Chapter 5 (conjunction) is on the borderline of the two; mainly gram-

Chapter 5 (conjunction) is on the borderline of the two; mainly gram- matical, but with a lexical component in it. The distinction between grammatical and lexical is really only one of degree. and we need not make too much of it here. It is important to stress. hov.""eVer, that when we talk of cohesion as being • grammatical or 1exical', we do not imply that it is a purely formal relation, in which meaning is: not involved. Cohesion is a semantic relation. But, like all components of the semantic system, it is realized through the lexicogramm.atical system; and it is at this point that the distinction can he drawn. Some forms of cohesion are realized through the grammar and others through the vocabnlary. We might add as a footnote here that certain types of gram.ma.ti.cal co- hesion are in their: turn expressed through the intonation system. in spoken Eng)ish. For example. in (~:6] Did !hurt your feelings1I didn't mean to. the second sentence coheres not only by ellipsis. with I didn~ t mean to pre- supposing hurt your feelings. but also by conjunction, the adversative mean- ing 'but' being expressed by the tone. Phenologically this would be: fi.>. did I I hurt your 11'1!EUNGS H 4 A I I didn't I MEAN I to // the second sentence having the rising-falling tone 4· For an explanation of the intonation system, see section S-4 and the references cited there. r .2 Cohesion and linguistic structure 1.2.1 Texture aru1 structure A text, as we have said, is not a structural unit; and cohesion, in the sense in which we are using the term, is not a structural relation. Whatever rela- tion there is among the parts of a text- the sentences~ or paragraphs. or turns in a dialogue- it is not the same as structure in the usual sense. the relation which links the parts of a sentence or a clause. StructUre is, of course, a unifying relation. The parts of a sentence or a clause obviously' cohere' with each other. by virtue of the structure. Hence they also display texture; the elements of any structure hav~ by definition. an internal wllty which ensures that they all express part of a text. One

I.2 COHESION AND LINGUISTIC STRt.TCTUJtB 7 cannot change text in mid-sentence, so to speak; or rather .. if one does, there will always: be a break in the structure. with something being inter- polated which is not structurally a part of the same sentence~ as in Hamlet's [1: 7] Then I will come to my mother by and by- they fool me to the top of my bent- I vvill come by and by. or, more conversationally, [I:8] •.. But what I want-to know is-yes. some ice,_ please- what this government think they're doing when they spend all that money on building new schools. What's wrong with the old ones? In general. any unit which is structured hangs together so as to form text. All grammatical units - sentences, clauses, groups. words - are internally • cohesive • simply because they are structured. The same applies to the phonological units, the tone group, foot and syllable. Structure is one means of expressing texture. If every text consisted of only one sentence, we should not need to go beyond the category of structure to explain the internal cohesiveneu of a text: this could be explained simply as a function of its structure. But texts are usually not limited to one sentence; on the contrary, texts consisting of one sentence only are faidy rare. They do exist; there are public notices, proverbs, advertising slogans and the like, where one sentence by itself comprises. a complete text, for example [:r;g] a. No smoking. b. Wonden never cease! c. Read The Herald every day. But most texts <'<tend well heyQild the «mfines of • single S<Dtence. In other words. a text typically extends beyond the range of structural relations, as these are normally conceived o£ But texts cohere; so cohesion within a text- texture- depends on something other than structure. There are certain specifically text-forming relations which cannot be accounted for in terms of constituent structure; they are properties of the text as such. and not of any structural unit such as a clause or sentence. Our use of the term COHESION refers specifically to these non-structural text-forming relations. They are. as we have suggested. semantic relations. and the teXt is a semantic unit. 1.2.2 Cchesicn within the smtence? Since cohesive relations are not concerned with structure. they may be

8 found just as wdl within a sentence as between sentences. They attract less notice within a sentence, because of the cohesive strength of gramm.atic.al structure; since the sentence hangs together already, the cohesion is not needed in order to make it hang together. But the cohesive relations are there all the same. For example [r;xo] If you happen to meet the admiral~ don't tellhimhisship's gone down. Hcre the him and his in the second b.lfhave to be decoded by reference to tlre atlmiral, just as- they w-ould have had to be if there had been a sentence boundary in between. Similarly: [I :I r] Mary promised to send a picture of the children, but she hasn't done. Here done equals sent a picture of tlut children~ and it is quite irrelevant to this whether the two are in the same sentence or not. Cohesive relatiQnS have in principle nothing to do with sentence bound- aries. Cohesion is a semantic relation between an element in the text and some other dement that is crucial to the interpretation of it. This other element is also to be found in the text (cf 1.2-.4 below); hut its location in the ten is in no way determined by the grammatical structure. The two elements, the presupposing and the presupposed, may be structurally re- lated. to each other, or they may not; it makes no difference to the meaning of the cohesive relation. However, there is a sense in which the sentence is a significant unit for cohesion precisely because it is the highest unit of grammatical structure: it tends to determine the way in which cohesion is EXPRESSED. For example, if the same entity is referred to twice within the same sentence, there are rules governing the form of its realization. These are the rules of pronominalization. It is the sentence structure which determines. within limits., whether at the second mention the entity will he named again or will be referred to by a pronorm. For example. we cannot say [r:~z] John tookJobn'shatolf and hung John's hat on a peg. Assuming that there is only one 'John • here, and only one • hat' • then this identity of reference must be expressed by the use of pronominal forms: John took his hat off and hung it on a peg. This sort of thing can be accounted for by reference to sentence struc- ture; the rdation between an item and another one that presupposes it could be explained as a stru<:ttual relation. In the preceding sentence, for

1.2 COHESION AND LINGUISTIC STRUCTURE 9 cxampk. the words one :and it both. in diflerent ways. presuppose the word item; and this presupposition could be incorporated into the structure of the sentence. But this would be misleading. Ouly certain irutances of cohesion could he treated structurally. and ouly when the two items, the presupposing and the presupposed. happened to occur within the sazne sentence. But, as we have seen, the question whether the two fall within the same sentence or not is irrelevant to the nature of the cohesive relation; cohesion is a more genera) notion, and one that is above considerations of structUre. Moreover only certain kinds of cohesive rdation are governed by such rules; mainly those involving identity of reference, which under certain conditions must be signalled by a reference item (Chapter :z.). Cohesion that is expressed through sulmitution and ellipsis (Cbapt<ors 3 and 4) is unaft'ected by the sentence structure; and so is lexical cohesion (Chapter 6). In the case of conjunction (Chapter 5), there are special forms to express the various conjunctive rdations where these are associated with gram- matical structure; eo~ [I: 13a]. which is non-structural, with its struc- tunl countetpart [I:IJh]: [r: 13} a. lt's raining.- Then let's sray at home. b. Since it's raining. let's stay .at home. Regardless of the presence or absence of .a structurallinl::. the semantic re- lation that provides cohesi~ namely that of cause, is the same in both. For these reasons cohesion withln the :rentence need not be regarded as essentially a distinct phenomenon. Cohesion is a general text-forming rela- tion, or set of such relations~ certain of which, when incorporated within a sentence structUre. are subject to certain restrictions - no doubt because the grammatical condition of'being a sentence' ensures that the pans go together to form a text anyway. But the cohesive relations themselves .are the same whether their elements are within the same sentence or not. As a general rule, the examples: cited in this book will he of cohesion across sentence boundari~ since here the etfect is more striking and the meaning is more obvious: ~ve ties between sentences stand out more clearly because they are the ONLY source of textwe, whereas within the sentence there are the structural relations as welL In the description of a text, it is the intersentence cohesion that is significanty because that rep- resents the variable aspect of cohesion. distinguishing one text from an- other. But this should not obscure the fact that cohesion is not, strictly speaking, a relation 'above the sentence •. It is a relation to which the sentence, or any other form of gram.rnatical structure, is. simply irrelevant.

.IO I .2 ·3 Cohesion and discourse stTucture It will be clear from what has been said above that cohesion is not just another name for discourse structure. Discourse structure is, as the name implies, a type of '>tructnre~ the term is used to refer to the structure of some postulated unit higher than the sentence. for example the paragraph. or some larger entity such as episode or topic unit. The concept of cohesion is set up to account for relations in discourse~ but in rather a different way. without the implication that there is some structural unit that is above the sentence. Cohesion refers to the range of possibilities that exist for linking something with what has gone before. Since this linking is achieved through relations in MEANING (we .are excluding from consideration the effects of formal devices such as syn- tactic parallelism, metre and rhyme), what is in question is the set of mean- ing relations which function in this way: the semantic resources which are drawn on for the purpose of creating text. And since, as we have stressed, it is the sentence that is the pivotal entity here - whatever is put together within one sentence is ipso facto part of a text - we can interpret cohesion. in practice, as the set of semantic resources for linking a SBNTJ3NCE with what has gone before. This is not to rule out the possibility of setting up discourse structures, and specifyingthestructureof some entity such as a paragraph or topic unit. It is clear that there is structure here. at least in cert2in genres or registers of discourse. But it is doubtful whether it is possible to demonstrate generalized structural relationshi~ into which sentences enter as the realiz- ation of functions in some higher unit, as can he done for all urfits below the sentence. The type of relation into which sentences enter with each other differs from that which holds among the part or sub-parts of a sen- tence. We cannot show~ for example, that there is any fi.mctional relation between the two sentences of [1 : I] a hove, such that the two form a con- figuration of mutually definlng structural roles. (It may on the other hand he possible to show something of the kind precisely hy in'\~king the con- cept of cohesion; if Chapter s.) Whereas within the sentence, or any similar unit, we am specify a limited number of possible structures. such as types of modiflcation or subordination, transitivity or modal structures and the like. which define the relations among the p.atts, we carmot in the same way list a set of possible structures for a text, with sentence dasses to fill the structural roles. Instead we have to show how sentences, which are structurally independent of one another, may he linked together tbrough particular features of their interpretation; and it is for this that the concept of cohesion is required.

:1.2 COHESION AND LINGUISTIC STRUCTURE 11 J.Z.4 Cohesion llS a senumtic relation To say that two sentences cohere by virtue of relations in their meaning is not by itself very precise. Practically any two sentences might be shown to have something to do with eaclt other as far as their meaning is concerned; and although in judging whether there is teXtUre or not we certainly have recourse to some feeling about how much the sentences do actually inter- relate in meaning. we could not give any very explicit account of the degree of relatedness that is needed or how it is to be measured. But there is one specific kind of meaning relation that :is critical for the creation of texture: that in which ONE ELliMl!NT IS INTERPRRTBD BY REFERENCE TO ANOTHER. What cohesion has to do with is the way in which the meaning of the elements is interpreted. Where the interpreta- tion of any item in the discourse requires making reference to some other item in the discourse, there is cohesion. Consider the example [I!I4] He said so. This sentence is perfectly intdligible as it stands~ we know what it means, in the sense that we can •decode' it semantically. But it is UNINTER- "P:RETABLE, because we do not know who ~he~ is or what he said. For this we have to refer elsewhere, to its 6 context' in the sense of what lwi gone before. Now it is also true that, given just the sentence [r;~s] John ,.;d everything. we do not know who •John• is, or what he said, either. But there is an important difference between examples [I: 14] and [ 1: I5 ]. In [I: 14]. the items he and so contain in their meaning an explicit signal that the means of their interpretation is available somewhere in the environment. Hearing or reading this sentence, we know that it links up with some other passage in which there is an indication of who 'he• is and what he said. This is not the case with John or everything, neither of which necessarily presupposes any such source of further imerpretation. We now come to the more complex pan of the picture.lt is easy enough to show that he and so are cohesive; there is no meam of interpreting them in their own right, and we are immediately aware of the need to recover an interpretation from elsewhere. There are systematically related ques- tions which express this: Who said so? What did he say? By the same token we can readily recognize the cohesive effect of a sentence such as; [1 :>6] Lying on the flooc.

12 INTR.ODU CTION Here there i:s no explicit :signal of presupposition. in the form of a word like he or 10; the cohesion is provided by what is left out. and again we can ask the relevant question Who is? Notice however that there is now some ambiguity as reg:rrds the information to be :supplied; the actual text might have been [r:r7) What was John doing when you came ln? Lying on the floor. in which case lying would have to be interpreted as UtflS lying not is lying. And there are .till further possibiliti"' ., illwtrated by' [1:18} What is your favourite pastime? Lying on tbe floor. These show that cohesion is. a relational concept; it is not the presence of a particular class of item that is cohesive, but the relation between one item and another. This point emerges very clearly with another type of cohesion, which would. otherwise be difficult to explain.. We said with reference to example [t;rs) that there is nothing presupposing about tbe item John; the sen- tence Julm said everything does not in itself confer the automatic right to ask for an interpretation of John. as he said everything does with regard to he. But we may have a sequence such as: [r:rg] I was introduced to them~ it was John Leathwall and his. wife. I had never met John before, but I bad heard a lot about him and had :some idea what to expect- Here John does have a cohesive function - because it is reiterated. This form of cohesion is lexical (Chapter 6}; it consists in selecting the same lexical item twice, or selecting two that are dosely related. The two in- stances may or may not have the same referent; but the interpretation of the second will be referable in some way to that of the first. Compare what was said aboutexam.ple [1:3) above. Another eD.mple would be: [I :20] Jan sat down to rest at the foot of a huge beech-tree. Now he was so tired that he soon fell asl~; and a leaf fell on hi~ and then another, and then another~ and before long he was covered all over with leaves. yellow~ golden and brown. Here leaf ties with beech-tree. The two are clearly not identical in reference, since tTu and kaf are not synonymous; but the intctpretation. of leaf de-

1.2 COHESION AND LINGUISTIC STitUCTUR.l!. 13 pends on beech-tue- we 'know' that the leaf was a beech-leaf, and if the sentence had continued before lcng he was covered all over with oak-leaves we should have rejected it as a mistake. This illustrates: the force of cohesion; and it also illustrates the fact that cohes.ion depends not on the presence of explicitly anaphoric items like w and Ire, but on the establishment of :l semantic relation which may take any one of various forms. One other form it may take is that of conjunction, expressed by means of items such as but, later on, in that case (Chapter s). Here the cohesion resides in an abstract relation between one proposition and another. This may be a matter of the CONTENT of the propositions. how they are rdated. to each other as phenomena; for example [1:21] First, he took a piece of string and tied it carefully round the nedr of the bottle. Next, he passed the other end over a branch and weighted it down with a stone. Or it may be a matter of their role in the discourse, how they are related in the perspective of the speaker or writer. for example [1:22] Fttst. he has no experience of this kind of work. Next, he showed no l>i.gn of being willing to learn. Here next refers to succession in the argument. not to any sequence of events in time. A very large number of diJferenr words and phrases occur as expressions of conjunction; but they all fall into a few sets representing very general types oflogical relation. Thus the concept of cohesion .accounts fur the essential semantic relations whereby any passage of speech or writing is enabled to flmction as text. We can systematize this concept by classifying it into a small number of distinct categories- reference, substitution, ellipsis, conjunction, and lexi- cal cohesion; categories which have a theoretical basis .as distinct TYPES of cohesive relation. but which also provide a practical means for describing and analysing texts. Each of these categories is represented in the text by particular features - repetitions, omjssiom, occurrences of certain words and constructions - which have in common the property of signalling that the interpretation of the passage in question depends on something else. If thar • something else • is verbally explicit, then there is cohesion. There are, of course, other types of semantic relation associated with a text which are not embodied in this concept; hut the one that it does embody U. .in some ways the most important, since it is common to text of every kind and is, in fact, what makes :a text a text.

14 INT:RODUCTION I. 3 Cohesion and linguistic context 1.3.1 The domain cf cohesive relations The simplest form of cohesion is that in which the presupposed element is verbally explicit and is found in the immediately preceding sentence; for example [1:23] Did the gardener water my hydrangeas? - He -said so. We shall treateis as the norm for purposes of illustration and discussion; not only became it is simpler in practice but also because it is, as we have suggested, the paradigm case of cohesion from a theoretical point of view, since the boundary between two sentences represents a minimal break in structural continuity. There .are two kinds of departure from this norm. First, the presupposed element may be located dsewhere, in an earlier sentence, perha~ or in the following one; secondly. it may not be found in the text at all. Let us consider these in turn.. Cohesion as we have said is not a structural relation; hence it is unre- stricted by sentence boundaries, and in its most normal form it is simply the presupposition of something tlut has gone before, whether in the pre- ceding sentence or not. This form of presupposition, pointing BACK to some previous item, is known as .ANAPHORA. What is presupposed anaphori- cally may be in the sentence immediately preceding, but it may also be in some earlier sentence; in the following example. he refers back to Henry: [I:24} The first years of Henry's reign, as recorded by the admiring Hall, were given over to sport and gaiety. though there was little of the licentiousness whlch characterized the French Court. The athletic contests were serious hut very popular. Masques, jousts and spectacles followed. one another in endless pageantry. He brought to Greenwich a tremendously vital court life. a central importance in the country's affairs and, above all, a great naval connection.* Or it may be the whole of some longer passage; here the such presupposes everything that precedes: I I: 25] Travelling with huge retinues of strlf and servants, medieval monarchs demanded a series ofhouses to take care of their needs.

I.J COHESION AND LINGUISTiC CONTEXT Ij Their requirements were large. Government went where they went -(it was still the King's government)- with aU its-attendant staff" and visitors. Tbey were responsible for a large number of followers, and visitors had to be entertained in s:tyle. They were expected to dispense patronage and to entertain on a lavish scale. During the winter festival of Christmas, h&ting twenty days. they nominally kept open house. Richard II, notoriously prodigal. entertained over ten thousand every day at his palaces. and even more over Christmas. No single home cottld possibly cope with the organization and material products needed on such a scale.* As might be expected, the tendency is different with different types of cohesion. Where tbe cohesive element is something like he or one, which coheres by direct reference to, or substitution for, another item. the pre- supposed element is typically a specific item in the immediately preceding sentence. This is the most usual pattern in the case of reference and sub- stitution. Cluracterisrically these intances also tend to form COHBS1VE CHAINS, sequences in which it, for example. refers bo1ck to the immedi- ately preceding sentence - but to anothec it in that sentence, and it is necessary to go back three, four or more sentences,. stepping across a whole sequence of its, before fmding the substantial clement. An example of this is [I: 25 J above, which has a cohesive chain medieval nwnarchs ... their ... thq ... they ... they ... they, leading finally to Richard II as a 'pecific instance of a medieval monarch. Here is another example in which three such cohesive chains intertwine, initiated by Short,johnson over J()f'dmt and Johmcn; I r: 26] Short places Jvhnson over JvrJan squarely in the tradition of expressionist drama. He says that Johnson is a 'typical Briton'. an 'English Everyman •. He regacds ilie play as an imaginative presentation of the mind of a man who has just died. But, be adds, Priescley is more interested in Johmon Jiving than in John- son dead. In this the pb:y is expressionist in its approach to theme. But it is also so in its use of unfamiliar devices- the use of masks. the rejection of the three or four act lay-out of the plot. And, finally. he points to the way in which Johnson moves quite freely in and out of chronological time. t It may be helpful to tabulate the ties forming these three chains: * Olive ;md Nigel Hamiltan, R~yal Grrmwich. The Greenwich Bookshop. t Gartth Llnyd Evam, ]. B. ~ry- 771z Dranllllist, I-Jcinenunn.

I6 INTltODUCTION (i) (ii) (ili) Short ]uhnson wer Jtmian Johnson Sentence 1: Short Johmtm over Jordmt Johruon (in J over]) Sentence 2: t Johnson Sentence 3: the play a man who has just died Sentence4: t Johnson (> x) Sentences: 1 the play ... its 1 Sentence 6: it ... its Sentence 7: Jobnson Where the cohesion takes the form of conjunction, with expressions like but, so, in that case, later on, the presupposition typically involves a pas- sage longer than a single sentence. This han:lly needs.iUustraring. hut here is one example, a passage of Carlyle in which the conjunction on the other hmul dearly relate• to the whole of the preeeding paragraph' f I :n] HDw much is still .alive in England~ how much has not yet come into life! A Feudal Aristocracy is still alive, in the prime oflife; superintending the cultivation of the land, and less comciously the distribution of the produce of the land, the adjustment of the quarrel. of the land;judging, .oldiering, a<ljusting; everywhere governing the people, - so that even a Gurth, horn thrall of Cedric, lacks not his due parings of the pigs he tends. Govern- ing;- and. alas, also game-preserving. so that a Robin Hood. a William Scarlet and others have, in these days. put on Lincoln coats, and taken to living, in some universal-suffrage manner~ under the greenwood tree! How silent, on the other hand, lie all Cotton-trades and such like; not a steeple-chimney yet got on end from sea to sea! Lexical cohesion differs aga~ in that it regularly leaps over a number of sentences to pick up an clement that has not figured in the intervening rext' (I:28] I screamed, and my scream went W2fting out on the night air! And some neighbours who - they were my nearest neighho~ but they were still some distance away - came rushing along. They were awfully good, and they .aid afterward. they thought I'd been being murdered. Well, I couldn't've made more noise if I had been! But I'd surprised myself- really, rhe sound that

1.3 COKESION AND LINGUISTIC CONTEXT I7 went floating out on the air l didn't know [ had it in me, and they said it would make my fortune if l sent it to Hollywood.. And I may say it surprised the thief sufficiendy that he dropped my handbag and fied. Fortunately I wasn't between him and the door, so there was no harm done and I didn't lose anything. - Fortunately for him, or fortuna.tdy for you? - Oh. for me; they generally carry knives. - I know; someone was murdered in the main hotel qutte recently. - Oh yes, yes. although people did say that there were wheels v.-ithin wheels in that. But you get between a fleeing thief and his exit, and he's bound to be carrying a knife. But anyhow, the only thing I )ost was my voice. I couldn't speak for a week afterwards. Here lost (in lost •. , my VQiw) resumes the lose (in didn't lose anything), the resumption being sigtuUed by the conjunctive item anyhow; and voice re- lates back to saeam, noise and sound. Resumptions of this kind can span large passages of intervening t~ especially in informal conversation. So fu we have considered cohesion purely as an anaphoric relation, with a presupposing item presupposing something that has gone before it:. But the presupposition may go in the opposite directio~ with the presup- posed dement following. This we slu11 refer to .as CATAPHORA. The distinction only arises if there is an explicitly presupposing item present, whose referent dearly either precedes or follows. If the cohesion is lexical with the same lexical item occurring twice over. then obvibusly the second occurrence must take its interpretation from the first; the first can never be said to point forward to the second. If John follows John. there is no possible contrast between anaphora and cataphora. But an item such as this .and lrere CAN point forward, deriving its interpretation from something that follows. for example: [ r: 29] This is how to get the best results. You let the berries. dry in the sun, rill all the moisture has gone out of them. Then you gather them up and chop them very fine. The presupposed element may, and often does, consist of more than one sentence. Where it does not, the cataphoric reference is often si~ed. in writing with a -colon: but although this has the effect of uniting the two parts into a single orthographic sentence. it does not imply .any kind of structural relation between them. The colon is used solely to signal the cataphora, this being one of its principal functions.

I8 lNTli.ODUCTION There remains one further possibility, namely that the information required for interpreting some element in the text is not to he found in the text at all, but in the situation. For example, given [r:Jo] Did the gardener water those plants? it is quite posslblc that those refers back to the preceding text, to some earlier mention of those particular plants in the discussion. But it is aJso possible that it refers to the environment in which the dialogue is taking place- to the 'context of situation', a:s it i~ ca1Jed- where the plants in question are present and can be pointed to if necessary. The interpretation would be 'those plants there, in front of us'. This type of .reference we shall cali.EXOPHORA, since it takes. us outside the text altogether. Exophoric reference is not cohesive, since it does not bind the two elements together into a text. One might reason that, meta- phorically speaking, the plants form part of the text; but this seems rather pointless, because there could be no significant contrast here between the presence of cohesion and its absence - one would have to assume that. in the absence of cohesive reference to them. the plants would have com- prised a -text on their own. But exophot'a is of interest at several points in the discussion, particularly with reference to the definite article as a text- forming agent, and it will be brought up where relevant. The line between exophoric and amphoric reference is not always very sharp. In dramatic dia1ogue, for examp)e, the mere presence or absence of a stage direction would change the picture, eg [I: 3 I] How sweet the moonlight sleeps upon this bank! Here will we sit, and let the sound of music Cre-ep in our ears. If the stage directions specify something like 'a grassy bank', then for the reader this and here become anaphoric; otherwise, they were exophoric. The significance of the exophoric potential is that~ in instances where the key to the interpretation ls not ready to hand, in text or situation, the hearer or reader COI>."STTtUCTS a context of situation in order to supply it for himsel£ So we supply the grassy bank in our imagination, and the pro- ducer need not put one on the stage. This is an essential clement in all imaginative writing. It may be helpful here to draw attention to the distinction between co- hesion as a relation in the system, and cohesion as a process in the text. • Cohesion' is defined as the set of possibilities that exist in the language for making text hang together: the potential that the speaker or writer has .at

1.3 COHESION AND LINGUISTIC CONTEXT 19 his disposal. This is a purely relational concept, and directionality comes into it only if one of the dements in the cohesive relation is BY rrs NATUJU! cohesive. in that it is inherently 'pointing to • something else; in this case there is a Jogical dependence. and hence a significant opposition IN T.HE SYSTEM between pointing hack. (anaphora) and pointing forwards (cata- phora). But cohesion is also a process. in the sense that it is the instantiation of this relation in a text. A text unfolds in real time, and directionality is built into it;. hence of the ~·o elements embodying the cohesive relation, one always follows the other. In the system: a b In the text: a b implicitly anaphoric explicitly anaphoric (explicitly) cataphoric (time) John1 John he: John, :he John In the text it is natural for the element occurring second to depend for its interpretation on the one occurring first; hence, anaphora is the unmarked and cat:.tphora is the marked term in the opposition. Cataphor.a occun; only as an EXPLICIT relation, with the first element always being one that is inherently presupposing. Thus cohesion as a process always involves one item pointing to another; whereas the significant property of the coheslve relation. as we have stressed above. is the fact that one item provides the source for the interpretation of another. 1.3.2 Text and situaticm We should now say a little more .about the nature of a text. and its relation to a context of situation. Let us begin with an example; {I: 32] Although the light was on he went to deep. Although the house was unfurnished the rent was very high. Although he was paid a high salary he refused to stay in the job. These three sentences dearly have something in common; they are not just three sentences picked at rarulom from a corpus of written English. What they have in common is a certain degree of grammatica1 similarity: parallel struCtures. with repetition of the item aithough. They cou~d. how- ever, be written in any other sequence without disturbing the organiza- tion of the passage as a whole, such as it is; whatever it is that gives unity ro this 'text' it does not depend on the order in which the sentences are arranged.

20 This sort of grammatical parallelism is not irrelevant to internal cohe- sion; it is a common feature not on1y of poetry but of many other kinds of discourse as well. .But by itself it does not make a string of sentences into a text. The sentences in r I : ]2 J could be said to form a text, but if so it is .a text of a very special kind: a text about language. in which the sentences are CITATION FORMS- tlut is., items introduced for the purpose of saying something about them. A set of citation forms that are related ONLY by their grammatical parallelism is a familiar feature of texts about language; and [I: 32) is in fact taken from a textbook of Chinese for English-speaking students. The sentences in it, together with their Chinese equivalents. form part of a drill. The passage illwtrates, in an extreme form, a general principle concern- ing decisions about what is and what is not a text. We do not, in fact, evaluate any specimen of language- and deciding whether it does or does not constitute text is a prerequisite to any further evaluation of it- without knowing something about its context of situation. It is the context of situation of this passage, the &et that it is part of a language textbook,. that enables us to accept it as text. A set of sentence; that in any other environ- ment would not constitute a text is admissible as such in the restricted context of a book about language. Since the present book will be full of citation forms we need not discuss them further here; the effect of their occurrence in a situation to whieh they are inappropriate can be seen in Ionesco' s play The Bald-headed Primatlonna.. But they illustrate the general principle that the hearer or reader, when he is determining, consciously or unconsciously, the status of a specimen of language, invokes two kinds of evidence, the external as well as the internal: he uses not only linguistic dues but also situarional ones. Lingustically, he responds to specific features which bind the passage together, the patterns of connection, inde- pendent of structure, that we are referring to as cohesion. Situationally. he takes into account all he knows of the environment: what is going on, 'What part the language is playing, and. who are involved. The internal and the external aspects of' texture' are not v..-holly separ- able, and the reader, or listener, does not separate them when responding unconsciously to a passage of speech or writing. But when the linguist seeks to make explicit the basis on which these judgments are formed, he is bound to make observations of two rather different kinds. The one con- cerns. relations within the language, patterns of meaning realized by gram- mac and vocabulary; the other concerns the relations BI!TWBEN the language and the relevant features of the speaker's and hearer's (or writer's and reader's) material, social .and ideological environment. Both these aspects

!.3 COHESION AND LINGUiSTIC CONTEXT 21 of a text fall wicl1in the domain of linguistics. The linguistic patterns. which embody • and. at the same time .also impose structure on, our experience of the environment., by the same token also make i.t possible to identify what features of the environment are relevant to linguistic be- haviour and so fOnn part of the context of situation. But there are two sets of phenomena here, and in this hook we are concerned with the UNGUISnc factors that are characteristic of texts in English. The situational properties of texts, which are now heglnning to be studied in greater de- tail and with greater understanding, constitute a vast field of enquiry which lies outside our scope here. Some of the factors of most immediate relevance are summ.ari2ed in the pa-ragraphs that follow. The term SITUATION, meaning the •context of situation' in which a text is embedded, refers to an those extra-linguistic factors which have some bearing on the text itsel£ A word of caution is needed about this concept. At the moment~ as the text of this Introduction is being com- posed~ it is a typical English October day in Palo Alto, California; a green hillside is visible outside the window, the sky is grey~ and it is pouring with rain. This might seem part of the "situation • of this text; hut it is not, because it has no relevance to the meanings expra<>ed. or to the words or grammatical patte.ms that are used to express them. The question is, w-hat are the external factors a£recting the linguistic choices that the speaker or writer makes.. These are likely to be the nature of the audience. tbe me- dium. the purpose of the conununication and ro on. There are types of discourse in which the state of the weather would form part of the con- text of situation, fur example. language-in-action in mountaineering or sailing; but writing a book about language is not one of them. As a rule. the features of the situation are relevant at a rather general level. That is to say~ if we think of the example of a lecture on current affitirs to an adult evening class, what matters is not that it is John Smith ta1king to Messrs Jones,. Robinson,. Brown and others on a particular Toes- day evening in Bumley, but that it is a lecturer addressing a gathering of adult students within the framework of a given social institution. This is not to deny either the individual characteristics of speakers or writers or the imporbnce of studying the distinctive quality of a particular authoc' s style. It is merely to emphasize that many of the features of a text can be explained by reference to generalized situation types. 1.3.3 Cempotrents of tht: conuxt of situation, an.J register The concept of CONTBXT OF SITUATION was formulated by Malinow- ski in I92J. in his supplement to Ogden and R..ichards' The M£aning of

22 INTROD UCTlON Meaning. and subsequently elaborated by Firth. particularly in a paper written in 1950 called •Personality and Lmguage in society'. It has been worked over and extended by a number of linguists, the best-known treatment being perhaps that of Hymes in • M odds of interaction of lan- guage and social setting'. Hymes categorizes the speech situation in terms of eight components which we may summarize as: form and content of text. setting~ participants. ends {intent and effect). key. medium, genre and interactional norms. It will be noted that. in this view of the matter, the text itself forms part of the speech situation. A more abstnct interpretation, intended as a basis for D.EIUVING the features of the text from the features of the situation, had been offered by HaUiday, Mcintosh and Strevens in The Linguistic Sciences tmJ L.mgunge Te.aching. They had proposed the three headings FIELD, MODI!• and TENOlt (to adopt the terminology preferred by Spencer and Gregory in Linguistics and Style). These are highly general concepts for describing how the context of situation determines the kinds of meaning that are expressed. The FI.ELD is the total event, in which the text is functioning. together with the purposive activity of the speaker or writer; :it thus includes the subject-matter as one element in it. The MODE is the function of the text in the event, including therefore both the channel taken by the language - spoken or written, extempore or prepared - and its genre. or rhetorical mod.e, as narrative. didactic, persuasive, 'phatic communion' and so on. The TENOR refers to the type of role interaction, the set of relevant sociai relations, permanent and temporary, among the participants involved. Field. mode and tenor collectively define the context of situation of a text (see the further discussion in Hallid.ay's Language .md Social Man). The linguistic features which are typically associated with a configura- tion of situational features - with particular values of the field, mode and tenor- constitute a REGISTER. The more specifically we can characterize the context of situation. the more .specifically we can predict the properties of a text in that situation. If we merely name the subject-matter, or the medium. it wiU teli m very little; we could talk of a • register of marine biology' or a 'newspaper register', but this hardly enables us. to say any- thing of interest about the types of text: in question. But if we give some in- formation about aB three categories of field, mode, and tenor, we begin to be able to make some useful observations. For instance, if we specify a field such as • personal interaction, at the end of the day. with aim of inducing contentment through recotmting of familiar events •, with mode 'spoken mQnologue, imaginative narrative, extempore • and tenor • intimate, mother and three-year-old child'. we can reconstruct a great deal of the

1.3 COHESION AND LINGUISTIC CONTEXT 23 language of this kind of bedtime story. especially if we go further and describe the CONTEXT OF CULTURE (another of Malinowski's concepts) which will tell us, among other things, what are the familiar events in the life of a child with the given socio-cultural background. The register is the set of meanings, the configuration of semantic patterns, that are typi- cally drawn upon undcr the specified condiri~ along with the words and structures that are used in the realization of these meanings. The fact that we can say of any given text. with some assurance, whether or not it 'Satisfies a d.escripri<?'n of the context of situation such as the one just given, shows how real the notion of register is. In general, if a passage bangs together as a text, it will display a con- sistency of register. In other words, the texture involves moce than the presence of semantic relations of the kind we refer to as cohesive. the de-- pendence of one element on another for its- interpretation. It involves also some degree of coherence in the actual meanings expressed: not only, or even mainly, in the CONTENT, but in the TOTAL selection from the semantic resources of the language, including the various interpersonal {social- expressive-conative) components- the moods. modaliries. intensities, and other forms of the speaker's intrusion into the speech siruatioa The concept of COHRSION can therefore be usefully supplemented by that of REGISTER~ since the tWO together effectively define a TEXT. A text is a passage of discourse which is coherent in these two regards: it is co- herent with respect to the context of situation, and therefore consistent in register; and it is coherent with respect to itsei£: and therefore cohesive. Neither of these two conditions is sufficient without the other, nor does the one by necessity entail the other. Just as one can construct passages which seem to hang together in the situational-sem.antic sense, but fail as texts because they lack cohesion, so also one can construct passages which are beautifully cohesive but which fail as texts because they lack consis- tency of register - there is no continuity of meaning in relation to the situation. The hearer, or reader, reacts to both of these things in his judg- ment of texture. Under normal circumstances, of course, we do not find ourselves faced with 'non-text', which is ·non-sense' of a rather esoteric klnd. Texture is a matter of degree. It is almost impossible to construct a verbal sequence which has no texture at all - but this, in turn. is largely because we insist on interpreting any passage as text if there is the remotest possibility of doing so. We assume, in other words,. that this is what language is for; whatever its specific function may be in the particular instance, it can serve this function only under the guise of text. If one can imagine a situation

INTllODUCTION in which someone is faced with a string of words picked at random fi:oma dictionary, but which has been made to look or sound as if it was struc- tured, then it is safe to predict that he will go to great lengths to interpret it as text, and as related to wme accessible features of the situation. The nearest we get to non-teXt in actual life, leaving aside the works of those poets and prose writers who deliberately set out to create non-text, is probably in the speech of young children .and in had translations. Two further points are worth making. in connection with the text and its context of situation. One is that the relation of text to situation is very variable, in terms of the relative weight which the text has to bear. There are certain types of situation in which the non-linguistic factors clearly dominate and the language plays an ancillary role: for example, a non- verbal game, like football, in which there are a few verbal insttuctions from player to player; or joint operations on objects, building, assembling, cooking, cleaning and the like. Here it is impossible to interpret what is said or written without situational information; one must know what :is going on. At the other end of the scale are types of activity in which the Language is the whole story, as in most formal or infonnai discussion on ·.abstract themes, such as those of business. politics and intellectual life. Here the language may be totally self-sufficient and any relevant situa- tional factors are derivable from the language -itsel£ The qualiry of texture, and the forms of cohesion which provide it, difier very much as hetwccn these two poles. One question on which a great deal of further study is needed is the relation between texture and situation type: the different ways in which texts of different kinds are constructed so as to form seman- tic whales. The second point concems what Ellis calls DELICACY OF FOCUS in situational analysis. We obviously cannot draw a clear line between 'the same situation' and 'different situations •; any nvo contexts of situ3tion w:iH be alike in some respects and not in others, and the .amount of detail needed to characterize the situation will vary according to what we are interested in - what distinctions we are trying to make betv.reen one in- stance -and another~ what fearuces of the text we are trying to explain and so on. Questions ]ike £are these two teXts in the same register?' .are in themselves meaningless; we can only ask in what respects the texts:. and the situations, ace alike and in what respects they dilfer. If a child turns around from talking to his &ther and starts talking to his uncle, we are not called on to decide whether the situation has changed or not; but we shall be interested to note whether there are linguistic signals of the dift"erence in personal relationships. This aJfects our notion of a text. Up to now we have

I.J COHESION AND LINGUiSTIC CONTEXT 25 been discussing this. on the assumption of an all-or-nothing view of tex- ture: either a passage forms text, or it does not. In real life we so seldom meet non-text thar we can aiford to adopt such a deterministic view: we are not required in practice to decide where a text begins and ends. :But in fact there are degrees of texture, and-if we are examining language from this point of view, espe<:ially spoken language, we shaU at times be uncer- tain as to whether a particular paint marks a continuation of the same text or the beginning of a new one. This is because texture is .really a • more-or- less• affair. A partial shifi in the context of situation- say a shift in one situational factor. in the fidd of discourse or in the mode or tenor- is likdy to be reflected in some way in the texture of the discourse. without destroying completely the continuity with what has gone before. It is worth pointing out in this connection that continuity of subject- matter is neither a necessary nor a sufficient condition for the creation of texture. Subject-matter .is neither more nor less important than other fea- tures of the context of situation as a determinant of text; it is simply one of the &cton that enters into the picture. And where there is continuity of subject-matter within a text, as we typically find it, the texture is. not necessarily the result of this ; the following example is about mathematics, but cohesion is provided, especially in the last sentence. more by the lexical patterns. of complicated . .. difficult ... easy and greater time .. , long •.. short than by any linking of specifically mathematical concepts: [1 :33] Throughout the long history of mathematics. men have always wished that they could calculate more quickly. fu each mathema- tical discovery was made and knowledge advanced a little the calculations facing mathematicians became more and. more com- plicated and demanded an even greater rime. There are some peopk who like doing long and dilficult arithmetic, but most of us do not and a.re eager to fmish our sums in the shortest and easiest way.* A text, then, can be thought of as the basic unit of meaning in language. It Js to semantic structure what the sentence is to lexicogramrnatical struc- ture and the syllable to phonological st:ruc:ture. It is a unit of situational- scmantic organization: a continuum of meaning-in-context, constructed around the semantic relation of cohesion. According to the particular situational-semanticconfiguration, orltEGISTER, of the text, so the forms taken by the cohesive rdarion will differ: texture in informal conversation

INT11.0DtJCTION is quite unlike that in fonnal written language, which is one reason why the former looks strange when written down and the Jatter sounds odd when read aloud. A text therefore normally has continuity of register; it <fits' a given set of situational features, a pattern formed by the nature of the communicative event (field), the place' assigned to bnguage acts within the event (mode) and the role-relationships of those who are participating (tenor). This fit does not by itsdf ensure the kind of continuity we :asso- ciate with texts; we often fed, in looking at children's v,rriting for example, that it OUGHT to hang together precisely because it is making sense in the situation, but in fact it does not, This reveals the existence of the other .aspect of texture, which is cohesion. The meaning relations which constitute cohesion are a property of text as such, and hence they are general to texts of all types, however much they may differ in the parti- cular form they take in one text or another. Texture results &om the combination of sem.1.ntic configurations. of two kinds: those of register. and those of cohesion. The register is the set of semantic configuratiorn that is typically associated with a particular cuss of contexts of situation, and defines the substance of the text: WHAT IT MEANS, in the broadest sense, including alJ the components of its meaning, social, expressive, communicative and so on as well as repre- sentational (see 1.3-4 below). Cohesion is the set of meaning relations that ls general to ALL CLASSES of text, that distinguishes text from 'non-text • and interrelates the substantive meanings of the text with each other. Cohesion does not concern what :t text means; it concerns how the text is constructed as a semantic edifice. 1.3-4 The pL:zu of cohei;ion in th£ linguistic system Table I summarizes the main components in the linguistic system. show- ing where cohesion comes in relation to the rest. 'Ibere are three rru:jor functional-senuntic components.. the IDEA- TIONAL, the INTEllP.R.RSONAL and the TEXTUAL. The IDEATIONAL component is that part of the linguistic system which is concerned with the expression of 'content', with the function that language has of being ABOUT something. It has two parts to it. the experiential and the logi.ca1, the former being more directly concerned with the representation of experience, of the 'context of culture • in Malinowski' s terms, while the latter expresses the abstract logical relations which derive only indirectly from experience. The INTERPERSONAL component is concerned wi.th the social. expressive and conative functions of language. with expressing

I.J COHESION AND LINGUISTIC CONTEXT 27 the speaker's • angle': his attitudes and judgments, his encoding of the role .rdationships in the situation. and his motive in sayi.ng anything at aH. We can summarize these by saying that the ideational component represents the speaker in his role as observer, while the interpersonal component represents the speaker in his role as intruder. There is a third component, the TEXTUAL,. which is the text-forming component in the linguistic system. This comprises the resources that language has for creating text, in the same sense in which we have been using the term all along: for being operationally rdevant,. and cohering within itself and with the context of situation. In part, the textual component operates like the other two. through systems a5sociated with particular ranks in the grammar{see 7-4-I bdow). For example. every clause makes a selection in the system of THEME, a selection which -conveys the speaker's organization of the clause as a mes- sage and which is expressed through the normal mechanisms of clause structure. But the textual component also incorporates patterns of mean- ing which are realized outside the hierarchical organization of the system. One of these is INFORMATION structur~ which is the ordering of the text, independendy of its construction in terms of sentences, clauses and the like, into units of information on the basis of the distinction into GIVEN and NEW: what the speaker is treating as information that is recoverable to the hearer {given) and what he is tr-eating as non-recoverabJe (new). This aspect of the meaning of the text is realized in English by intonation, the infor- mation unit being expressed as one TO Nil GROUP. The remaining part of the textual component is that which is concerned with cohesion. Cohesion is closely related. to information structure, and indeed the two overlap at one point (see 5.8.2 below); but information structure is a form of structure, in which the entire text is blocked out into elements having one or other function in the total configuration- every- thing in the text has some status in the 'given-new' framework. Cohesion. on the other hand. is a potential for relating one element in the text to another. wherever they are and without any implication that everything in the text has so.me part in it. The information unit is a structural unit. although it cuts across the hierarchy of structural units or constituents in the grammar (the 'rank scale' of sentence. clause and so on); but there are no structural units defined by the cohesive rdation. Cohesion, therefore. is part of the text-forming component in the linguistic system. It is the means whereby elements that are structurally unrelated to one another are linked together. through the dependence of one on the other for its interpretation. The resources that make up the

28 cohesive potentia) are part of the total meaning potential of the language, having a kind of catalytic function in the sense tha~ without cohesion, the remainder of the semantic system cannot he effectivdy activated at all. 1.3.5 The tManing cf cohesion The simplest and most general forms of the cohesive relation are "equals' and 'and •: identity of reference. and conjoining. We shall discuss the meanings of these and of the other forms of cohesion. and related mean- ings in other parts of the linguistic system, in a rather summary way in