Chapter 6 Lexical Cohesion 6.r The class of' general nouns' In the previous four chapters, we have described the various types of grammati<:al cohesion: reference, substitution and ellipsis, and conjunc- tion. In order to complete the picture of cohesive relations it is necessary to take into account also lexical cohesion. This is the cohesive effect achieved by the selection of vocabulary. On the borderline between grammatical and lexical cohesion is the cohesive fm1ction of the class of GHN£RAL NOUN. We can speak about a borderline hcrc because a general noun is itself a borderline case between a lexical item (member of an open set) and a gram.matica1 item. (member of a dosed system). The class of general noun ls a small set of nouns having generalized reference within the major noun classes, those such as 'human noun'~ • place noun'. • fa-ct noun' and the like. Examples are: pt!ople, person, man, woman, chilJ, boy, girl [human] ueabl.re fnon-human animate] thing, object [inanimate concrete count) stuff [inanimate concrete mass} business. JtjJair, matter [inanimate abstract] move [action] plau [place] questiou. idea [fact} These items are often neglected in descriptions of English; but they play a significant pan in verbal interaction,. and are also an important source of cohesion in the spoken language. The fOllowing examples illustrate their cohesive function: [6: r] a. Didn't everyone make it dear they expected the minister to
6.I. THE CLASS OF 'GENI!RAL NOUNS' 275 resign ? - They did. But it seems to have made no impression on the man. b. 'I should like ro be a littl~ Jargcr, sir, if you wouldn't mind,' said Alice: 'three inches is such a wretched height to be'. 'It's a very good height indeed!' said the CaterpiUar angrily. rearing itself upright as it spoke (it ~s exactly three inche<> high). 'But I' m not used to it ! ' pleaded poor Alice in a piteow tone. And she thought to herse)f. 'I wish the creatures wouldn • t be so easily offended ! ' c. What shall I do with aU this crockery ? - Leave the stuff there; someone'H come and put it av.d.y. d. We .all kept quiet. Th:at seemed the best move. e. Can you tell me where to stay in Geneva? I've never been to the plae<. f. Henry seems convinced there's mone-y in dairy farming. I don't know what gave him that idea. As these examples show~ a general noun in cohesive function is almost always accompanied by the reference item the. This the is anaphoric, and the effect is that the whole complex • the + general noun • functions like an anapboric reference i.tem. The most usual alternative to the is a demon- strative, and if a demonstrative occurs it usually carries the tonic: cfi that idra in example [6: rf). This relates to the fact that the general noun itself does NOT carry the tonic, if it is functioning cohesively; a fact v.-hich holds true even when it occurs in final position, which is the unmarked location of tonic prominence. Hence in f6: Ia, d, e and f) it would be highly improbable, and strongly contrastive, to assign tonic prominence to man, nwve, place and iJeil. The ahove gives us some indication of the status of general nouns. From a lexical point of view, they are the superordin.ate members of major lexical sets, and therefore their cohesive use is an instance of the general principle whereby a superordinate item operates anaphorically as a kind of synonym(see 6.2 below). From a grammatical point of view, the combination of general noun plus specific determine£, such as the man., tlu> thing, is very similar to a reference item. There is little difference between it Jeems to have ~t~ade very little impression on the man and it seems to have matk very little impression 011 him: in both instances interpretation i~ possible only by reference to something that has gone hefore. But it is not the case that there is no difference at all : the form with general noun,
276 LEXICAL COH.F.SION tlu: man, opens up another possibility, that of introducing an inte£penonal clement into the meaning, which is absent in the case of the personal pronoun. (It may be worth stressing once again that the fact that general nouns are very general in meaning, and therefore often interpretable only by reference to some element other than themselves, does not make them unimportant in the language. Since they require recourse to another item. that item must he located earlier within the same text; and this means that they play a significant role in making a text hang together.) The expression of interpersonal meaning. of a particular attitude on the part of the speaker, is an important function of general nouns. EssentiaUy the attitude conveyed is one of familiarity, as opposed to distance, in which the speaker assumes the right to represent the thing he is re-ferring to as it impinges on him personally; hence the specific attitude may be either contemptuous or sympathetic, the two being closely related as forms of personal involvement (if the mean!ng of diminutives in many languages). There are quite a few general nouns which have this inter- personal element as an inherent part of their meaning, especially those referring to human beings, for example idi()t. fool, devil, dear; and these are supplemented. at any one moment in time, by a host of more or l~s slang terms differing widely fmm one social group and one generation to another. But whether or not it is inherently attitudinal in meaning, a general noun in cohesive function can .always be accompanied by an attitudinal Modifier. So we have examples such as the dears. the poor dears; and also the stupid thing. the lucky fellow and so on: [6: 2) a. I've been to see my great-aunt. The poor old girl's getting very forgetful these days. b. Alice caught the baby with some difficulty, as it was a queer- shaped little creature. and held out its arms and legs in all directions, 'just like a star-ftsh', thought Alice. The poor little thing was snorting like a steam-engine when she caught it. -. c. Henry• s thinking of rowing the Atlantic. Do go and talk. to the wretched fool. These forms with interpersonal elements in their meaning have certain special features when they are used cohesively. The general nouns may be ti':lnSCategorized up the scale inanimate-animate-human, with creature being used for human .as well as animate, and thing for all three (cf 3.2.3 above). The adjectives cannot he submodified, by words such a~ very, nor can they he compared. Only adjectives with an attitudinal meaning can
6.2 TYPES OF REITERATION 277 occur; it would not be possible to say the fat man in [6: ra]- or if one did, by virtue of its occurrence in this context fat would become attitudinal. A general noun i.n cohesive function can in fact accept only non-defining Modifiers; since it refers back to the entire nominal group ;,vith which it i.s to be identified for its interpretation, it carries over any defining ele- ments from this nominal group, and hence it must -itsdf remain undefined. Attitudinal adjectives are by their nature non-defming. Here is a Shakes- pearean example (the good man, referring to Lear); f6: 3} AU blest secrets, All you unpublish' d virtues of the earth. Spring with my tears l be aidant and remediate In the good man's distress l & a corolbry of their carrying over of definiteness, general nouns of the human class are very frequently used in anaphoric reference to penonal names. It is interesting that the other use of these nouns., when there is an attitudinal element present either in the noun itself or in the form of modi- fication, is as vocatives: tenns of .abuse or endearment, you crazy fool! and the like. There they are exophoric instead of anaphoric; :md this nnder- lines the fact that the typical context in which they function is a referential one, so that like reference items they refer either to the situation or to the preceding text. The interpersonal element of attitude, however, although it is fre- quently associated with the cohesive use of general nouns, is by no means always present; this kind of anaphoric reference does not necessarily embody any attitudinal meaning. The following are entlrel y neutral: f6:4] a. J've just read John Smith's essay. The whole thing is very well thought out. b. Robert seems very worried about something. I think you ought to kve a talk with the boy. Here the items thing and brq refer :maphorically to Jafm Smith's e.ssay and Rabrrt respectively~ and again the identity of reference is slgnalled. by the presence of the anaphoric reference item the. 6.2 Types of reiteration Thus the use of general nouns as cohesive .agents depends on their occur- ring in the context of reference ~ having the same referent as the item which they presuppose, this being signalled by the accompaniment of 01 reference item.
278 LE-XICAL COHE-SION This use of general words as cohesive demcms, however, when seen from the lexical point of view, is merely a special case of a much more general phenomenon which we may term REITERATION. Reiteration is a form oflexical cohesion which involves the repetition of a leA-ica.l item, at one end of the scale~ the use of a general word to refer back to a lexical item, at the other end of the scale; and a number of thi.ngs in between - the use of a synonym, near-synonym, or superordinate. Let us illustrate each of these in turn. [6:5] a. There vtas a large mushroom growing near her, about the same height as herself; and, when she had looked. under it, it occurred to her that she might .as v.rell look and see what was on the top of it. She stretched herself up on tiptoe., and peeped over the edge of the mushroom, ... b. Accordingly ... I took leave, and turned to the ascent of the peak.. The climb is perfectly easy ... c. Then quickly rose Sir Bedivere, and ran, And leaping dov.n the ridges lightly. p)ung' d Among the bulrush beds, and dutch' d the sword And lightly wheel' d and threw it. The great brand Made light'nings in the splendour of the moon ... d. Henry's bought himself a new Jaguar. He practically lives in the car. In (a), there is REPETITION: mushroom refers hack to mushroom. In (b) climb refers. back to ascent, of which it is a SYNONYM. In (c) brand refers back to sword. of which it is a near SYNONYM. In (d), car refers hack to Jaguar; and car is a SUPERORDlNATE of Jaguar- that is., a name for a more general class {as vehicle is a superordinate of car, spo<m of teaspoon, cut of pare, and so on). All these are cohesive in e...xactly the same v.tay as the GEN.ERAL WORDS illustrated ih [6: 1-4]; the latter differ only in level of generality. All these instance:;; have in common the- fact that one lexical item refers back to another, to which it is related by having a common referent. We shall refer to this general phenomenon as R.EIT.ERA T ION. A reiterated item may be a repetition, a synonym or near-synonym, a superordinate, or a general word; and in most cases it is accompanied by a reference item. typically the. At the same time, there is no sharp dividing line between these forms., consis.ting of a related lexical item plus anaphoric the, and. the personal
6.2 TYPES OF REITERATION 279 reference forms such as he and it. We can in fact recognize a continuum, or 'dine'. of cohesive elements; for example (adapting f6:sb] above): ( The ascent 1 Thedimh (6: 6] I turned to the ascent of the peak. "\, The task : The thing is perfectly ~ It Here we have (r) the same item repeated, (2} a synonym. (3) a super- ordinate, {4) a general noun and (5) a personaJ reference item. Here ascent and climb arc lexical items whose interpretation IN THrs INSTANCE is shown (by anaphortc' the) to be idcncicai with that of an earlier lexical item to which they are related either by repetition (ascmt) or by synonymy (climb). The same ls true of task, except that task :is a more generaherm. higher in the lex\cal taxonomy; so the cohesive environment of the word task adds specificity to it - when we interpret the task by reference to the ascent of the yak we identify the particular kind of task referred to. The word Jhing is an even more general term which is being used in exactly the same way; but it is still more specific than it, because it usually excludes people and animals. as well as qualities. states and relations, and it always excludes facts and report&. Mo:st general of all is the reference item it; but even it is nm a' pure' phoric element sinceitJikewise embodies some specifici.ty, though only minimal: it excludes people. The form it comes closest to being an alternative realization of genera1 noun + reference item, as in the thing. Hence the boundary between lexical cohesion of the type we are calling REJTERATIO~. and grammatical cohesion of the REFERENCE type, is by no means clear cm; the class of general nouns provides a form of cohesion that lies somewhere in between the two, and is interpretable as either. Here we are interpreting it as lexical cohesion, and bringing it under the general heading of what we are calling reiteration. When we talk about .REITERATION, therefore. we are including not only the repetition of the same lexica1 item but also the occurrence of a related item~ which may be anything from a .synonym or near synonym of the original to a general word dominating the entire class. Let us categorize these as above: any instance of reiteration may be (a} the SAMB WOltD, (b) a SYNONYM or NEAR-SYNONYM, (c) a SUPERO:RDINATE or (d) a Gl!ll;l!llAL WORD. For example: [6: 7] There's a boy climbing that tree. a. The boy's gojng to fall ifhe doesn't take care.
280 b. The Jad's: going to fall if he doesn't take care. c. The child's going to fall if he doesn't take care. d. The idi:ot' s. going to faH if he doem' t take e2re. In (a), boy is repe-ated. In (b), the reiteration takes the form of a synonym lad; in (c), of the superordinate term child; and in (d), of a general word idiot. It is typical of such general words, at least those referring to people, as we have seen, that they carry a connotation of attitude on the part of the speaker, usually one of familiarity (derogatory or intimate). Here is another example, this time with a non-human referent: {6: 8) There's a boy climbing the old elm. a. Thatelm.i:m'tverysafe. c. That tree i:m' t very safe. d. That old thing isn't very safe. Here (a) repeats elm, (c) selects the superordinate ttu, and (d) the general word thing. It is difficult to find a synonym of the same degree of specifi- city in this example; we could find one in a series like: There's a boy dimbing alang the rafters. (a) Tlw~ rafters ... (b) Those beams , , , (c) These timbers ... (d) Those things . ... The category of superordinate, illustrated in (c), refers to any item whose meaning includes that of the earlier one; in technical terms, any item that dominates the earlier one in the lexical taxonomy. There are often severai possible superordlnate terms, words that are intermediate between the lowest level, represented by (a) and (b), and the highest, represented by (d). That is to say, there may he various degrees of generality intermediate between the presupposed item itself. eg: elm in [6:8}, on the one hand, and a very general word like thing on the other. Words wi.th intermediate status are more open to modification, though still with a tendency to some evaluative meaning, eg: this eminmt authM. The general words, which correspond to major classes of lexical items, are as we have said very commonly used with cohesive force. They are on the borderline between lexical items and substitutes. The substitutes one and Jo can be thought of as being as it were the highest points in the lexical taxonomy of nouns and verbs respectively; as such, they con- stitute a dosed dass, and so acquire a purdy grammatical function. Next below them come the general words, such as thing, person, make, do and so on; these, although limited in number, are not dearly hounded and it i'> hardly possible to compile a definitive list of them. They do function more or less as lexical items. ro when they occur cohesively we can treat
6.2 TYPBS OF REITERATION 281 them as instances of lexical cohesion. But there is no sharp line between substitutes and general words- because there is no very sharp line between grarnrn:1r and vocabulary: the vocabulary. or lexis, is simply the open- ended and most 'delicate' aspect of the gram.m.ar of .a language. Not all general words .are med cohesively; in &et, ouly the nouns are, for the reason noted above, namely that a general word is cohesive only when in the context of reference - that is, when it has the same referent as whatever it is presupposing, and when it is accompanied by a refereru:e item. All the types of lexical cohesion that we have considered up to this point have involved identity of reference; no matter whether the reitera- ted item has been a repetition. a synonym. a superordinate or a general word, it has been assumed to share a common referent with the originaL Keeping to this assumption for the moment we can shift our point of view from the grammatical to the lexical and look at reference from the lexical angle, .interpreting -it as a means of avoiding the repetition of lexlcal items and thus making it dear that if the lexical item had been reiterated it would have had the same referent. The simplest illustration of this is provided by proper name§. Suppose we have {6:9] John took Mary to the dance. John was left all alone. -how do we know whether it's the same John? The answer to this is, .if you want to make it dear that it is the same John, don't call him John; call him he. In other words, we use a reference item; and this conveys the meaning • the present sentence is related to the lao;t one by the fact that both contain a reference to the ..ame individual". This does not mean that a repeated proper noun can never have the same referent as it had on its first occurrence; the second John couLD refer to the same person as the first- we simply do not know whether it does or not. If John is repeated, we need some further signal to tdl us how to interpret it. With common nouns, the means arc readily available; the signal is given by a reference item, typically the. So for example in [6:Io]Just then a Fawn came wandering by: it looked at A1ice with its large gentle eyes, but didn't seem at a11 :&ightened ... 'What do vou call voursdf?', the Fav.rn said at last. the signals 'the Fawn referred to on this: second occasion is the same Pav.m as that referred to in the (or some) preceding sentence'. From this it would seem that it is not the repetition of the word Fawn that has. the cohesive effect, but only i.ts repetition accompanied by an
2-82 LEXICAL COHBSION anaphoric reference item. This might suggest that there was no distinct category of LEXICAL COHESION; that what we are caUing •lexical cohesion' was merely the reiteration of a lexical item in a context of grammatical cohesion, the cohesion being simply a niatter of reference. But that is not, in fact, the whole story. It is true that lexical reiteration, where the reference is identical, is usually made explicit by means of an anaphoric reference item.. But there are other types of lexical cohesion which do not depend on identity of reference; patterns of word occur- rences which by themselves give a separate, purely lexical dlmension of internal cohesion to a text. 6. 3 Lexical relations as cohesive patterns The most immediately obvious type offexica1 cohesion is that illustrated by the Fa<..vn in [6: roJ, where the same word is repeated and has the same referent on both occasions. We have already seen that it .is not necessary for the second instance to be an exact repetition of the same word; it rruy be any kind of what we have called REITERATION -synonym, superordinate, or general word. We ha"r"e assumed up to this point, how- ever, that there must be identity of reference between the two. and this suggested that 'lexical cohesion' was to be interpreted simply as an accom- panying feature that may be associated with grammatical reference. [t is not necessary for two lexical occurrences to have the same referent, however, in order for them to be cohesive. Consider the following examples: {6: I I] Why does this little boy have to w£iggle all the time? a. Other boys don't wriggle. b . .Boys always wriggle. c. Good boys don't wriggle. d . .Boys should be kept out of here. In (a). boys ties with boy although they arc not coreferential. This could he explained as cohesion by comparative reference, in view of the item other; but in (b) there is no identity of reference and no reference item either, yet boys still coheres with boy. It would be possible to use a personal reference item INSTEAD OF boys here (they always wriggle); this reflects the weak relation of coreference that does exist between the two - boys refers to • all boys • and therefore by implication :includes • this little boy·. In (c), however, there is neither the implication of inclusion nor any form
6.3 LEXICAL RELATiONS AS COHESIVE PATTERNS 283 of reference whatever; yet still there is the same cohesive relation between boys and boy. Nor is this relationship in any way dependent on the presence of other items suggesting the same general referential environ- ment; it is not the wriggling that provides the context, as (d) shows. Many instances of cohesion are purely lexical, a fWlcrion simply of the eo- occurrence oflexical items, and not in any way dependent on the relation of reference. A lexical item, therefore, coheres with a preceding occurrence of the same item whether or not the two have the same referent, or indeed whether or not there is any referential relationship between them. Let us summarize the possibilitiec: with another example. The second occurrence may be, as far as reference is concerned, (a) IDENTICAL, {b) INCLUSIVE, (c) EXCLUSIVE or (d) simply UNRELATED. So for example: [6: 12] There's a boy climbing that tree. a. The boy's going to fall if he doesn't take care. b. Those boys are always getting into mischief. c. And there's another boy standing underneath. d. Most boys love climbing trees. In (a) the bey has the ume referent as a boy has; the reference item he could be used instead. In {b) those boys includes the boy referred to pre- viously, and others as wdl; here we could have a reference item they on the basis of the weak coreferentiali:ty referred to in Chapter :z. {2.4.L3), -w-here the relation is one of inclusion; if example [6: IIb] above. In (c) annther boy ex dudes the boy referred to in the first sentence; here there is explicit NON-identity of reference, and in such instances we cannot have a reference item to replace boy - we can however have a substitnte or eUiptical form, another one or another. In (d), most ~s bears no referential relation at all to the boy previously mentioned~ we cannot gather from {d) whether the boy in question likes climbing trees or not, and the speaker does not necessarily know, or care. This is borne out by the fact that he eo old make it exp\icit either way, by the use of a particular intonation pattern: (6:r2] a~. //IMOSTboyslovedimbingtrecs// d". //4 MOST boys f/ I LOVE climbing trees// where (d') means 'just a~ that one does' v.-hile (d") means 'whereas I'm not sure about that one•. Ch:tracteristicaily in (d). where there is no rela- tion of reference between the two occurrences of boy, there is more lexical repetition overall~ here nvt only bor but also climb and tree are
284 repeated. and this compensates. as it were, for the lack of any referential connection. Properly speaking, reference is irrelevant to lexical cohesion. It is not by virtue of any referential relation that there is a cohesive force set up between two occurrences of a lexical item; rather, the cohesion exists as a direct relation between the forms themselves (and thus is more hke sub- stitution than reference). So for example there is cohesion betv.-een the two occurrences of wriggle in [6: I ra]; the question whether they refer to the same wriggling is one which, fortunately, does not arise. Compare: [6:r3] :a. Henry presented her with his own portra.it.. As it happened, she had alw:ays wanted a portrait of Henry. b. The Forthright Building Society required, apparently, that a borrower should sign. seal and deliver the mortgage deed in the presence of a solicitor, so" that the solicitor would sign it as the witness. This is quite a common requirement. Where a borrower is legally represented. his own solicitor will usually be the witness to the borrower's execution of the mortgage deed..* In (a), the second occurrence of portrait is indefinite; but it is still cohesive. The last sentence in {b) contains the items borrower. w~ss, :solicitor and mortgage deed, a11 of which arc repetitions and as such cohere with the earlier occurrences; hut the whole discussion is hypothetical and the quest- ion of coreference is simply not applicable, or decidable. 6.4 Collocation We now come to the most probkmatical' part of lexical cohesion, co- hesion that is achieved through the association of lexical items that regularly eo-occur. We have seen that lexical reiteration takes place not only through repe- tition of an identical lexical item but also through occurrence of a different lexical item that is systematically related to the ftrst one., as a synonym or superm:dinate of it. This principle applies quite genera.lly. irrespective of whether or not there is identity of reference; so, for example. in [6: 1 I] we could have had chi/Jren instead of boys throughout and the effect would still have been cohesive. * ~ Lzgd s;tk of Buying <1 Houu. Coruumen· Association, 19{jj.
6.4 COLLOCATION 285 Furthermore, W"e find that the (;ohesive eftect is still present if in place of chilJren we now have girls: [6: •4.] Why does this little boy wriggle all the time? Girls don,t wriggle-. Girls and boys are hardly synonyms, nor is there any possibility of their having the same referent; they are mutually exclusive categories. Yet their proximity in a discourse very definitely contributes to the texture. There is obviously a systematic relationship between a pair of words such as boy and girl; they are related hy a particular type of oppositeness, called COMPLEMBNTARITY in Lyons' classification. We can therefore extend the basis of the lexical relationship that features as a cohesive force and say that there is cohesion between any pair of lexical items that stand to each other in some recognizable lexicosemantic (word meaning) relation. This. would include not only synonyms and near-synonyms such as dimb ... asunt~ beam ... rafter. disease ... illness. and super ordinates such as elm ... tree, boy , .. child. skip ... play, but also pairs of opposites. of various kinds, complem.entaries such as boy , .. girl, stand up . . . sit rkwn, anton·yms such as like . .. hate, wet . .. dry, crowded . .. deserted, and converses such as order .. . obey. Jt also includes pairs of words drawn from the same ordered series. For example. if Tuesday OCCU!"S in one sentence and Thursday in another, the effect ¥rill be cohesive; si milady Jollllt' . .. cent. Mtth . , . south. colonel . .. brigadier. Likewise with any pairs drav.rn from uruJrdered lexical sets, like basement ... rcnj. road ... rail, red ... green. The members of such sets often stand in some recognizable semantic relation to one another; they may be related as pan to whole, like car ... brake, box ... lid. or as part to part, like mouth .•. chin. verse •.. clwrus (or- rtfrain); they may he co-- hyponyms of the same superordinate term,. ie both members of the same more general class, such as chair •.. table (both hyponyms of forniture), walk ... drive(bothhyponyms of go); and so on. The members of any such set stand in some kind of semantic relation to one another, but for textual purposes it does not much matter what this relation is. There is always the possibility of cohesion between any pair of lexical items which are in some way associated with each other in the language. So we will find a very marked. cohesive effect deriving from the occurrence in proximity with each other of pairs such as the following, whose meaning relation is not easy to classify in systematic semantic terms: laugh" .. joke, blade ... sh.Jrp,gartlen ... dig. ill ... doctor, try ...
286 LEXICAL COHESiON succeed, bee .•• honey, door . . : windr1w, king ... crown, boot ... row, sun- shine .•. cloud. The cohesive effect of such pairs depends not ro much on any systematic semantic relationship as on their tendency to share the same lexical environment, to occur in COLLOCATION with one .another. In general, any two lexical items having similar patterns of coUocation - that is, tending to .appear in similar contexts - vviU generate a cohesive force if they occur in adjacent sentences. This etfect is not limited to .a pair of words. It is very common for long cohesive chains to be built up out of lexical rdations of this kind, with word. patterns like candle ..• flame ... flicker, hair ... comb ... curl . .. wave, poetry ... literaturl! . .• reader ... writer . .. sJyle, sky ... sunshine •.. doud ... rain weaving in and out of successive senten~ Such patterns occur freely both within the same sentence and across sentence boundaries; they are largdy independent of the grammatical structure. Rather than citing a number of short passages to iUustrate dill we will quote one long paragraph which is a typically rich reserve of such collocational cohesion; note the imponance of the title in this regard: A RIDE ON AN AVALANCHE Few Y osemite visitors ever see snow avalanches and fewer still know the exhilaration of riding on them. In all my mountain- eering I have enjoyed only one avalanche ride. and the start was so sudden and the end. came so soon [ had but little time to think of the danger that attends this sort of travel. though at such times one thinks fast. One fine Yosemite morn1ng after a heavy snow- falL. be1ng eager to see as many avalanches as possible and wide views of the forest and summit peaks. in their new white robes before the sunshine had time to change them, I set out early to climb by .a side canyon to the t-op of a commanding ridge a little over three thousand feet above the Valley. On account of the looseness of the snow that blocked the canyon I knew the climb would. require a long rime, some three or four houn as I estimated; but it proved far more difficult than 1 had anticipated. Most of the way I sank w;~ist deep, almost out of sight in some places. After spending the whole day to within half :m hour or so of sund-own. [ was still several hnndred feet below the summit. Then my hopes were reduced to getting up in time to see the sunset. But I was not to get summit views of any sort that day, for deep trampling near the canyon head, where the s:now- was
6.4 COLLOCATION 287 strained~ started an avalanche, and I was swished down to the foot of the canyon as if by enchantment. The wallowing ascent had taken nearly all day, the descent only about a minute. When the avalanche started I threw myself on my back and spread. my arms to try to keep from sinking. Fortunately, though the grade of the canyon is very steep, it is not interrupted by precipices large enough to cause outbounding or free plunging. On no part of the rush was I buried. I was only moderately imbedded on the surface or at times a little below it, and covered with a veil of back-streaming dust particles; and as tbe whole mass benearh and about me joined in the flight there was no friction, though I was tossed here and there and lurched. from side to side. When the avalanche swedged and came to rest I found myself on top of the crumpled pile wlthout a bruise or scar. This was a fine experience. Hawthome says somewhere that steam has spiritualized travel; though unspiritual smells, smoke, etc. still attend steam travel. This flight in what might be called a milky way of mow-stars was the most spiritual and exhilarating of aU the modes of motion I have ever experienced. Elijah's Right in a chariot of fire could hardly have been more gloriously exciting. (from The Yosemite by John Muir, 1912) Examples of chains of collocation.aJ cohesion are: mcuntaineering . Yosemite ... summit peaks ... climb ... ridge; hours ... whok da)l . (sundown ... sunset ... ) all day ... minute; wallowing ... sinking .. . buried ••. imlwdded; ride ... riding ... ride ... travd ... travel .. . travel . . . flight ... motion ... flight. The analysis .and interpretation of lexical ~tteming of this kind is a major task in the further study of textual cohesion. Here we shall simply group together all the various lexical rdations that do NOT depend on referential identity and. are NOT of the form of reiteration accompanied by the or a demonstrative - in other words. aH lexical cohesion that is not covered by what we have called 'reiteration•- and treat lt under the general heading of COLLOCATtON, or collocacional cohesion. without attempting to classify the various meaning rdatiom that are invo~ved. But it should be home in mind that this is simply a cover term for the cohesion that results from the ccroccurrence of lexical items that are in .some v • .-ay or other typically associared with one another, because they tend to occur in similar environments: the specific kinds of eo-occurrence
288 relations are variable and complex, and would have to be interpreted in the light of a general semantic description of the English language.* 6.5 The general concept oflexical cohesion The suggested framework for the description of lexical cohesion 1s as follows: Type of lexical cohesion: I. Reiteration {a) same word (repetition) (b) synonym (or near-synonym) (c) mperordin~te (d) geru:ral word 2. Collocation Referential relation: (i) same referent (ii) inclusive (iii) exclusive (iv) unrelared The effect of lexical, especially collocationa1. cohesion on a text is subtle and difficult to estimate. With grammatical cohesion the effect is relatively clear: if one comes across the word he, for example, there is no doubt that some essential information is called for, and that the identity of the he must be recovered from somewhere. Reference items, substi- tutes and conjunctions all explicitly presuppose some element other than themselves. In lexi.cal cohesion. however, it is not a case of there being particular lexical items which always have a cohesive function. EVERY lexical item MAY enter into a cohesive rdation, but by itself it carries no indication whether it is fUnctioning cohesively or not. That can he established only by reference to the text. This seems to suggest that what we are ca.lling lexical cohesion carries no meaning; that it is simply an incidental consequence of the fact that discourse does not wander at random from one topic to another but runs on reasonably systematic lines with a certain consistency of topic and predictability of development. In general. of course, this is true; most discourse is well organized, and the patterned occurrence of lexicaJ items is a natural consequence of this. But this does not imply that lexical eo- * For a :mon: extended discussion of this point. and of lexical cohesion in general, sec Ruqaiya Hasan: LAngw<lge in tM Imaginative Conltxl, a S<lti~Jlinguistic muiy -of starif's t..td by thi14ren, Lon- don, Routledge & Kegan Paul (Primuy Socia~ti<m, Language and EducMion. ed Buil. Be=stcln). forthcoming.
6.5 THE GENERAL CONCEPT OF LEXICAL COHESION 289 hesion has no meaning. Without our being aware of it, each occurrence of a lexical item carrie~; with it its own textual history. a particular collo- cational environment that has been built up in the course of the creation of the text and that will provjde the context within which the item will be incarnated on this particular occasion. This environment determines the < instantial meaning •, or text meaning, of the item. a meaning which is unique to each specific instance. In reading or listening to text, we process continuously, and therefore by the time any given lexical item is taken in, its context has already been prepared; and the preceding lexical environment is perhaps the most significant component of this context. It frequently provides a great deal of hidden information that is relevant to the interpretation of the item concerned. There are many examples of this in the long paragraph from John Muir quoted above. To consider just one of these: an inspection of the collocational environment of the item sunset shows that it ties with sundown in the preceding sentence, and less immediately • with the words (long) time ..• hours • .. (whole) a.,r in the slightly less immediate context. These two collocational themes come together in the phrase withjn hnlf an hour of sundown. This environment defines sunset in the context of time, as an event preceded by .a fixed and limited interval, and sets the stage for the passage which serv~ as the immediate environment for sunset, namely in time- to see the sunset. The result is twofold. On one hand, when we meet this phrase in timE to see the sunset we interpret it with what has gone before in mind, and this defines the unique instantial meaning of the word sunset on this oc-casion. On the other hand, the fact that we do this has the effect of making the word sunset. when it does occur, cohesiv"e with the related items that have preceded it, and hence of giving it a significant part in the creation of texture. The lexical environment of any item. includes, naturally, not only the words that are in some way or other related to it, in the terms discussed in this chapter, but also all other words in the preceding p.usage, and aU of these contribute to its specific interpretation in the given instance. But it is the occurrence of the item IN THil. CONTEXT OF RELATED LEXICAL ITEMS that provides cohesion and gives to the passage the quality of text. The relatedness is a matter of more or less; there is no clearly defined cutoff point such that we can say that sunset, for exampJe. is related to just this set of words and no others. But we can say that it is more closely related to some than to others; and it is the closeness of the relationship that determines the cohesive effect. The relative strength of the collocational tension is really a fun<:tion of
290 LBXJCAL COHESION two ldnds of rdatedness, one kind being rdatedness in the linguistic !'.ystem and the other being relatedness in the text. "What we are calling rela- ted lexical items are related in the linguistic system. In the linguistic system there is: a closer relationship between sunset and sundown than, say, between sunset and day; the latter are, in turn. more closely related than sunset and summit, or sumet :!nd m<mntain. although there is some relation- ship here too, less remote than, say • between sunset and sight or sunset and t>stimate. There are degrees of proximity in the lexical system, a function of the relative probability with which one word tends to eo-occur w:ith another. Secondly~ in the text there is relatedness of another kind. relative proximity in the simple sense of the distance separating one item from another. the number of words or clauses or sentences in between. The cohesive force that is exerted between any pair of lexical items in a passage of discourse is a function of their relative proximity in these two respects. There is a very close proximity between sunset and sundown as regards their relatedness in the linguistic system; they are morphologically related, both containing the element .sun, and they are also near-synonyms, sunset referring to a particular event considered as a perceptual phenomenon, and sundown referring to tbe same event considered as defining a moment in time. If tbe two occur in adjacent sentences, they exert a vety strong cohesive force; this would be progressively weaker the greater the textual distance between them. There is a third factor in6uencing the cohesive force between a pair of lexical items in a text, and that is their overall frequency in the system of the language. A word which enters with equal readiness into collocation with words of every possihk range of lexical meaning effects relatively little cohesion with any of them. Words such as go or man or know or way can hardly be said to contract significwt cohesive rdations. because they go with anything at all. Since, roughly speaking, words of this kind are also those with high overall frequency in the language, in general the higher the frequency of a lexical item (its overall frequency in the system) the smaller the part i.t plays in lexical cohesion in texts. When ana.lysing a text in respect oflexica:l cohesion, the most important thing is to use common sense. combined with the knowledge that we have,. as speaken of a language. of the nature and structure of its vocabu- lary. We have a very clear idea of the relative frequency of words in our own language, and a ready insight (if we do not submerge it beneath the weight of the demand for formal procedures of analysis) inro what con- stitutes a significant pattern and what does not. In assessing the lexical cohesion of <1. text we can s.afely ignore, as we certainly would do without
6.5 THE GENERAL CONCEPT OF LEXICAL COHESION 291 even thinking about it, repetitive occurrences of fully grammatical (closed system) items like pronouns and prepositions and verbal auxiliaries, and also oflexical items of very high frequency !ike take and do and good and the others mentioned above. An exception to this appears just when such words occur in special senses. with restricted patterns of collocation; for example takings in the sense of earning~ or gocd in a specificaUy moral context. Again, common sense needs to be brought into play. There i") likely to be tw significant cohesion between two occurrences of goad of which one is in a moral sense and the other an exclamation meaning 'agree'; whereas there might be quite a significant tie between the first of these and a different but related word such as virtue or judgment. In the coding scheme suggested in Chapter 8. we have used a single heading for all instances of coUocation.al cohesion. making no differentia- tion either according to the different kinds of collocational relation or according to different degrees of cohesive force. A full interpretation of lexical cohesion would require further differentiation on both these counn: but such a treatment demands a separate study and is beyond our scope here. There remains one point to be added to round off this limited discussion oflexicat cohesion. A lexical item is not bound to a particular grammatical category, or to a particular morphological form. For example~ there is just one lexical item boy, which has the forms boy, boys. boy's and boys'. Similarly talk, talks, talked and talking all represent a single lexical item talk. There are no perfectly clear criteria for deciding just how far this principle can be extended. For example, go,goes. going. gone and went are all orre lexical item, and so are good. better and be-st; so also presumably are noun and (where these have the sense of ·norm') nominal, nominalize and twminalization. Rather more doubtful are pairs like tooth and dental, map and cartographic, town and urban; even more doubtful, perhaps, a set such as young, youth and juvenile. fu the last resort it does not much matter, since such sets and pairs are cohesive anyway; hut it is ofien possible to be guided by the context - the doubtful cases are generally doubtful precisely because they are sometimes the same word and sometimes not. so that pairs like tooth and dental may he used. either as morphological varianu of the same lexical item or as diiferent lexical items. This, like many other linguistic points, is well brought out by forms of linguistic humour. an expression like tlU' orchiepiscopal gaiters is playing on the fact that archiepiscopal can be interpreted as simpiy a morphological variant of the item arr:hhlshop. although usuaUy it functions as a related but separate :item. On the other side of the line would be pairs like starve and hunger. or
292 disease and ill, which are cdated by synonymy but probably never treated as forms of the same word. The concept of the lexical item, therefore, is not totally clearcut; like most linguistic categories, although dearly defined in the ideal, it pre- sents many indeterminacies in application to actual instances.. Despite this -indeterminacy- and it may be remarked that the term LRXICAL l'l'EM is rather less indeterminate than the folk-linguistic term WORD - it is an essential concept for the understanding of text. However luxuriant the grammatical cohesion displayed by any piece of discourse, i.t will not form a text unless this is matched by cohesive patteming of a lexical kind. A final example: [6: r6] Sing a song of sixpence, a pocket full of rye, Four-and-twenty blackbirds baked in a pie, When the pie was opened, the birds began to sing, Wasn't that a dainty dish to set before a king? The king was in his counting-house, cotmting out his money, The queen was in the parlour, eating bread and honey, The maid was in the garden, hanging out the clothes. Along came a blackbird and pecked off her nose. There is reiteration of the same word, eg: pie , .. pie. king ... king; of a near-synonym, eg: eating ... pecked; of a superordinate. eg: pie . .. dish. sixpence, .. money, blackbird. , . bird; dish might perhaps ako be interpreted as a general word in the modern sense (• anything nice'; cf: dishy). There ls also collocational cohesion. eg: king ... queen, parlour . .. garden, dish ... eat, rye ... hread. The rhyme provides a good illustration of the amount of lexical cohesion, and the varied nature of lexical cohesion, that is charac- teristic of even a very -short text.* * In ~ recent unpublished paper, based on research in spoken discourse, J. McH. SiDcbir mggeru; that p;atterns of lexical ~ohesion across uttet"anCe boundaries m.ay be used by speakers to locateindiYidwl <:onceptua.i tbme$, or ORJFNTATIONS • .By choosing tO repeat the voca- bulary of a pcevioui speUer, one signals willing~ ro negotiate in his tertnS; by usiDg synonyms or paraphrase, one signals the opporite. Words of reference like pronouru, and diiptioU synbX {tg one-word answer~ to questions) reali= othet selettions of orientation.