If we had taken isolated sentences from real-life texts. they would have looked something like the follow- mg: [7:4-} a. Two rolled otflt and stopped. as though arrested by a witch's wand, at Mrs Oliver's feet. b. This is a one with animals too. animals that go in water. c. Administration spokesmen were prompt to say it should not be considered any mch thing. d. You could see them coming on him, before· your very eyes. e. I expect you will get this but I'll send it if you want. f. It was the morning caught for ever. g. So he proposed having his discovery copied before parting with it. These are typical examples of what people say and write - except that they do not say or write them in isolation. In interpreting them, we build
7.2 THE GENI!.RAL MEANING OF COHE-SION jOT in. along with other environmental factors. the continuity element; we do not even notice the indeterminacies and :all the dUierent meanings we could • read in', because the lens is alread-y in place before the picture comes along to be interpreted. But the process of interpretation goes on, and the patterns of cohesion have played a central part in it. The point is perhaps obvious enough. But we often fail to realize just how much of our interpretation depends on this continuity with what has gone before. It is not only the referents that we have to supply~ the meaning of two and it in (a), this in {b). it in (c). them and him in {d), this in {e), it in {f), he in (g). Nor is it simply information of the kind that is demanded by the too in (b), the such in {c) or the so in {g): •tn addition to what?', 'any thing such as what?', 'why did he propose having it done?' Taking these sentences by themselves we have no idea, or rather we have only the haziest idea, how to interpret the things - the objects, events and so forth- that are encoded in the grammatical structures. and the lexical items. What kind of rolling took place? What does with mean, in with animals? ln what ways. are things coming on him ? What can be made of you will get this but I' 11 send it? What sort of discovery is to be copied., how, and why? We cannot begin to visualize the morning, and we do not know whether it is a morning that has been mentioned before or one that is to be identified exophorically, as unique or at )east recognizable und:er the circumstances. We do not know whether the spokesmen for the Administration are talking about an object, an institution or a lengthy passage of text - a fact or report. There is nothing unusual or mysterious about any of these examples~ but they are out of focw, and will click into place onJy when we put them in their textual environment and satisfy the queries which they arouse. It is hardly necessary to do this in order to demonstrate the point at issue. However, the reader may feel deceived if nothing further is said, so here is the immediately preceding eo-text for each of the above examples: (.a) Joyce, a sturdy thirteen-year-old, seized the bowl of apples. (b) This mobile's got fishes,. yours has animals. (c) During the hearing on Wednesday, Inouye said the questions fur- nished by Buzhardt • should serve as a substitute, admittedly not the very best, but a substitute for cross-examination of Mr Dean by the President of the United States •. {d) Spots. AH over his face and hands - (e) Nothing else has come for you except Staff Bulletin no 2.
]02 THE MEANll"G OF COH.ESION (f) There on the mugh thick paper. reduced to their simplest possible terms, were the stream, glittering :and dimpling, the stone arch of the bridge flushed in morning sunlight, the moor and the bills. (g) The nobleman, it appeared, had by this time become rather fond of Nmna and Pippa. He liked. it might be said, the way they comported themselves. Anything approaching a •fuU' interpretation is likely to need. much more information than is recoverable from a preceding sentence or two. For example, in the text containing (g), two pages earlier, wa~ the senten<:e It was ,a highly indecent picture. Preceding (a). at intervals. there have occurred it was to be ,a Hallowe'en party ... and lvf:rs Oliver was partial to apples. Moreover the whole text has, in turn, been preceded by other texts containing accounts of Mr:s OHver .and her fondness for apples, as well as associated references to witchcraft. In the same way (c) has been preceded not only by six columns of detailed news but by six months of almost daily reporting centring around the Watergate affair:. In (d) the chaotic absence of cohesion is used as a comic device to suggest information being extracted from some- one against his will, though in fact {as the audience knows from the preceding text) the reluctance is feigned and the information is false: Patch: Mind yo1.4 Sam, it may not be that at aU. We can't tell what poor old Slivers has got - Mellock ~Who's Slivers? [As they do not reply, Grindley shaking his head at Patch. Ursula cuts in.l Ursula: Is he the man you had locked. in that cabin ?fAs they do rwt reply] Heis,isn'the? [They nod.] Well. what's the matter with him? Patch: It was the only thing we could do. you know. Until the doctor came. Mellock [not lildng thisJ: The doctor? Ursula: Come on. What's the matter with him? (They are obviously reluctant to answer.] He was taken ill, wasn't he? Patch:. All hot and flushed. Then breaking into spots. Gridky [warningly]: Bob! You know, we promised. Unula: Don't be idiotic. You've gottoteH us. Patch fwith feigned reluctafla]: Spots. All over his face and hands -
7•3 THE MEANING OF DIFFERENT KINDS OF COHESION 303 Gridley: You could sec them coming on him, before your very eyes. About that size. [Indicates.} No bigger. [Sfwws them.]* Cohesive ties, especially those ·with the immediately preceding text, are only one source for the information iliat the reader or listener re- quires. Both situational and more remote textual information are necessary components. But it is surprising how much can ofien be recovered simply from the presuppositions carried by the cohesive elements. The ongoing continuity of discourse is a prima-ry factor in its intelligibility. This illustrates the meaning of cohesion as a whole. It provides, for the text, which is a semantic uni~ the sort of continuity which is achieved in units at the grammatical level - the sentence. the clause and so on - by grammatical structure. Like everything dse in the semantic system, cohesive rdations are realized through the lexicogrammar. by the selection of structures, and of lexical items in structural roles. Our inten- tion in rhis book has been to survey the lexicogrammatical resources in question, and show their place in the linguistic system. But the cohesive rdations themselves are relations in meani:ng, and the continuity which they bring about is a semantic continuity. This is what makes it possible for cohesive patterns to play the part they do in the processing of text by a listener or a reader. not merely signalling the presence .and extent of text but actually enabling him to interpret it and determining how he does so. 7-3 The meaning of different kinds of cohesion We have discussed cohesion under the five headings of reference. substi- tution, ellipsis, conjunction and lexical cohesion. The classification is bated on Hnguhtic fOrm; these are the categories of cohesion that can be recognized in the lexicogrammatical system. In terms of the resources which are brought into play, they are all lexicogramm<Jtical phenomena of one kind or another. Reference, substitution and ellipsis are clearly grammatical, in that they involve dosed systems: simple options of presence or absence, and systems such as those of person, number, proximity and degree of comparison. Lexical cohesion is, as the name implies, lexical; it involves a kind of choice that is open-ende~ the selection of a lexical :item that is in some way rdated to one occurring previously. Conjunction is on the border- line of the grammatical and the lexical; the set of conjunctive elements can probably be interpreted grammatically in terms of systems, but such *].B. Prie:sdcy, Bus 1111 :h~ &at~ ('11lf" Plqs of]. B. Prindey, Vol.2). Heinenunn.
304 an. interpretation would be fairly complex, and some conjunctive ex- pressions involve lexical seJection as well, eg: moment in fiom that nwment on. This tells us about what form cohesion takes. what resources of the linguistic system are drawn on in the expression of cohesive relations. But it does not tell m about those rdations themselves. lf we ask what is the NATURE of cohesive relation~ as distinct from what form of EXPRESSION they take, we get a different answer - one still in terms of the linguistic system, but giving a different kind of explanation. We are now asking about the nature of cohesion considered as a set of rdations in language; whereabouts in the linguistic system are these relations located? In other words., what do the different kinds of cohesion mean? If we look at cohesion from this point of view, we shall be able to recognize three kinds. These are the three different kinds of relation in language. other than the rdation of structure. that Jink one part of a text with another. In the most general terms they are (1) relatedness of form. (2) rdatedness of reference. (3) semantic connection. The way these correspond to the various types of cohesion is as follows: Nature of cohesive relation: Rdatcdness of form Relatcdness of reference Semantic connection Type of cohesion: Substitution and dlipsis; lexical collocation Reference; lexical reiteration Conjunetion 7·3·' General principles behind the different types We have referred to aspects of this general picture at various places in the discussion. It has been pointed out that reference, while it is ex- pressed by grammatical means, is actually .a sem.anti<: relation, a relation between meanings of particular instances rather than between words or other items oflinguistic form. Substitution an.d ellipsis. on the other hand, are formal relations between elements at the lexicogrammaticallevel. It has also been sho'\\"'11 that various consequences follow from this distinction. In substitution. and ellipsis it is always possible to 'restore • the presupposed item (replacing the substitution counter, or filling out the empty structural slot); in reference, typically, it is not. On the other hand a substitute has to preserve the grammatical function of the pre- supposed item; whereas there is no such restriction on reference, which is independent of this sort of formal constraint. Lexical cohesion has some-
7·3 THl! ME-ANING OF DIFFERENT KIND'S OF COHRSION 305 thing of both types. The relation itself is a formal one. between items of the vocabulary irrespective of any referential identity; but lexical cohesion is typically used in contexts where there is identity of reference, and for this reason the cohering lexical item is usually accompanied by tire:, or other anaphoric reference item. Why these two difft>rent types of cohesive relation. one formal the other semantic? This can be explained by the fact that there are two possible channels for the recovery of information: the situation, and the kxt. The concept of SITUATION was discussed in Chapter z. It is a very simple notion., designed to account for the fact that language takes place in social contexts and makes connections with the realities that make up those contexts. The relevant realities are by no means necessarily to be found in the surrounding stage properties., the furniture of the material environment. A social context is a much more abstract conceptio~ a kind of semiotic structure within which meaning takes place; the' realities • of which it is made up may he of an entirely intangible kind. But equally they may reside in the persons and the objects that figure in the imme- diate vicinity; and if so, reference will have to be made to them. This is what we have called exophoric reference. The semantic level in the linguistic system is, among other thlngs. an interface between language and the realities of the outside world. So the exophoric connections with the environment are connections made at the semantic leveL This accounts fm reference. Reference is a semantic relation linking an instance of language to its environment. and reference items are in principle exophoric. The basic meaning of him is • that man out there •. We can see this dearly in the first and second person forms me and yQu, which refer to the roles of speaker and addressee in the communication situation; and also in the demonstrarives with their sys- tem of proximity. •near me' (this) or' not near me' (that), v.-1th sometimes a third term 'not near either of us' (yon), as in [7:5] Yon C:.S,;us bath a lean and hungry look. Secondly. in any connected passage of discourse it will be necessary to refer back to something that has been mentioned already, making explicit the £act that there is identity of reference between the two. There is still, no doubt. .an ultimate referent beyond the language. which defines the nature of the identity between. the two instances.. But the immediate referent of the second instance is the first instance; and it is this imme- diate referent. the previous mention, that now constitutes the relevant
3o6 THE MEANING- OF COHESION environment, not the extralingui~tic referent. Prohabiy,alllanguages adapt their reference items to this function. extending them from exophoric to endophoric use. (This formulation ls not intended to imply that such a development ha~ taken place in the known history oflanguages-, hut rather that it is a development that has probably taken place in the evolution of human language .as a whole.) Thus in English nearly all reference items are also regularly endophoric. In those types of situation in which the perceptual environment is not part of the relevant social context, uses of language which are far removed from • language in action'. endophoric reference takes precedence over ex.ophoric as. a means of e5tahlishing identity. In this way the process of idemification of the referent becomes a cohesive or text-forming proces~. Why do we refer to 'John~ as him rather than .as John? Because John is vague, whereas him is definite. John could be any old John; but him means 'that particular individual whose identiry we have established and agreed upon'. We refer to John as him rather than as John in order to signal that his identity is a feature of the environment. And the same principle applies to the other reference items. The environment has been extended from the situation to include the text. In that case, if the relation of reference may he endophoric as well as exophoric - if a reference item can refer to an element in the text as well -as to an element in the situation - we may well ask why languages have evolved a second relation of a different kind, that of substitution~ to relate one linguistic item to another. Here the key to the answer lies in the concept of contrast, in the sense of contrastive information. In connected discourse there are very many occasions where we need to repeat some item precisely where there is no identity of reference. For example [7:6) WouldyoulikethiHeapot? -No, I want a ~uare one. Here the second speaker does not use the reference i.tem it, because he does not, in fact, want the object referred to. But he does establish con-- tinuity of a different kind, one based not on referential identity in the given instance but on the identity of the linguistic elements involved. The continuity lies not in the meaning but in the form. The use of the substi- tute one means • supply the lexical item that just figured as Head of the nominal group'. The relation between the mu instances is a relation established at the lexicogrammaticallevel. h is not, of course. without its semantic aspect; but the semantic implication is of a different kind. The general class of objects. in this case
7•3 'IRE MEANING OF DIFFERENT KlNDS OF COHESION 307 'teapots', constitutes the link between the two. But the significance of the continuity that is established by the use of the substitute is that it is con- tinuity in the environment of contrast. Example [7:6] is a typical instance of cohesion through substitution, where the meaning is 'a non-.identicaJ member of the identical class• _ The contrast may take m:my different forms. The meaning • non- identical member of the identical dass' is merely one that is characteristi- cally associated with. the use of the nomina] substitute one. Bur the con- trast may be in any of the systems associated with the element in which substitution occurs. With the nomina1 substitute, it may be found in the Deictic or Numerative element as well as in the Epithet; while a verbal substitute is typically accompanied by a contrast in polarity, in mood or in modality. In order to express this sort of continuity in the environment of con- trast, the cohesive relation that is appropriate is one that is established not at the semantic level whete there is an implication that the co~ve factor is an extralinguistic one. but at the Jexicogrammati.cal level. Here the implication is that the continuity is essentially a linguistic continuity, that lies in the words themselves: the meaning of substitution is • this is the same word that we had before •. It is thus inherently a textual. not a situarional rdati.on. and is used in exophoric contexts only with a special effect. that of creating the illmlon that the presupposed item. has occurred before. We have used the formulation 'contrast" or 'contrastive information' to draw attention to the special feature which distinguishes substitution from reference. This might suggest that there is always some negation involved: •not what was .referred to previously, but {a different one, etc} •. This is the typical form that the contrast takes; but it is not the only form.. Consider an example such as {7: 7] I want three teapots. 1"11 take this on~ and this one, and this one. Here the contrast simply takes tbe form of new information;. we are talking about teapots (one), and the teapot in question. not specified before. is now being specified (this). In ellipsis, which as we have seen is closely rdated to substitution, this is the usual interpretation; for example [7: 8] What are you doing?- Buying a teapot. Here the eJlipsis of I am displays the continuity and the remainder is thereby signalled as new information. Likewise: [7:9] How many teapots are you buying?- Three.
308 This is the general principle undedylng the difference between reference. on the one- hand, and substitution on the other. Reference is a semantic relation. in which a meaning is specified through the identification of a referent; the source of identification is the situation, so that the relation of reference is basically an exophoric one. It becomes incidentally cohesive~ when the identification is mediated th.rongh rhe presence of a verbal referent in the preceillng text; this then becomes the presupposed element. and the text replaces the situation as the rdevant environment within which the relation of reference is estabJished. Substitution/ellipsis is a formal {lexicogrammat:ical) relation, .in which a form (word or words} is specified through the use of a grammatical signal indicating that it is to be recovered from what has gone before. The source of recovery is the text, so that the relation of substitution is basically an endophoric one. It is inherently cohesive., since it is the pre- ceding text that provides the relevant environment in which the pre- supposed item .is located. Conjunction. the third and. final type of cohesive relation. differs from both of these in that it is cohesive not by continuity of form or reference but by semantic connection. Some relation is established between the meanings of two continuous passages of text, such that the interpretation of the second is dependent on the relation in whi.ch it stands to the first. This relation may be one of two kinds~ either it is present in the ideational meanings, as a relation between things - for example between two events in a narrative; or it is present in the interpersonal meanings, as a relation between elements or stages in the communication process - for example between two steps in an argument. Either of these may be represented as a form of semantic connection between a pair of adjacent clauses; the former as in [7: 10a], the latter as in {7: mb]: [7'10] a. Jackfelldownandbrokehiscrown. And Jill came tumbling after. b. Forhe'sajollygood.fdlow~ And so say all of us. A hrief further discussion of each of the three types of cohesion is given in the following three subsections:. 7.3.2 R£jerenct Reference is the relation between .:m element of the text and something else by reference to which it is interpreted in the given instance. Reference
7·3 THE MEANING OF DIFF:BlU!NT KINDS OF COHBSION 309 is a potentially cohesive relation because the thing that serves as the source of the interpretation may itself be an dement of text. A reference item is one whose interpretation is determined in this way. The interpretation takes one of two forms. Either the reference item is interpreted through being IDBNTIFIED WITH the referent in question; or it is interpreted through being COMPAli.ED WITH the referent - explicitly not identified with it. but brought into some form of comparison with :it. In the former case, where the interpretation involves identifying, the reference item functions as a Deictic and is always specific. Deixis is the identifying function in the nominal group; and fOr cohesive purposes the identification must he specific. Hence the set of reference items includes all the specific deictics (pronouns and determiners) except the interroga- tives. The interrogatives cannot he cohesive since they contain only a UQUEST FOR. specification, not the specification itsd£ Personal Demonstrative Exis- Possessive tenrial Refer- I~ you, mine, yours, ours my, your, our the ential we, he, his, hers, (its). his, her. its, their this~ these she,U. theirs • that, those """ s they.one Inter- who whose whose whkh, what rogative whpt Specific pronouns Specific determiners In other words. all :reference items of this type are specific, because their interpretation depends on identity of reference. This does not imply that the referent, where it is itself an element of the text (i£ where the reference is anaphoric), must necessarily also be specific. A reference item can relate anaphorically to any dement whether sp~c or not; for example {7:1I) I can see alight. Let's follow it. where it refers ro a light. But the specificity is conferred by the reference rdation. Since this involves identity, • a light' thereby becomes • the
310 THE MP.ANING OF COHESION light', ie the IJght that was just mentioned- and specified in the process., so here • the 1ight that(I've said) I can see'. This is why it is possible to have sentences such as: [7:xz] Nobody ever believeshe'sgoingtolose. where he means the person being considered as an instance, here 'the person whose belief is in question'. In this case the presupposed item is not only non-speci:6c; i.t Js also being sa:id to be non-existent. A considerable amount is now known about the rules of pronominaliza- tion, in the sense of personal reference within the sentence; but this is not a cohesive phenomenon and lies outside our scope. The question of the interpretation of reference items in oontexts of potential ambiguity has also begun to be studled, .and this, though not our primary concern here, does need to be briefly mentioned. Here the question is, how does the listener or reader identify which of two or more possible items in the text a reference item refers to. For example if we come across: a sentence ruch '-' [7:13] Spurs played Liverpool. They beat them. how do we know who beat who? Various grammatical criteria have been proposed,_ in terms of transitivity or of mood; suggesting that a reference item will preserve the structural function of its referent on one o-r another of these dimensions. For example, if transitivity is the determining factor~ a reference item functi.nning as Actor will refer to just that one among the possible referents that has the Actor function. If mood is the determining factor~ a reference item functioning as Subject wiU refer to just that one among the possible referents that has the Subject function. Example [T .14] satisfies both transitivity and modal criteria: 7"4] The chased the robbers. They caught theDL cops (transi- Process Go.! Proc= Go.! tivity) (mood) Sub- Pred.i- I Complement Subject Predi- Complement ject cator ea tor Here they refers to the cops and them to the- robbers. But consider [7: 15] The cops chased the robbers. They eluded them.
7•3 THE MEANING OF DIFFERENT KINDS OF COHBSJON 3II Here the only possible interpretation is that they refers to the robbers arul them to the wps~ this involves a reversal of the roles in both structures - and yet we find no difficulty in interpreting it. Similarly: [ . 61 J bn d B·lt' h { a. But he wouldn't giveit to bim. 7.1 o wante 1 s orse. b B he ldn' h' r · . ut wou t pay tm tor Jt. It i~ clear that (a) and (b) require opposite interpretations. In (a), he is Bill and him is John. whereas in (b) it is the other way round. There is no feeling tlut either is more acceptable, or more cohesive, than the other. Since reference is a semantic relation. the criteria are to be found not in the grammar but in the semantics. It is the meaning that enab1es us to disambiguate in such instances. If there is a grammatical tendency to be had recourse to in those instances where the meaning does not resolve the problem, it is likely to lie. as Hasan suggests elsewhere, neither in trans- itivity nor in mood but in theme. This again is to be expected. since it is the thematic structure which is the text-forming structure in the clause (see 7.4-1 below). The particular combination of circumstances that-is required i.n order to produce an ambiguous reference item in pcecisdy the kind of environment where transitivity, mood and theme are all incon- gruent with each other is so odd that no example of it can he very con- vincing; but here is an attempt: f7:I7] I. These the were given their porue> children by gnndparen"- (transitivity) (mood) Subject (theme) ii They" re staying here~ now. In the second sentence. they is Actor (in transitivity). Subject (in mood) and Theme fm the-me). In the first sentence, Actor. Subject and Theme are all different items: the Actor is their grandparents, the Subject is the chibiren and the Theme is tkse ponies. It seems that, if anything. the preferred interpretation of they is these ponies - and that in spite of the preference of English for human Subjects. If this is so. it suggests that, to
312 THE. MEANING OF COHES10N the extent that there is any grammatical criterion at all, it will be- found in the them:e structure rather than in the transitivity or modal structure. Note that this does not apply to substitution. If the first sentence had been followed by the question Which ones? or elliptical Which?. the more likdy interpretation would have been which grmulparents l It seems therefore that no dearcut grammatical rules can be given fOr assigning a .reference item to one among a number of possible text referents~ since the assignment is typically made on semantic grounds. If there is more than one referent for the identification of he or it or this. the referent is the one that makes most sense in the context. This is not to say that ambiguity cannot arise; it can. and not infrequently does. There may he no dearly predominant candidate for the status of£ making most sense'; and in that event, as a last resort, we may appeal to the grammar - probably to the fheme-rheme structure. 0rHER TIHNGS BEING EQUAL, it seems that the most probable target of a cohesive reference item is the Theme of the preceding sentence. This seems to hold even if the reference item is not itself thematic; compare: (7:r8] These ponies the children were given by their grandparent<. Have you seen them.? where them still seems more likely to refer to these ponies . .But given the range of POSSJBLE targets in a connected passage, it is unlikely that any purely grammatical principles could suffice for resolving the issue. and the semantic principle of 'making most sense • • difficult as it may be to make explicit, is the only one that could really be expected to apply. As regards restrictions on reference, these again are not our main concern; we are concerned with what does happen. not with what does not. But these Rnd also to re8.ect semantic considerations -often ones that are reflected in the grammar also. Here is just one example: [7:19] a. An old. man came in with his son. }Th dirty b. An old man came in with his overcoat. ey were very · The second sentence is acceptable following (a) but not, or at best doubtful, following (b). Old tnmt and overcoat are too different to be brought within the same presupposition; and this is related. to the fact that they cannot be coordinated: an old man ami bis son C4nte in, but not an old man mu/ his cvacOdt came in. There are instances where a reference item is used when strictly spahng the relation is not one of reference. An example will illustrate this: [7:20] Arthuc's very proud ofhis chihuahuas. I don,.t like them.
313 This is ambiguous; it could :m.ean 'I don't like Arthur's chihuahuas', or it could mean "I don'tlike chihuahuas(in general; cf: I don't like tke things)·. The second meaning is anomalous; them is not roreferential with the nomina) group Atthur' s chihuahuas. It is no doubt to be explained as being ooreferencial with the noun Head cbilmahuas taken on its own. without the Modifier. Compare in this regard the comment on [3:52] in 3.2.6 above; and also [7:4e), 1 expect you will get this but I'H send it if you want, where it and this both refer to Staff Bullain no z, considered as an object since it is Goal of the verbs get and senJ). but they refer to different copies. Finally there are instances where the reference item, because of its specificity, serves to disamhiguate preceding sentences that otherv\<-ise in themselves are ambiguous; for example (7:2I] f d rather like to see a pJay. It"s at the Ambassador's. Here the it shows that the meaning of the first sentence is 'thereys a play r d like to see •. The context of such ambiguities is very often of the kind illustrated by [7: 21}, namely a clause that is structured as a simple propo- sition but which is in fact incongruent. The congruent form of expression here would be a clause of the IDENTIFYING type, one with an equative structure sllCh as There's a play r d rather like to ~. Comparison differs &om the other forms of reference in that it is based not on identity of reference but on non-identity: the reference item is interpreted~ not by being identified with what it presupposes~ but by being compared with it. The expression • non-identity • is actually misleading~ because one possible form of comparability is identity. But tbe identity is not the criterion; being identical is just one of the ways in which two things may be like or unlike each other. In the comparative type of reference. the presupposed element takes on the role of a reference POINT. It serves as a standard, to which something dse is referred in terms of its likeness, in general or particular; and 'the same' is one kind of likeness. In this way the comparison provides the source of interpretation for the reference item; and where the presupposed dernent is also in the text, there is cohesion between the two. For example, more presupposing oysters in [7:22] 'I like the Walrus best,.' said Alice-: •because, you see, he was a little sorry for the poor oysters •. 'He ate more than the Carpenter though; said Tweedledee. When likeness takes the value of sameness, comparison resembles other
314 THE MEANING OP COH11SION forms of reference in being specific: 'same' implies • the same •. For this reason same and other comparatives of identity are typically accompanied by the, or some other specific determiner . .By contrast to this, when like- ness takes the value of non-identity (similarity or difference), the reference is typically non-specific; and comparatives expressing non-identity are unlike all other forms of reference .in just this respect. So we usually find the- same place but a similat' place, an other place (written as one word, another)~ adiffirenr platt'. We can summarize the meaning of reference by using the term co-INTER'l'RETATION. There is a semantic link between the reference item and that which it presupposes; but this does not mean that the two necessarily have the same referent. It means that the interpretation of the reference .item DEPE.'•ms IN SOME WAY on that of the presupposed. Co- reference is one particular form that eo-interpretation may take- where the two items do, in fact, refer to the same thing. But the genera! concept that lies behind the cohesive relation of reference, and by virtue of which personals, dernonstratives and comparatives are alike in their text-forming capacity, is that of eo-interpretation. A reference item is one which is interpreted by reference to something else. It is this principle of eo- interpretation that defines its role in the semantics of the text. 7·3·3 Substitution and ellipsis Wtth substitution there is no implication of specificity. The substitution relation has no connection with specifying or identifying a particular referent; it is quite neutral in this regard. So specific forms such as the empty one and non-speci£c ones such as an empty one are both equaUy likdy. The fact that the nominal substitute tme has evolved from the same source as the indefinite article might suggest that substitution is inherently non- specific; but the meaning of the substitute one is countability. not indefi- niteness. This is reflected in the fact that the plural of the substitute is ones, while the plural of the indefinite arride is some; and some is also the 'mass • furm of the article. whereas there is no form of substitute availabk in the context of a m.ass noun. We have referred already to the main distinction between substitution and nominal reference. In reference there is typically identity of referent. Substitution is used where there is no such identity. This requires a device which makes the connection at the lexicograiU!Ilatical level at what we called the level of'wording~. since the cohesion takes the form of'same element in the language (same -wording) but different referent'. The
3 I) essence of substitution therefore is contrast: a new referent is being defined - and hence there is no substitution for proper names. The contrast is not necessarily in the reference; it may be in some interpersonal element in the meaning- the modality, key. attitude etc- but the principle is the same. Reference implies that there is identity of meaning between the presupposing item and that which it presupposes. while substitution lmplies non-identity of meaning. This is illustrated by the use of substitu- tion and ellipsis in responses; the function of a response is to supply missing information. or confirmation- that is to supply something that is New. and it is this that provides the environment in which the substitute or dliptical item occurs. For example. (7:23] Did you cook the dinner?- No;John did. The distinction between reference and substitution or ellipsis is however less clearcut with verbs and clauses than with nouns. Note the difference between (7:24" and b]: {7:z4] Are they selling the contents today ? a. No. they're doing it tomorrow. (reference) b. No~ they are (doing) tomorrow. (sub- stitution or dlipsis) The first, being referential. makes it an assumption that they are selling. and merely supplies the time; it parses the question as 'when are they selling the contents? •. and has a them..atic structure of the identifying type~ it is equivalent to 'the time when they're selling=tomorrow', with the verb embedded in the Theme. An alternative form for (a) would be It's tomorrow they're doing it. The second does not assume the selling but states lt, because the meaning which provides the contrastive environment far the substitution - namdy the polarity - is New. The substitute form of the answer parses the question as • are they selling the contents?" with 'today • either as given or as additional rdevant informa- tion; its thematic organization is •the fact=that they are selling; but tomorrow'. For this reason in a question-answer sequence with no possible fix:us other than the polarity. the reference form is not an appro- priate form of answer; the following exam_ple shows this: [7: 2.5] a. Are they selling the contents?- Yes, they're doing it. -No. they're not doing it. The substitute or elliptical forms of the answer, on the other hand, would be entirely appropriate:
316 THE MEANING OF COH.BS!ON [7:25] b. Are they selling the contents-?- Yes, they are{doing). -No, they're not (doing). Conversely. when the sense requires that the focus is elsewhere (because the process itsdf is not in question, but only the circumstance -locative, temporal, etc- with which it is associated) the substitute or el1iprica1 form is ruled out. [n [7:26}, fm: example, the answer presents 'I. sleep' as if it was new infor1ru1tion, and hence is rather odd: [7:26] Do you sleep on the couch?- No; I do (do) on the sofa. Here the reference is also ruled out, but for a different reason; we do not say I do it on the scfa because sleep is not a verb of action. But with other types of process. reference would be acceptable: Do you c()()k t>Vef1 day? - No, I Jo it every other iky. Two final examples: ! D .h . {a. No, she does it for pleasure. {reference) 7:2.7 r oes lie rtnt b. No; she does. (do) for pleasure. (substitution ror pro t. lh . ) ore pm Here (b) is unlikely because it presents 'she paints" as new information, whereas the form of the question suggests that the fact that she paints is not at issue; the appropriate meaning is rather 'the reason she paints is for pleasure', as expressed in (a). This illustrates the general principle of substitution and dlipsis, with their meaning of • continuity in the environment of contrast •. What is carried over is a FOllM. a word or structural feature; and this happens in an environment where the referential meanings are not identicaL The structural environment, on the other hand, tends to remain fairly constant; examples such as [7: 28a and b] are un1ikely because they involve too great a structural shift: (7:28] a. Would you like this book to read?- I've already done so. b. Give me a hook to read. I have (done) this one. whereas following Read this book! in (a). and Have you read these books? in (b), the rubsritute forms would be quite unexceptionable. Why does the speaker not simply repeat the same word? He can do. of course: [7: 29] I've had an offer for this.-rn make you a better offer. But notice what happens. In order to signal this as a reiteration (and if it was not a reiteration it would not be cohesive). the speaker has to shift the tonic away from t1fer on to better. :But ojfer is a lexical item: hence the
JI7 placing of the focus on an earlier item is 'marked • and strongly contrastive, a function of the systent.ic opposition between [7:3oa and b]: [7:3o) a. I'll make you a better OFFEll(ummrkerl focus) b. I'll make you a BETTER offer(nurkedfocus) Yet this opposition is irrelevant in the context of {7: 29]. There is only one possible meaning here, not two; the context requires the fOcus on better. hut it also requires that it should be an unmarked focus. and this can be achieved ouly by the use of a structure in which there is no lexical item following better, ro that bettr:r either is the last word in the information unit {ellipsis: I'll make you a better) or is followed on1y by a grammatical word,. one which does not carry information focus (substitution: I"ll make you a better one). Both these have urunarked focus. [n other words, the substitute and elliptical forms are preferred because they create cohesion without disturbing the information structure of the discourse: without assigning prominence of a kind which is irrelevant in the given environ- ment. Between substitution and ellipsis the difference in meaning is minimal We defined ellipsis as substitution by zero; we could equally well have defined substitution as explicit ellipsis. Ellipsis is characteristic particularly of responses: responses to yes/no questions, with ellipsis of the proposition (No he tlidn~t; Yes I have, etc), and to WH- questions, with ellipsis of alJ elements hut the one required (In the drawer; Next weekend, etc). But whereas there is a significant difference in meaning between elliptical or substitute forms on the one hand and the corresponding 'filled out' forms on the other, there is hardly a significant dia'erence between the two cohesive forms themselves. For example, [7:3I] Let'sseeifGrannycan1ook after the shop for us. a. She MIGHT look after the shop for us. hi. She MIGHT do. bii. She .MIGHT. As we saw in the last paragraph~ (a) differs from (b) in that it makes explicit the Given element look after the shop for- us, and in doing so imposes a marked information structure in which MIGHT look cftn iN shop J(IT us is specifically contrasted with might look 4}ier- THE SHOP for us; whereas in (b) the distribution of information i.s neutral - the tonic fills in its unmarked place. This is dearly a meaningful choice on the part of the speaker. But between (bi) and (bii) there is hardly any difference in the meaning. There are many contexts in which ouly one or the other is
]I8 THH MEANING OF COHESION possible; for numerous speakers of English, for example, only (ii) could. occur here. Where both are possible, the substitute form .appears slightly more explicit in its sense of'same form in a different environment": {7: J3] Has Smith reacted to that paragraph about him in the paper? -No he hasn•t. -He hasn't done yet; but when he reads it carefully he may think again. (7:33) Have an apple.- I'll take this.- The other one·s better. The use of do in [7:32} and one in [7:33] suggests in each case .a somewhat more pointed contrast than would be achieved by the elliptical form he hasn't yet, the other•s better. And a clausal substitute may serve to dir ambiguate in certain reported speech contexts: [7:34) Will Granny look after the shop for us?- She ham't said. The elliptical form may mean either •she hasn't said that she will" or 'she hasn't said whether she wilJ or not', whereas the substitute form she htun't said sg could only mean the former . .But in many instances the distinction between substitution and dlipsis is scarcely noticeable, and c;m be treated. fur practical purposes as a matter of free variation. 7·3·4 Lexical cohesion: reiteration and wUocation Lexical cohe5ion is 'phoric: • cohesion that is established through the structure of the LE XIS, or vocabulary. and hence (like substitution) at the lexicogrammatic:allevel. To recapitulate this point: Linguistic level at which 'phoric' rdation is established Semantic Lexicogrammatical { Grammatical Lexical Type of cohesion Reference Substitution and. ellipsis Lexical cohesion Lexical cohesion embraces two distinct though related aspects which We referred to as REITERATION and COLLOCATlON. I. Reiteration. This is the repetition of a lexical item, or the occurrence of a synonym of some kind, in the context of reference; that is, where the
7-3 THE ME.ANING OF DIFFERENT KINDS OF COHESION 319 two occurrences have the same referent. Typica11y, therefore, a reiterated lexical item is accompanied by a reference item, usually the or a demon- strative. The complex consisting of the plus reiterated lexical item is therefore cohesive hy reference. But since reiteration is itself cohesive in its own right. as shown by the fact that cohesion takes p1ace even where there is no referential £elation (cf next paragraph). such instances constitute a double tie and are interpreted here in this way {see Chapter 8, 8.J and note 2 to Text r). 2. Collocation. As remarked above, the repetition of a lexical item is cohesive in i~ own right, whether or not there is identity of reference, or any referential relation at all between the two. The cohesion derives from the lexical organization of bnguage. A word that is in some way associated with another word in the preceding text, because it is a direct repetition of it, or is in some sense synonymous with it, or tends to occur in the same lexical environment, coheres with that word and so contri- butes to the texture. The following passage contains illustrations of both these types: [7: 35] Soon her eye fell on a little glass box that was lying Wider the table: she opened it, and found in it a very small cake. on which the words • EAT ME' were beautifully marked in currants. 'Well, f11 eat it/ said Alice, 'and if it makes me larger, I can reach the key; and if it makes me srn.aJler, I can creep under the door; so either way rn get into the garden, and I don't care which happens!' She ate a little bit, and said anxiously to herself. 'Which way? Which way?' holding her hand on the top of her head to feel which way it was growing, and she was quite surprised to find that she remained the same size: to be sure, this generally happens when one eats cake, but Ali.ce had got so much into the way of expecting nothing but out-of-the-way things to happen, that it seemed quite dull and stupid for life to go on in the common way. So she set to work. and very soon finished off the cake. The second occurrence of cake, in when one eats cake (second paragraph), is without reference item; there is no referential link with the first occur- rence, but the repetition i.tsdf constitutes a tie. The third occurrence, in very soon finishd off the cake, is with a reference i.tem; here, therefore. there are two ties., one of reference. the referential identity being sho'Wl1 by the, and one of reiteration. Other instances of lexical cohesion i:n the
320 THll: MEANING OF COHESION passage are provided by eat .•. eat ... ate ..• eats; open .•• key .. , ~kM; larger , , • smaller .•. {the same} size; mokes larger ... makes sm.1ller ... gwwing; MJ>PMS •.• happen. The principle behind both types is the cohesive effect achieved by the continuity of lexical meaning. This may be combined with a referential relation but does not depend on this for its eifect. The cohesion ls a func- tion of the relation between the lexical items themselves. which has both a semantic aspect- synonymy~ hypo-nymy, metonymy, etc- and a purely lexical o:r oollocational aspect. the mutual expectancy between words that arises from the one occurring frequently in the environment of the other, or (a better way of !ook.ing at it} of the two occurring in a range of environments common to both. The whole of the vocahu1ary of a language is intemaiiy structured and organized. along many dimensions~ which collectively determine 'what goes with what'; these tendencies are as much pan of the iinguistic system as are the principles of grammati- cal structure, even though they are statable only as tendencies. not as •rules'. It is the essentially probabilistic nature of lexical pattetning which makes it effective in the creation of texture; because they lie outside the bounds of structure. and are not consttained by structural £elation.ships~ the lexical patterns serve to transform a series of unrelated structures into a unified. coherent who~e. 7·3·5 ConjunctWn Conjunction is somewhat different from the other cohesive relations. It is based on the auumption that there are in the linguistic system forms of systematic relationships. between sentences. There are a number of pos- sible ways in which the system allows for the parts of a text to be con- nected to one another in meaning. There are certain elementary logical relations inherent in ordinary language; doubtless these derive ultimately from the categories of human experience. and they figure importantly in rhe sociolinguistic construction of reality, the process whereby a model of the universe is gradually built up over countless generations in the course of semiotic interaction. (They can be regarded as departures from the idealized norm represented by formal logic; but it is worth remembering that in the history of human thought the concepts of formal logic derive. however indirectly, from the logic of natural language.) These logical relations are embodied in linguistic structure, in the form of coordination, apposition. modifica- tion, etc. Analogous to these are certain non-structural, text-forming
7·3 THB MEANING OF DIFFJlRRNT KINDS OF COHESION 321 relations which are what we are calling conjunctive relations. Conjunctive relations are encoded not in the fOrm of gram:matical structures hut in the looser~ more pliable form oflinkages between the components of a text. The specific cor.yunctive relations are those of 'and', 'yet', 'so' and 'then'; and each of these may occur in either an 'external' or an 6intern:al' context. The latter distinction, which derives fmm the functional basis of the semantic s~tem, determines the Jocus of the cozyunction; the cot9un.ction may be located. in the phenomena that constitute the content of what is being said (external), or in the interaction itself: the sociaJ pro- cess that constitutes the speech event (internal}. Here is a further set of exampJes of each. 'and' 'yet' 'so' • then' External Internil They gave him food and They gave me fish to clothing. And they looked eat. And [ don't like .after him till he was fish. better. They looked after him well. Yet he got no better. He drove into the harbour one night. So they took his licence away. He stayed there for three years. Then he went on to New Zealand. That must he Henry. Yet it can't be; Henry's in Manchester. We're having guests tonight. So don't he late. He found his way eventually. Then he'd lefr his papers behind. CotYnnction does not depend either on referential meaning or on identity or association of wording. Conjunctive relations are not • phoric.,; they represent semantic links between the elements that are constitutive of text. There are numerous possible ways of interpreting conjunctive rdations;. the fourfold scheme we have adopted here is simply the one we have fOund most helpful in the quest for a general characterization of cohesive relations which would not be • closed., - which would allow further subdassification as and when needed. A purely structural approach would suggest other modes of classification. based for example on the traditional categorization of subordinate clauses. As already- noted, there are structural analogues of the conjunctive relations; here are some examples of the way each is expressed:
322 THB MEANING OP COHESION Structural Qogical) Textual (conjunctive) paratactic hypotactic ' and' Alw, ... . .. and ... besides .•. ' yet' However,, .. although ••. . .. yet ... '•o ' Consequently~ ... because ... . . . so •.. ·then~ Subsequently~ ... ... then ... after ••• (ht the same way.,. the • phoric' relations of reference, and substitution and ellipsis. are also found as structure-forming relations within the sentence.) But from our present standpoint it is the nature of text, rather than the organization of grammar, that has determined the interpretation and presentation of the systems involved. 7.3.6 Summary The semantic basis of cohesion in English texts can be summarized as follows (and if the Tabk at the end of 7·3 above). Cohesion consists ( r) in continuity of lexicogrammatical meanmg ('relatedness of form';. phoric) U!XICAL COHESION SUBSTITUTION BlllPSES collocates reiterations (repetitions and synonyms) superordinates general terms ,---------1---------. substitutes ellipses clauses ---- groups{nominal}- words verbal Substitution and ellipsis are relevant especially in the environ- ment of discontinuity of reference.
7·3 THJ! M.BANING OF DIFFER.BNT KINDS OF COHESION J2J (2) in continuity of referential meaning (' relatedness of reference': phor;c) personal (communication :role of referent) RlwBRENCE demonstntive (proximity of referent) comparative (similarity to preceding referent) (3) in semantic connection with the preceding text (non-phoric) additive adversative CoNJUNCTION j causal ;n {ideational meaning (external) terms of interperscnal meaning (internal) temporal These are the cohesive relations. In categorizing them in this way. it is perhaps useful to add a reminder o£ the difference betv.reen the cohesive relations themselves and the means by which they are represented in the linguistic s.ystem... The COHESIVE lUU.ATIONS THEMSELVES can be interpreted as being either lexicogrammatical in nature (1} or semantic, the ~tter being either refcrential (2) or colliunctive (3). The type of cohesion, in other words, is either one that depends on semantic relations m the linguistic system or one that depends on leximgrammatical relations. :But the BXPRESSION of cohesive relations involves both the semantic and the lexicogrammatical systems in all cases: that is, both choices in mean- ing. and their realization in words and structures. Thus even where cohesion is achieved through the setting up of a purely formal relationship in the text. such as the substitution of one for the noun expressing the Thing, the CHOICES that are involved, not only in Lhe selection of the particular thing-meaning itself but equally in the identification of it with a preceding thing-meaning. are semantic choices. And conversely. even where the cohesive relationship is a semantic one. it has to be realized. in the lex:ioogrammatical system; for example. identity of referential meaning as expressed through the grammatical system of third person pronoWIS. Here is a final summary table:
324 THB MEANING OF COHESION RepesentaJWn in linguistic systrm Semantic Type'![ cohesive Telation Conjunction Additive. adversative. causal and temporal relations; cxtemal and internal Reference Identification: by speech role by proximity by 'P"cificity (only) Reference point Lexical cohesion Collocation (similarity of lexical environment) Reiteration (identity of lexical reference) Substitution Identity of potential reference (class meaning) in context of non- identity of actual (irutantial) reference 7.4 Cohesion and the text Lexicogrammatical (typically) Dis~Coucse adjnncts: adverbial groups. prepositional groups Personals Demonstratives Definite article Comparat~es Same or associated lexical item Same lexical item; synonym; superordinate; general word Verbal. nominal or clausal substitute Verbal, nominal or clausal ellipsis Texture involves much more than merely cohesion. In the construction of text the establishment of cohesive relations is a necessary component; but it is not the whole story. In the most general terms there ace two orher CQmponents of texture. One is the textual structure that is internal to the sentence: the organiza- tion of the sentence .and its parts in a way which relates it to its environ- ment. The other is the • macrostructure' of the text. that establishes it as a text of a particular kind ~ conversation, narrative, lyric. commercial correspondence and m on.
7·4 COHBSJON AND THJI TEXT 325 7·4·1 Texture within the sen.tenu: T'he main components of texture within the sentence .in English are the theme systems and the information systems (if the summary at the end of Cb.pter I). These have been outlined in anartide by Hatliday.* The theme systems are those concerned with the organization of the clause as .a meSS<~ge: its structure in terms of a THEMB and a remainder (known as the RHEME), and a wide range of thematic: variation that is associated with this structure in one way and another. The following examples give an idea. of the semantic range that is involved: [7:36] a. Jolm', annt I left him tlm duckpress Rheme b. Jolm I was left thi, duckpress by hi' aunt Rheme c. This duckpress I John's aunt left him 1heme Rlreme d. What John's aunt left him I was this duck press Theme: Identified Rheme: Identifier e. The way John got t.his duckpress was by a legacy from his aunt Theme: Identified Rlreme: Identifier £ Bequeathing this dud:press was what John•s aunt did for him Theme: Identifier Rheme: Identified The information systems are those concerned with the organization of the text into units of information. This is expressed in English by the intonation patterns, and it is therefore a feature only of spoken English. In written English. punctuation can be used to show information struc- ture. although it cannot express it fully. and most punctuation practice is a kind of compromise between information structure (punctuating according to the intonation) and sentence structure (punctuating according to the grammar). The intonation of spoken English expresses the informa- tion structure in a very simple way. Connected speech takes the form of an unbroken succession of intonation units, or TON.E GltOVPS as they are generally called; and each tone group represents what the speaker chooses to encode as one piece of infOrmation, one unit of the textual process. • 'Notes on traruitivity and theme: iu &glhh'. Put :z,J-nwl of .l..itJgJmtics J, t9()7.
326 Each information unit is then structured in terms of two elements. a NEW element, expressing what the speaker i.s presenting as information that is not recoverable to the hearer from other sources; .and a GIVEN elenrent. expressing what the speaker is presenting as information that is recoverable to the hearer from some source or other in the environment - the situa- tion. or the preceding text. The Given dement is optional; the New is present in every information unit, since without it there would not be a separate information un:it. In the following examples. the information unit bonndaryis shown by// and the New element is printed in SMALL CAPITALS: (7:37} a. 1/ JOHN's AUNT UFr IDM THIS DUCKPRESS // b. //JoHN's AUNT lefr him this duck press// c. f/ JoHN'S At=NT //left him TIDS DUCXPRESS fl d. I/ John WAS l.llFr THIS DUCKPRllSS I! by HIS AUNT 11 e. i/ JoHN v.= left this duckpr= 11 by ms AUNT 11 f. //JOHN /1 W~o\S LBFT THIS DUCKPIESS BY InS AUNT lf g. If THis DUcKPREss J f JoaN' s A t.'NT left him 1/ The number of possibiiities is very large indeed, and the combination of thematic systems with information systems gives a paradigm which. with a clause of average length, runs into the tens. and hundreds of thousands. Since each one has a different textual meaning this might seem 1mm:mage- .ably complex - until it is realized that this enormous number of different textaal structures within the sentence is the result of combining a nwnber of related hut independent choices each one of which is by itself very simple. If there are only twenty different choices, each of only two possi- bilities, this already yields over a million forms. In bet, of ooun:e, things are not quite as simple as that; the number of possibilities depends on the structure of the sentence, the choices are not fully independent. so that not all the theoretically possible combinations occur. and not all choices are limited to two options. But it is this principle on which the sentence is structured internally in its role as the realization of text; and this internal texrure is the structural counterpart of cohesi.on. Neither cohesion alone, nor internal textual structure alone, suffices to make of a set of sentences a text. Texture is a product of the interaction between the two. 7.4.2 The texture <>/discourse The third and final component of texture is the structure of discourse. By this we mean the larger structure that is a propeny of the forms of dis-
7·4 COHESION AND THE TEXT 327 course themselves: the structure that is inherent in such concepts as narrative~ prayer, folk-ballad, formal correspondence, sonnet, operating instructions., television drama and the like. 1t is safe to say that every genre has its own discourse structure. It might seem as if informal, spontaneous conversation had no structure of its own over and above the internal organization of each sentence and. the cohesion betv.-een the sentences. But the work of Harvey Sacks and Emanuel Schegloffhas shown beyond question that conversation is very highly structured. There are definite principles regulating the taking of turns in conversation, and one of the functions of some of the items operating cohesively as conjunctives (Chapter 5) is that of marking and holding turns. There are several types of what Sacks and Schegloff call • adjacency pain •. ordered sequences of two elements in a conversation that are related to each other and mutually presupposing, like greetings. invitations, or question-answer sequences. The discourse structure of <~ conversation is in turn reinforced by the cohesion, which explicitly ties together the related parts, bonding them·more closely to each other than to the others that are not so related; hence Goffman's observation that 'there tends to be a less meaningful rdatiomhip between two sequential interchanges than between two sequential speeches (ie turns) in an inter- change' (Imer.a;on R;tual, p 37). Other forms of discourse are more obviously structured than conversa- tion; and some, notably narrative, have been studied in considerable detail in a variery of different languages. There is no need here to labour the point that the presence of certain elements,. in a certain order, is essential to our concept of narrative; a narrative has. as a text, a typical organization, or one of a number of rypical organizations, and it acquires texture by virtue of adhering to these fonru:. Literary forms, including the • strict' verse forms - culturally established and highly-valued norms such as those of metre and rhyme scheme, ddining complex notions such as the sonnet~ iambic pentameter blank verse, and the like - all fall within the genera1 category of discourse structures. They are aspects of textUre, and combine with intrasentence structure and intersentence cohesion to pro- vide the tota1 text-forming resources of the culture. 7·4·3 The rok rif. linguistic analysis The linguistic analysis of a text is not an interpretation of that text~ it is an explanation. This point emerges clearly. though it is often misunder- stood, in the context of stylistics, the linguistic analysis of literary texts.
328 The linguistic analysis of literature IS not an interpretation of what the text means; it is an explanation of why and how it means what it does. Similarly, to the extent that linguistic analysis is concerned with evalua- tion. a linguistic analysis of a text is not an evaluation of that text; it is an explanation of how and why it is valued as it is. A linguistic analysis of a literary text aims at explaining the interpretation :and evaluation that are put upon that text. The role of linguistics is to say how and why the text means what it does to the reader or listener, and how and why he evaluates it in a certain way. This point can be genetahzed to the study of texts as a whole. The analysis of cohesion, together with other aspects of texture. will not in general add anything new to the interpretation of a text. What it will do is to show why the text is interpreted in a certain way; :including why it is ambiguous in interpretation whereVer it is so. It will explain the nature of conversational inferences, the meanings that the hearer gets out of the text without the speaker having apparently put them in -presuppositions from the culture, from the shared experience of the participants, and from the situation and the surrounding text. It is the text-forming or • textual' component of the semantic system that specifically provides the linguistic means through which such presuppositions are made. Similarly the analysis of cohesion will not tell you that this or that is a good text or a bad text or an dfective or ineffective one in the context. But it will tell you something of '\VHY YOU TIDNK it is a good text or a bad text, or whatever you do think about it. It is in this perspective that in the final chapter we suggest means for describing the cohesive patterns of a text. The intention is to provide for a reasonably comprehensive picture of this aspect of texture; and in t:1ris way to offer an insight into what it is that makes a text a text.