Topic and the representation o f discourse content

A cco rd in g to de B eau gran d e, this node-sharing is a graphic correlate of T o p ic (19 8 0 : 94). C learly, w hat de Beaugrande understands as to p ic is w hat m ay be described as a topic en tity (see section 4 .3 ). W e have already argued that a discourse top ic is a m uch m ore com plex concept. H o w ever, de B eau gran des claim , based on his an alysis of sim ple text, is indicative of how far it is possible to take an extrem ely lim ited view of top ic w hen the data studied is so lim ited. In fact, we m ight go furth er and state that m uch of the research reported in the literature on issues like top ic, text-structure and text-con ten t has been restricted to such unrepresentative discourse data that the fin d in gs are unlikely to have m uch w ider application in the analysis of discourse. T h e discourse analyst m ay glean useful insights into som e aspects of sim ple text from this research, but he cannot forever restrict him self to in vestigating versions of m aterial like T h e farm er and the don key or T h e rocket in the desert . O ne of the issues de B eaugrande (19 8 0 : 92) show s an aw areness of, but does not in vestigate, is the fact that the heavy use of sentences in com prehension m odels keeps us from addressing the question of how long a stretch of text people actually process at one tim e . It seem s unreasonable to suggest that whole narrative texts, for exam ple, are processed in one single sw eep. I f there are sm aller un its of discourse, w hat are their boundaries like, w hat com ponents do they contain, and how are they internally organised? T h ese are questions w e shall attem pt to answ er in the course of C h ap ter 5.

124

4
Staging and the representation of discourse structure

T h e lin e arisa tio n p ro b le m O ne of the constraints on the speaker / w riter is that he can p roduce only one w ord at a tim e. W hen he orders these single w ords into sentences, and those sentences into texts, he confronts w hat has com e to be called the linearisation p rob lem . H e has to choose a b egin n in g point. T h is point w ill influence the hearer / read ers interpretation of everyth in g that follow s in the discourse since it w ill constitute the initial textual context for everyth in g that follow s. C o n sid er just tw o types of invented exam ples. F irst of all, consider the effect of an identical attributive description being preceded b y differen t evaluative com m ents: (1) a. I cant stand Sally Binns. Shes tall and thin and walks like a crane. b. I do admire Sally Binns. Shes tall and thin and walks like a crane. In a the attributes tall and thin and walks like a crane m ust be assum ed to be unattractive, aw kw ard, un gain ly. In b those sam e properties are now endow ed w ith elegance and grace. C o n sid er next the effect of linear sequencing on the in terpreta tion of events in tim e w here the listener can be expected to derive differen t im p licatu res from differen t ord erin gs (L e v e lt, 1 9 8 1 : 9 1 ) : (2) a. She married and became pregnant, b. She became pregnant and married.

4 .1

T h e re is, as L ev e lt rem in ds us, an ordo naturalis, w hereby it is assum ed that, if there is no cue to the con trary, the first-m entioned event happened first and the second-m entioned event follow ed it. It is, then, open to the hearer / reader to draw im plicatures from that

125

ord erin g, im plicatures w hich w ill be constrained by both the content of w hat is said and stereotypical expectations based on previou s experience (cf. discussion in 2 .4 ). W e are fam iliar, in the field of visual perception, w ith effects p rod u ced b y presen tin g the sam e stim ulus in a different context. A block of colour p roduced in the centre of a light surround m ay be perceived as bein g m uch darker than that sam e block of colour presented in the centre of a dark surround. S im ilarly a line p resented in a given context is perceived as being longer than a < --------------- > >------------------ < line of the sam e length w hich is presented in a different context. In a sim ilar w ay, un derstan d in g of verbal input is processed against the relevant background of the im m ediately precedin g co-text (w ithin , of course, a specified con text). T h e sam e sequence of w ord s m ay take on a differen t v alu e (W iddow son, 19 78) w hen it is uttered in a differen t co-text. W e shall consider this effect, first w ith respect to the internal structure of m essages at the sentence level, and then w ith respect to the organisation of larger stretches of discourse. Them e We shall discuss the linearisation process at this level on ly very b riefly. T h is m eans w e are obliged to cut several corners in our discussion . In particular w e shall speak of the them atic organisation of the sentence. It is im portant to appreciate, how ever, that in com p lex and com pound sentences a separate them atic organisation w ill be assigned to each clause (for an extended discussion of processes of them atisation in E n g lish , see H alliday, 19 6 7 ). It is, fu rth er, go in g to be necessary in this section to cite as exam ples several sets of constructed sentences in order to dem ons trate the p otentially contrastive effects of differen t structures. W e shall use the term theme to refer to a form al category, the left-m ost constituent of the sentence. E ach sim ple sentence has a them e the starting point of the utteran ce and a rh em e, everythin g else that follow s in the sentence w hich consists of w hat the speaker 126 4 .2

states about, or in regard to, the starting point of the utterance (M ath esius, 19 4 2 ). T h e them e, then, is w hat speakers / w riters use as w hat H allid ay calls a point of d ep arture (19 6 7 : 2 1 2 ) . In m any cases (often considered to be the unm arked or neutral cases) the them e of d eclarative sentences w ill be a noun phrase (the gram m a tical su b ject), that of in terrogatives the interrogative w ord , and that of im peratives the im perative form of the verb . In ou r discussion w e shall focus on sim ple declarative sentences and consider their them atic, rather than their syntactic, structure. It is a strik in g feature o f E n g lish , as of m any other languages, that there exists a very w ide range of syntactic form s w hich can be used b y the speaker to con vey the sam e propositional or cognitive content. C o n sid er a few of the syntactic form s available in E n g lish : (3) a. b. c. d. e. f. g. John kissed Mary. Mary was kissed by John. It was John who kissed Mary. It was Mary who was kissed by John. What John did was kiss Mary. Who John kissed was Mary. M ary, John kissed her.

T h e sam e propositional content is expressed each tim e. In each case it is asserted that kissing w ent on and that Jo h n did the kissing and that M ary w as the one w ho w as kissed. If the only reason for having syntactic stru ctu re w ere to perm it us to express propositional content, it is hard to see w h y there should be such an im m ense variety of form s (only a few of w hich are listed above) to perm it the expression of that p ropositional content. W hy do we find this w ide variety of structures? A n um ber o f differen t answ ers to this question have been prop osed. A lice D avid so n (198 0 ) suggests T h e m ore m arked the con struction , the m ore likely that an im plicated m eaning w ill be that w hich the utterance is intended to con vey , w here her own sentence n icely, icon ically, dem onstrates the deliberate w ay in w hich she is m an ipulating the syn tax to m ake her point. She suggests takin g the active form as the norm al, unm arked, form for the declarative sentence and claim s that the passive m ay for exam ple be used to con vey a hum orous or derogatory effect. S o to the question 'D id John kiss M ary? a cautious friend might reply 127

'Well, M ary was kissed by Jo h n .' It is clearly not the case, how ever, that usin g the passive necessarily has a m arked effect. F ro m the discourse an alysts point of view , the most wideran ging and interesting approach m ust be that w hich considers the effect of usin g one sentential form rather than another in the context of discou rse. It is clearly the case that (3a g) could not all function satisfactorily as answ ers to the sam e question. A speaker p rod u cin g these utterances w ould have different assum ptions about the state of know ledge of his hearer, that is about his h earers p resup p ositions. T h u s , in answ er to the question 'What did John do?, (3a) seem s possible and so does (3e), but the rest seem less ap p ro p riate; (3b) seem s to be about Mary rather than Jo h n ; (3c) seem s to im p ly that the hearer already know s that som eone kissed M a ry and id entifies j/Ww as the individual w ho did it; (3d) seem s to im p ly that the hearer know s that Jo h n kissed som ebody and identifies the recipient as M ary (and m ay indeed, w ith contrastive intonation on M ary, indicate that it w as M ary rather than som e b o d y else w ho w as the re c ip ie n t); (3 ^ sim ilarly assum es the hearer know s that Jo h n kissed som eb od y; (3g) seem s m ore appropriate as an an sw er to the question what happened to Mary ? W ith sim ple exam ples like these, it seem s reasonable to suggest that w hat is p rim arily at issue is the judgem ent that the speaker m akes about w hat the hearer believes to be the case w ith respect to w hat he w ants to talk about. H allid ay dem onstrates, w ith an effective exam ple, the dislocating effect on a text of changing the them atic stru ctu re. T h e occasion in each case m ust be taken as one in w hich a reporter is announcing on a radio program m e w hat is happen in g at a reception for three astronauts w ho have recently com pleted a successfu l m ission :
(4) a. The suns shining, its a perfect day. Here come the astronauts. T h eyre just passing the Great Hall; perhaps the President will come out to greet them. No, its the admiral whos taking the ceremony . . . b. Its the sun thats shining, the day thats perfect. The astronauts come here. The Great Hall theyre just passing; hell perhaps come out to greet them, the President. No, its the ceremony that the admirals taking . . .
T h is passage w as p resented b y H allid ay at a System ic W orkshop in the early seven ties. F o r a sim ilar exam ple see H allid ay, 19 78.

128

H ere the speaker in a sim ply asserts a sequence of facts and opinions w hich he thinks w ill interest his listeners. (W e shall not discuss the internal stru ctu re of this sequence of assertions, m erely note that, h avin g set the scene, he clearly expects to report events as they occur in tim e, floatin g opinions w hen nothing of interest is happ en in g.) T h is speakers utterances could be seen as replies to a series of v ery general questions like whats going on?, whats happening now? T h e sp eaker in b on the other hand w ould have to be im p utin g a great deal of know ledge to his hearer. T h e first two clauses appear to answ er questions like whats shining?, ivhats perfect? T h e last sentence appears to contradict a b elief w hich the speaker im putes to his listeners, nam ely that they suppose the adm iral w ill be takin g som ething other than the cerem ony. It is hard for the p rocessor to construct a coherent m odel of w hat is go in g on from the text in (b), even though the propositional content is the sam e as that in text (a) and the cohesive links (see 6 .1 ) are m aintained. T h e problem H allid ay illustrates here is one w hich is fam iliar to m any w riters w ho pause in the m iddle of a paragraph, uncertain how to connect the next thing they w ant to say w ith the last sentence. It is som etim es possible to force a link w ith a connector like however or therefore, but som etim es it is necessary for the w riter to recast his proposed sentence, to reorganise the syntactic expression. W hereas in w ritten language w e generally only see the finished p rodu ct, so that w e have no indication of w here the w riter m ay have m ade such a correction, in spoken language we can som etim es observe a speaker reorganising w hat he w ants to say and thereb y p rod u cin g a differen t them atic stru ctu re: (5) a. (a departmental discussion about spending money) X : there was a gift of about 38 Y : well that isnt a gift + it is earmarked because + well + the money is + in about ig y s some money was . . . b. (a former Minister of Transport interviewed after a motor way accident in fog) I m going to introduce + mm + as a + certainly as a trial a + a measure of segregation + + this w ill - one cannot make it compulsory + because of the difficulties of enforcement c. (conversation between young woman and her aunt) cause there was a man in my fath ers in the Scouts . . .
I 2 ()

hes a county commissioner now . . . and eh one of his oldest scoutmasters . . . In a and b the speakers appear to have em barked on one structure, w hich they decide is un satisfactory, and m odify it in m idstream to produce a differen t them atic organisation. In c a m ore extensive reorganisation takes place as the speaker eviden tly realises that her aunt m ay not have access to the relevant inform ation that her father is in the Scou ts so she stops talking about this m an , announces that her father is in the S cou ts, and then after som e local interaction w ith her aunt, reverts to talking about the m an in his role as a scoutm aster of her fath ers . W hereas we m ay not be able to perceive this self-m onitoring process at w ork in w ritten language, it m ay be dem onstrated, by req u irin g sub jects to choose one of a set of possible continuation sentences, that there are preferred them atic sequences, in some genres of discourse at least, w hich w ill lead subjects to prefer m arked syntactic form s. T h u s , given a constructed text like th is: (6) or c. She was immediately surrounded by journalists. T h e re is a p reference for c as the continuation sentence, rather than b. W e suppose that this is because readers prefer to m aintain the sam e subject (or discourse topic entity - a notion to be developed in the next section). T h e effect becom es even m ore striking if there are no com p etin g anim ates, as i n : d. She was immediately buffeted by the wind. or e. T he wind immediately buffeted her. T h e passive (d) seem s to be the natural choice here. G iv en the choice of an active sentence w hich continues the subject and m arks the them e as agent, there is virtual unanim ity of preferences for the active form : f. All the journalists were immediately smiled at by her. or g. She immediately smiled at all the journalists. Som e recent studies have exam ined the distribution o f some >3 a. The Prime Minister stepped off the plane. b. Journalists immediately surrounded her.

sentential types in discourse genres of differen t kinds (see Jo n e s, 19 7 7 and Prince, 19 7 8 ). It seem s clear that som e sentential types have a p articularly narrow range of distribution . T h u s , in expo si tory prose, w h-clefts, in w hich, as Prince points out, the content of the introductory w h-clause is presupposed in form ation, have a p rivilege of distribution lim ited alm ost exclu sively to three fu n c tion s: (7) a. introducing the discussion as in: What is most striking in the behaviour of newts is . . . What is particularly worrying about the Cabinets view o f collective responsibility w . . . What I m going to talk to you about today is . . . b. summarising the discussion as in: What I have tried to argue then is . . . What we have been considering is . . . c. more rarely, to indicate explicit contrast as in: You may fin d peace in the bosom o f many religions. What is unique about what Christianity has to offer is . . .
W e are gratefu l to M ahm ou d A ya d from w hose an alysis w e draw and from w hose exten sive corp u s of w h -clefts we have borrow ed these exam ples.

W e have proceeded so far on the sim p lifyin g assum ption that the left-m ost constituent in the sentence is the gram m atical subject of the declarative sentence. T h is perm its a sim ple conflation, made by m any scholars, of the categories theme and grammatical subject. T h u s , in discussion of discourse one m ay find the term theme rather than gram m atical subject used (e .g . in C lark & C lark, 19 7 7 ). It is im portant to note that the left-m ost constituent (as in (3g)) is not alw ays the gram m atical subject. It is freq uen tly the case, for instance, in d eclarative sentences, that ad verbs or adverbial phrases m ay precede the gram m atical subject as in: (8) a. Late that afternoon she received a reply paid telegram . . . (64) b. In one place Betty saw the remains of the study safe . . . (64) c. Without hesitating Betty replied . . . (64) d. Then he went on . . . (65) e. In the meantime she would be the better of professional aid . . . An hour later a pleasant-looking middle-aged woman arrived and took charge. (65) (all from Freeman Wills Crofts, Golden Ashes, Penguin Books, 1959)

131

Staging and the representation o f discourse structure

T h e se extracts are from a detective novel w hich constantly thematises tim e ad verbials (as w ell as others). T h e direct link between w hat has gone before and w hat is asserted in the m ain clause of the sentence is then the ad verbially expressed relationship. In extracts from a travel brochure w e find , p redictably, m ore locational ad verbials th em atised : (9) a. b. c. d.

On some islands it is best if you . . . In Greece and Turkey, you are met at the airport . . . In all other places we make bookings . . . At the centres where we have our own representatives

you . . . e. In some centres we have local agents . . . f. On a few islands you have to collect them yourselves. . . (Aegina Club brochure 19 8 1, p. 3) In general it seem s reasonable to suggest that the constituent w hich is them atised in a sentence is, in som e sense, w hat the sentence is ab ou t , regardless of w hether or not the constituent is the gram m atical sub ject. W hen the gram m atical subject is them a tised, this seem s self-evid ent. T h u s in (a) Fred borrowed a hammer from John (b) John lent a hammer to Fred sentence (a) seem s to be ab ou t F red and (b) seem s to be ab out John. W here ad verbials of tim e w ere them atised, as in the exam ples (8) above, the sentence seem s to be about (or, put differen tly, to be an sw erin g the question) w hat happened n ext? W e shall discuss the im plications of this textual stru ctu rin g in the next section. M ean w h ile, w e should note that there is another set of adverbials w hich are freq u en tly them atised but w hich do not contribute to the structure of the discourse in the sam e w ay. T h is set includes what we shall call metalingual comments in w hich the speaker / w riter sp ecifically com m ents on how w hat he is sayin g is to be taken. He m ay com m ent on the structure of w hat he is say in g: let me begin by, first o f all I shall, I shall now turn to, in conclusion, fin a lly , etc. He m ay com m ent on his com m itm ent to belief in w hat he is sayin g: obviously, o f course, clearly as against perhaps, possibly, supposedly , etc. H e m ay produce one from a large n um ber of expressions w hich indicate how the recipient is to tag the content in his m em ory: in confidence, between you and me, frankly, briefly, etc. (F o r an 132

4-2 Theme extended discussion of adverbials of this kind, see B row n & I .evinson, 19 7 8 .) It is clear that this them atised m etalingual com m ent is not to be integrated w ith the representation of content w hich the recipient is con structing. It m erely gives him directions, 111 som e cases about the type and structure of m ental representation he should be con structing (Once upon a timep resum ably instructs the recipient to construct a fairy-tale m odel), in som e cases about the internal structure of the m odel (more importantly), and som e times com m ents on the reliability of w hat is asserted (perhaps). Som etim es, of course, h ed ges of this sort are not them atised but inserted w ithin the sentence, or they follow it, as in: (10) a. Frankly I dont think he will. b. I frankly dont think he will. c. I dont think he will, frankly.

It is hard to m ake jud gem en ts on the effect of differen t placings of ad verbials in sentences in isolation. Som e hearers feel these v aria tions p roduce no differen ce in m eaning, others perceive subtle nuances of d ifferen ce. L ik e m any issues concerning them atisation / linearisation / selection of syntactic structure, this issue is little understood. W e shall assum e in the rest of our discussion that theme is a form al category in the analysis of sentences (or clauses in a com plex or com pound sentence) and, follow in g D anes (19 7 4 ), we shall assum e that it has tw o m ain fun ction s: (i) connecting back and linking in to the p revious discourse, m ain tain ing a coherent point of view servin g as a point of departure for the further m ent of the discourse. 4 .3 develop

(11)

T h e m a tisa tio n a n d sta g in g T h e process of linear organisation w hich we have been exam in in g, largely at a sentential level in 4 .2 , produces the same sort of problem for the speaker / w riter in organising units larger than the sentence. W e m ay talk in general of th em atisation as a discoursal rather than sim p ly a sentential process. W hat the speaker or w riter puts first w ill influence the interpretation of everything that follow s. T h u s a title w ill influence the interpretation of the text w hich follow s it. T h e first sentence of the first paragraph will

133

constrain the interpretation not only of the p aragraph, but also nl the rest of the text. T h a t is, we assum e that every sentence form part of a d evelop in g, cu m u lative instruction w hich tells us how In construct a coherent representation. 4 . 3 .1

Staging

A m ore general, m ore in clu sive, term than thematisation (w hich refers on ly to the linear organisation of texts) is staging' T h is m etaphor is introduced by G rim es in a w ay which seem consonant w ith ou r use of thematisation : E v e ry clause, sentence, paragraph, episode, and discourse is organised around a particulai elem ent that is taken as its point of departure. It is as though tli< speaker presents w hat he w ants to say from a p articular perspective' ( I 9 7 5 : 3 2 3 )- G rim e s is here p articu larly concerned w ith how the linear organisation can be m anipulated to brin g som e item s and events into greater prom inence than others. T h u s an initial main clause w ill, icon ically, refer to an im portant event, w hile follow ing su b sid iary clauses w ill su p p ly su b sid iary inform ation. O ther sclio lars have w idened the application of G rim e s staging metaphoi T h u s C lem en ts ( 19 7 9 : 287) suggests: S tagin g is a dim ension ol prose stru ctu re w hich id entifies the relative prom inence given to various segm ents o f prose d isco u rse. T h is definition opens the door to far m ore than processes of linearisation, and perm its the inclusion w ithin stagin g of rhetorical devices like lexical selection, rh ym e, alliteration, repetition, use of m etaphor, m arkers of empha sis, etc. W e shall use stagin g not as a technical term , but as a general m etaphor to cover the exploitation of such varied pheno m ena in discourse. T h e notion of relative p rom in en ce arising from processes of them atisation and stagin g devices has led m any researchers, p articu larly in p sych olin gu istics, to consider staging as a crucial factor in discourse stru ctu re because, they believe, the w ay a piece of discourse is staged , m ust have a significant effect both on the process of interpretation and on the process of subsequent recall. In 4 .3 .2 w e shall exam ine som e w ork w hich relates to discourse stagin g. 4 .3 .2

Theme as main character I topic entity

In this section, we encounter uses of the term theme quite d ifferen t from the form ally constrained category w hich we

J 34

(follow in g H allid ay, 19 6 7) use to refer to the left-m ost constituent 111 the sentence or clause. W e find theme used som etim es to refer to (lie gram m atical sub jects of a series of sentences as in this rem ark by Kilt/. (19 8 0 : 2 6 ): T h e notion of a discourse topic is that of the 1 ommon theme o f the previou s sentences in the discourse, the topic <in lied from sentence to sentence as the subject o f their predication (our em phasis). T h e sam e term is also used, particularly in the im ycholinguistics literature, to refer not to a constituent, but directly to the referent of the constituent. T h u s Perfetti & G oldm an (1 <>74: 7 1 ) w rite: B y them atisation we m ean the discourse process l>y w hich a referent com es to be developed as the central subject of I lie d iscou rse (our em phasis). 'This latter usage leads n aturally to an interpretation of theme as m eaning m ain character . T h e discourse process of them atisation referred to b y Perfetti & G old m an then leads to the foregrounding of a referent, as d escribed in C h afe ( 19 7 2 ) , w h ereby a particular referent is established in the foregroun d of consciousness w hile other d iscourse referen ts rem ain in the background. T h e fo re ground or th em atised in d ivid u al, as P erfetti & G oldm an em pha sise, m ay be referred to b y a variety of differen t form al expressions. T h u s an in d ividual called D r Jon es can be them atised w hen identified in the discourse b y the expressions as the doctor or the surgeon or he, just as w ell as b y the repetition of the expression D r

Jones.
P erfetti & G o ld m an p erform ed a series of experim ents (19 7 4 ) in w hich they sought to dem onstrate the effect on the recall of sentences occu rrin g in a text of usin g a prom pt w ord referrin g to a them atised referen t, as opposed to usin g a prom pt w ord referrin g to a non-them atised referen t. T h e y w ere able to show that a them a tised referent occu rrin g as syntactic subject w as the better prom pt for sentence recall. S u ch a fin d in g is consonant w ith the observation that pieces of discourse about a m ain character are freq u en tly organised into sets of sentences in w hich the character is referred to b y the noun phrase acting as syntactic subject. A good exam ple of this is the them atisation of B ird ie in extract (14 ) in C h ap ter 3. P erfetti & G o ld m an s results m ay help to explain w hy one basic organisational m ethod for discourse production involves placing the m ain referen t in subject position. Sets of sentences structured in
1 3 5

this w ay m ay be easier to rem em ber. T h is m ethod is favoured by those w ho prepare en cyclopaedic entries, such as ( n ) , in which 'llii N ez P erces are them atised, b y obituary w riters, as in ( 12 ) and I > \ the w riters of ch ild ren s reading books, as illustrated by cx li.n l ( 1 3 ) . (E xp ressio n s used for the them atised referent in each exti.n 1 are italicised.) (n ) N EZ PERCES

The Nez Perces continue to bear the name given them l>\
French fur traders, referring to the custom of piercing then noses for the insertion of ornaments. They belong to I hr Sahaptin language family, in contrast to the other people ul the region, who speak languages of the Shoshonean branch ul the Uto-Aztecan stock. The Nez Perces number more limn 1,500, a reduction of about 2,500 since their first contact with whites. The great majority live on a reservation in Northern Idaho: less than a hundred live on the Colville Reservation 111 Washington. (Encyclopedia of American Ethnic Groups, Harvard Universil y Press, 1 <>8o) (12)

Mr Mitsujiro Ishii Mr Mitsujiro Ishii, who as a former Speaker of the Japanese

House of Representatives was instrumental in staging the 1964 Tokyo Summer Olympics and the 1972 Sapporo Wintri Olympics, died on September 20. He was 92. Ishii had served as Industry and Commerce Minister and in other cabinet posl* under the late Prime Ministers, Shigeru Yoshida, Nobusuke Kishi and Eisaku Sata, before retiring in 1972. He was speaker of the House of Representatives from February 1967 to Ju ly 1969. (from The Times, 25 September 1981)

(13)

Jack goes up the beanstalk again. He comes to the giants house and he sees the giants wife, (from Jack and the Beanstalk, Ladybird Books, Series 777)

T h e organisation of stretches of spoken discourse can follow a sim ilar pattern, as show n in extract (14 ) . (14) P: did you have any snow + during the holidays R : there was some actually on + at Hogmanay because we had some friends + a Greek friend of ours was visiting us and when he left the house + just after Hogmanay + you know he had been away about fifteen minutes then he rang the doorbell again + he said its snowing its snowing + he was really excited you know +

136

II we look at extract ( 1 3 ) as a set o f clauses, we can follow a method suggested by D an es (19 7 4 ) and represent its stru ctu re in tin follow in g w ay :

11 s )

Them ei (Jack) - Rheme! (goes up the beanstalk again) Them ei (he) - Rheme2 (comes to the giants house) Them ei (he) - Rheme} (sees the giants wife)

In each of the clauses of this piece of discourse the them e, or the larting poin t , is the sam e. I f we w ish to claim that the referen t luck is the them e of the discourse, we m ust be aware that we are Iusing this claim on the fact that Ja c V is them atised in each of the 11 ,1uses in the discourse. It is on the basis of discourses w ith this type o f fixed stru ctu re that the term them e seem s to have com e to he used as a general term in discourse analysis for m ain character iind sentential subject (as w ell as the left-m ost constituent). T h e possibilities for confusion w ith this varied use of the same term inology are ob vio u s. W e have already said that we shall reserve Ilie me for the form ally left-m ost sentential constituent. F o r the 'm ain character / object / id ea notion - exem plified by the referents 'M r M itsu jiro Is h ii in ( 12 ) and Ja c k in ( 1 3 ) we shall use the term w riters / sp eakers topic entity. In those exam ples the text was very sim p ly constructed so that the topic entity w as form ally thematised in each sentence. In the next extract we shall observe that an expression referrin g to the w riters topic entity is form ally the theme o f som e sentences, but not of all sentences. (W e shall not exam ine clauses in non-sentence-initial position, since that w ould involve a w ide-ran gin g technical discussion of them atisation which we have no space for here, but see H allid ay, 19 6 7 .)

( 16)

Mr William Serby Mr William Serby who died aged 85 on September 20 was

County Treasurer to Buckinghamshire County Council from 1929 to 19 6 1. He was commissioned in the Queens (R. W. Surrey Regi ment) in 19 15 and served in France until he was wounded in 19 16 . From 19 17 to 19 19 he served as liaison officer with the French and Russian forces in the North Russian Expeditionary Force. In 1926 he was appointed County Accountant to the Corn wall C .C .
1 3 7

During the Second World War he commanded the IIimiim Guard in Wendover and in later years was actively concri m il with the work of the R .N .I.B ., the Oxford Diocesan Houtd ill Finance, the Bucks Historic Churches Trust and in many lm til organisations in Wendover. In 1926 he married Jean Durns and they had one son >nnl two daughters. (from The Times , 25 September 1 < > H1) O bituaries such as extract (16 ) p rovide particularly clear exam pies of d iscourses w hich have only one w riters topic cut il v throu ghout. In this case, it is M r W illiam S e rb y . A n expression referrin g to this in dividual is them atised for the w hole discourse 111 the title, and for each o f the first and second paragraphs where expressions referrin g to him are m ade the starting poin t for wh.il follow s. T h e w riter m ight have continued, as in extract ( 13 ) willi expressio n s referrin g to the sam e individual them atised in each sentence and p aragrap h. In each of the subsequent sentences and paragrap h s, how ever, the w riter them atises a tim e-adverbial phrase. W e could say that, although the w riter continues with the sam e topic en tity, he organises w hat he w ants to say about this topic entity accordin g to differen t (tem porally determ ined) pers pectives on the in d ivid ual concerned. T h e them atised elem ents do not sim p ly produce a chronological list, but provide different points o f d ep artu re for con sid erin g the individual in different roles. It m ight be objected that the term topic en tity is unnecessary and that w hat w e are talking about here is sim p ly our old friend to p ic . W e insist that it is useful to distinguish betw een the topic entity / m ain character notion and the general pretheoretical notion of top ic as w hat is bein g talked ab out . One w ould hardly w ant to say that the to p ic of an ob itu ary w as the m an referred to b y the nam e at the top of the en try, except in speaking in som e kind of shorthan d. T h e re are m any aspects of the m an , physical character istics for instance, w hich w ould hardly be considered to be appropriate aspects for inclusion in an ob itu ary. T h e top ic of an ob itu ary m ight be m ore ad equately characterised in som e such term s as an appreciation of the notew orthy events and deeds in the life of X .

i 38

Titles and thematisation W e argued in C h ap ter 3 that the title of a stretch of ill i nurse should not be equated w ith the top ic but should be M'Khiiled as one possible expression of the topic. W e now w ish to IIIt|tose that the best w ay o f describ in g the fun ction of the title of a iliM i uurse is as a p articu larly pow erful them atisation device. In the llllc of extract ( 16 ) , the topic entity w as them atised, or, to express 1 In relationship m ore accurately, w hen w e found the nam e of an individual them atised in the title of the text, w e expected that Individual to be the topic entity. T h is expectation-creating aspect of 1 I n m atisation, esp ecially in the form o f a title, m eans that them a(iticil elem ents provid e not on ly a startin g point aro un d w hich w hat I> 1lows in the d iscourse is stru ctu red , but also a starting point which con strain s our interpretation of w hat follow s. T h is point m ay In illustrated b y usin g part of a text constructed b y A nderson et al.
I | .j

("> 7 7 : 37 2 )> ar*d reprodu ced here as ( 17 a and b ). (W e have provided the title in each case.) ( 1 7a)

A Prisoner Plans His Escape

Rocky slowly got up from the mat, planning his escape. He hesitated a moment and thought. Things were not going well. What bothered him most was being held, especially since the charge against him had been weak. He considered his present situation. The lock that held him was strong, but he thought he could break it.

The topic-entity of this fragm ent is the in d ivid u al nam ed R o ck y and, because of the them atised expression in the title, we can read this text w ith the interpretation that R o ck y is a p rison er, in a cell, planning to break the lock on the door and escape. In an exercise w hich we conducted usin g this text after w hich su b jects w ere asked to answ er several questions, we found that there w as a general interpretation that R o ck y w as alone, that he had been arrested by the police, and that he disliked bein g in prison. W hen w e presented exactly the sam e questions to another group who read the follow in g text, ( 17 b ) , w e received quite different answ ers. (17b) A Wrestler in a Tight Comer Rocky slowly got up from the mat, planning his escape. He hesitated a moment and thought. Things were not going well.

139

What bothered him most was being held, especially mm > iIn charge against him had been weak. He considered hi pi m ill situation. The lock that held him was strong, but he lliiiuulil he could break it. In an sw erin g questions on this fragm ent, subjects indicated I lull th ey thou ght R o ck y w as a w restler w ho w as being held in some Kmil o f w restlin g h o ld and w as p lann in g to get out of this hold. Koi ky w as not alone in a prison cell and had had nothing to do with tin police. B y p ro vid in g differen t startin g p oin ts in the them alm nl elem ents o f the differen t titles, w e effectively constrained the wity in w hich the piece of text w as in terpreted. (A nderson et al. ( i <>77) d iscu ss the differen t possible interpretations of the one piece ol test (w ithout titles) presented in (17 a ) and (17 b ) in term s o f knowledge stru ctu res or schem ata w hich are activated for the interpretation ol texts. W e shall discuss schem ata and related concepts in m ore detail in C h ap ter 7 .) E x tra c ts (17 a ) and (17 b ) provide a particularly dram atic illustni tion o f the effect of them atisation. T h e re are, of course, m any otln i easily recogn isable them atisation devices used in the organisation ol d iscourse stru ctu re. P lacin g headings and sub-headings w ithin .1 text is a com m on them atisation device in technical or public in form ation d ocum ents. It also occurs, you w ill have noted, in linguistics textbooks. W hat these them atisation devices have in com m on is not only the w ay they p rovide starting p oin ts for p aragrap h s in a text, but also their contribution to d ivid in g up a w hole text into sm aller chunks. T h is chu n kin g effect is one of the m ost basic of those achieved by them atisation in discourse.

4 .3 .4

Thematic structure

In (8) and (9) we dem onstrated the possibility of ele m ents other than the gram m atical subject occurrin g as the form al them e o f the sentence or clause, b y con siderin g a set of them atised ad verbial p hrases of tim e w hich occur in a detective sto ry, and ad verbial p hrases o f place w hich occur in a travel brochure. In ( u ) w e exem p lified the structure of an encyclopaedic entry w here the gram m atical sub ject, referrin g to the w riters topic entity, was con sistently m ade the them e of succeedin g sentences. T h e n in (16 ) we discussed an ob itu ary in w hich there w as an in terw eaving of 140

lie mi'S w hich related to the in d ivid ual, the topic entity, and them es which related to differen t tem poral fram es, an interw eaving which pt lim ited the w riter to represent his topic entity from different temporal persp ectives. H ere we consider the them atic structure of I hi ci* further passages: 11M)

This rug comes from the village of Shalamazar in the southern Chahar Mahal, but the design is woven in many of the villages. The design is one of those that fit into several possible
categories, involving as it does elements of bird, tree, vase and prayer types. The prayer mihrab may be omitted in some cases, but the vase is always present, as are the strikingly drawn birds . . . In rugs of this type excellent natural dyestuffs are very often found, and the quality varies from medium to quite fine.

Outstanding examples . . . (P. R. J. Ford, Oriental Carpet Design, Thames and Hudson
i 98 l > P- 3) <)bserve the seq uen ce: This rug (illustrated) the design T he design T he prayer mihrabl , , .. , , . r J f (details of design) the vase J ' b In rugs of this type the quality Outstanding examples T h e them atic organisation here gives a clear identification of (i) (ii) the w rite rs topic area the organisation of the p aragraph, m oving from a p ar ticular exam ple of a ru g typ e, through characteristic design, to generalisations about rugs of this type.

T h e them atic stru ctu re of the extracts w e have exam ined so far is relatively help ful to the identification of topic area and the organisa tion of stru ctu re. O ther cases are far less clear. Jo u rn alistic prose is often far m ore loosely stru ctu red : (19)

Due in the bookshops soon from Faber and Faber is a small

paperback which reveals more about the way British television drama is really produced than all the weekend symposiums and university gabfests I ve attended in the last ten years. It consists of seven chunks, one each from . . . The title, Ah

Mischief, comes from Hares contribution. He tells ol unlllg

nervously to visit . . .

(The Listener, 29 April 1982, p 14)

T h is set: Due in the book shops It The title He

m akes it clear that the w riters topic area is concern with a bonlt, though the stru ctu re o f the contribution is m uch less clnn ls m arked. It is possible, though it w ould need to be demonstrate)I, that less clearly m arked structure is m ore d ifficu lt for a recipient In process. T h e analysis of them atic structure in spontaneous speech pro vid es considerable problem s. W e have glossed over som e of the p roblem s of attribution of them atic structure in w ritten language, and w e shall do so again as we encounter spontaneous, converse tional speech. M u ch o f w hat is said is not readily related to the syntactic categories sen tence or clause (contra L a b o v (1966 ) who reported that about 7 5 % of utterances in most conversations are w ell-form ed b y an y criterion (w hen rules of ellipsis and general editin g rules are ap p lied, alm ost 98% w ould fall in this category) cited in L in d e & L a b o v , 19 7 5 ). In the follow in g extract, an attempt has been m ade to assign them atic stru ctu re: (20)

the environment I was living in was Berkeley + which is purely academic + no it wasnt purely academic it was em + it was basically academic I mean most of Berkeley is the university + its like a town + in which the university dominates the city + like Cambridge + or Oxford + the university is the hub of the city + and most of the people you found there kind of ancillary to the university + + em and you also got a lot of wasters there I mean people who dropped out of university and cant bear
leaving the place +

T h e them atic fram ew ork here is generally less sp ecific: the environment I was living in which no it it it I most of Berkeley 142 it in which the university most of the people you you I people

A (M M n it is possible to discern the speakers topic area but it is not MlMHiltle to discern the d evelop in g organisation of the text by Ithilt 11ig at the them atic stru ctu re. It is a characteristic of p rim arily t u t ' motional conversational speech in our data that the interactional |M|hvI, m arked by I and you, is frequen tly them atised (cf. also (2) In ( hiipter i , description of a rain bow ). T h is m arking gives a clear lllihriilion of the speakers view of w hat he is using language to 1In. I hematic organisation appears to be exploited by speakers / \vi Hers to provide a structural fram ew ork for their discourse, w hich 1 rliilcs back to their m ain intention and provid es a p erspective on svlmt follow s. In the detective sto ry cited in (8), the w riter shuttles .(limit, com m enting on the activities of a num ber of different individuals, located in differen t parts of E n glan d and E u rop e w illiin the space of two pages. T h e coherence of structure is im posed, p artly at least, because locally w ithin the text the author is m eticulous in relating events to each other in tim e. E ach new miverbial phrase m arks the fact that the scenario has shifted. T h e n levance of the various activities to each other, or to the plot, is not plain to the reader at this point. H e has to trust the w riter to restrict him self to the account of relevant activities and his w arrant for this unstained trust is that the author goes to such trouble to sp ecify the c om plex tem poral relationships of the activities he describes. In the travel brochure (9), w hat is essentially an un structured list of facts is given structure and arranged into paragraphs on the basis of different geographical locations. It is the differen t geographical locations, realised in them atic position, w hich form the fram ew ork of the discourse. A n yon e w ho has ever w ritten an essay is fam iliar w ith the problem of w here to start the essay, how to relate paragraphs to what has gone before, and how to relate sentences to w hat has gone before. W e all freq u en tly encounter prose w here the w riter has not paid su fficien t attention to them atic organisation. C o n sider this citation on the w rap p in g of a S w iss L em o n Oil soap tablet: (21) Li-mang is what the Chinese called the citrus lemon tree in 1 1 75 a d and some believe the Mongolians invented lemonade in 1099. Lemons, like other species of citrus fruits, have been cultivated for thousands of years and are native to Southeast ern Asia.
H

'Staging and the representation o f discourse structure

T h e re is m ore than one problem here, but one reason w hy this text reads rather odd ly is because of the them atised some in the co-ordinated second clause of the first sentence, follow ing the m arked stru ctu re in the first clause.

4 .3 .5

N atural order and point o f view

W e have already m entioned the notion of a natural o rd e i' for the presentation of a narrative sequence of events. A s L evel! ( 1 9 8 1) rem arks, it is natural to put the event that happened first before the event w hich follow ed it. A sequence of events in time, told as a n arrative in E n g lish , w ill often be presented in the order in w hich they happened and, often, w ith an unstated im plication of 1 relationship in w hich the second event in som e sense follow s from the first (e.g. was caused by). T h is type of non-logical inference has been characterised b y H orn (19 7 3 ) as post hoc ergo propter hoc. C o n sid er the follow in g passage. Ju s t before it begins, a violent storm has broken, w ith torrents of rain : (22) Between where I stood by the rail and the lobby was but a few yards, yet I was drenched before I got under cover. I disrobed as far as decency permits, then sat at this letter but not a little shaken. (W. Golding, Rites o f Passage, Faber & Faber, 1980, p. 191)

It is not stated that the narrator is dren ched by the rain (rather than b y , say, p erspiration) or w h y he w ishes to get under cover. It is not m ade clear w h y he disrobes or w hy he finds him self not a little shaken . T h e norm al assum ption of an English-speaking reader w ill be, how ever, that the series of events are m eaningfully related to each other, and he w ill draw the appropriate inferences that the narrator is drenched b y the rain, w ishes to take cover from the rain, disrobes because his clothing has been drenched by the rain , and is not a little shaken because of his im m ediately p reced in g experien ce in the violent storm . (F o r a discussion of in feren cin g see C h ap ter 7 .) W e stress that these inferences will In draw n b y an E n glish -sp eak in g reader because it appears that in other cu ltures there are rather d ifferent bases for narrative stru c tures (cf. G rim e s, 19 7 5 ; G rim es (e d .), 19 7 8 ; B ecker, 19 80). It is clearly the case that there are stereotypical orderings in genres other than those w hich ob vio u sly consist of a series o f events
H

4 .3 Thematisation and staging in lim e. T h u s L in d e & L a b o v report that 9 7% of the sub jects, in a 111 vey in w hich sub jects w ere asked to describe the lay-out o f their apartm ents, d escribed them in term s of im aginary tours w hich liiinsform spatial lay-outs into tem porally organised n arratives (19 7 5 : 924). T h e narrative tour in each case begins at the front iloor, just as it w ou ld if the in terview er w ere to arrive for the first I line at the apartm ent. A sim ilar alignm ent w ith the point of view of I lie hearer is taken by speakers w ho are asked to give directions in a nlrange tow n. T h e y alw ays begin, c o o p e ra tiv e ly , from the point where the en q u iry is m ade and then attem pt to describe the route as ii succession o f acts in tim e. In each of these cases then, there is a mitural starting point and the description is an attem pt to follow a natural progression. L e v e lt suggests that by adopting the stereoty pical pattern of the cu lture the speaker facilitates the listen ers com prehension ( 1 9 8 1 : 94) since both speaker and hearer share the mime stereotype. It seem s very likely that there are other constraints on ordering in types of discourse w hich are not sim p ly arranged as a sequence of events in tim e. V an D ijk (19 7 7 ) suggests that descriptions of states 11I affairs w ill be determ ined by perceptual salience so that the m ore milient entity w ill be m entioned first. H e suggests that norm al nrdering w ill con form to the follow in g pattern: Ut ) general - particular whole - part / component set - subject - element including - included large - small outside - inside possessor - possessed (van Dijk, 1977: 106) C o n sider the follow in g extract in term s of van D ijk s proposed constraints: (44) 'It was indeed a horrifying sight. 2T he walls alone stood, bare and gaunt and blackened, with cracked and split stone dres sings and gaps where the cornice had been dragged away when the roof collapsed. 3Within were heaps of wreckage, mostly brick and stone from internal walls which had fallen, but with occasional objects of twisted metal and quantities of broken glass. 4In one place Betty saw the remains of the study safe and
1 4 5

in another three stiek-like objects which she eventually classi fied as the barrels of shot guns. (Freeman Wills Crofts, Golden Ashes, Penguin Books, 1959, p. (>4) T h e first sentences describe the general, whole, large, outside T h e second sentence m oves to the w alls, part, including, large, outside. T h e third sentence m oves inside and begins to observe small included objects. T h e last sentence introduces small included particulars , initially un identified and then p articularised. It seems reasonable to suggest that, in general, in this descriptive passage the con strain ts w hich van D ijk proposes are adhered to. Sup p ose the constraints w ere not adhered to? It w ould follow then, that when the n o rm al ord ering is reversed , som e special effect (staging device, im plicatu re) w ou ld be bein g created b y the speaker / writci V an D ijk suggests that if the norm al ordering (gen eral-particular) is reversed as betw een (25a and b) the second sentence in (b) will be taken as g iv in g an explanation for the state of affairs described in the first sentence: (25) a. Peter always comes late. He wont be in time tonight eithci b. Peter was late again. He never comes on time.

V an D ijk s suggestion is certain ly of interest to the discourse an alyst. W e should, how ever, take note of L e v e lts w arn in g: the . . . question of how natural order relates to differen t dom ains ol discou rse, w ill n ever be answ ered exh au stively: there are as many natural orders as there are things to talk ab out ( 1 9 8 1 : 94). O ne ob vious constraint on ord erin g w hich m ay override the perceptual salien ce principle outlined by van D ijk , is the main tenance of a consistent point of view . F illm ore ( 1 9 8 1) has noted that a feature of literary discourse is the effect of a particuliu orientation or angle of vision on the w ay events are presented. T h u s , at the begin nin g of H e m in g w ays The Killers, the w ay the reader has to view the events is determ ined by the organisation ol the first sen tence: (26) The door of Henrys lunchroom opened and two men came in

T h e op en in g of the d o o r takes place before the appearance of the two m en. T h is ord erin g of events is com patible w ith the fact that the m en cam e in . T h e stru ctu re of this fragm ent reflects the view

of even ts w hich a narrator inside the lunchroom m ust have had. In another of F illm o re s exam ples, (27) below , the sequential stru c ture o f reported events is determ ined by the o rd er in w hich they were ob served , rather than by their m ost natural p hysical sequence: (27) The light went on. She was standing by the door.

In literature, the author freq uen tly assigns the role of narrator to one of his characters. T h e author then has to m anipulate the know ledge w hich the reader needs so that it can be p lau sibly known to and recountable b y the narrator. Several au th ors have explored the literary possibilities of recounting the sam e even ts seen through the eyes of differen t characters and interpreted d ifferen tly by them (cf. for tw o v ery d ifferen t literary typ es, The Moonstone b y W ilkie C o llin s and The Alexandrian Quartet b y L a w re n c e D u rrell). T h e problem for the author is to create a coherent v iew of a particular w orld. T h e problem is, of course, one w hich affects all our production of language. K u n o (19 7 6 ) and K u n o & K a b u ra k i (19 7 7 : 627) have pointed out that the variation in w hat they call cam era angles has an effect on the syn tax of sentences. I f the sp eaker is em pathising with one p articipant in a dom estic dram a rather than another, the same event m ay be described for exam ple b y sentence a or by sentence b : (28) a. John hit his wife. b. M arys husband hit her.

T h e sp eakers em pathy, his sym p ath y w ith one point of view rather than another, m ay also lead to a p articular choice of lexis. C onsider the follow in g p arad igm : (29)a. Mary, Queen of Scots, was executed by the English Queen. b. M ary, Queen of Scots, was assassinated by the English Queen. c. M ary, Queen of Scots, was murdered by her cousin, Elizabeth. In each case the agent referred to is the sam e in d ivid u al, the patient referred to is the sam e in divid ual, and the agent causes the patient to die. (T h e cogn itive content m ight be held to be the sam e.)

Staging and the representation o f discourse structure

H ow ever, in a the action is reported as a legal process (executed) sanctioned b y the constitutional m onarch ( the English Queen). In h the action is reported as an illegal, politically m otivated act (assassinated) sanctioned by the constitutional m onarch (the E n glish Queen). In c the action is reported as an illegal, crim inal act (murdered) perform ed by a close relation (her cousin Elizabeth). In each case the w riter reveals a differen t assessm ent of the character and m otivation o f the act. (F o r a discussion of the w ide range of factors in flu en cin g lexical choice in discourse, see D ow n in g, 19 80.) T h e question o f em p ath y , described by C h afe as arising because people are able to im agine them selves seeing the w orld through the eyes of others as w ell as from their ow n point of v iew (C h afe, 19 7 6 : 54), takes us far beyond the relatively form al investigation of the effects of them atisation into the general area of stagin g w hich we return to in the next section.

4 .3 .6

Theme, thematisation and staging

T h ro u g h o u t this chapter w e have attem pted to draw a distinction betw een the linearisation problem in term s of the cogn itive o rd erin g o f events, description , etc. and the linearisation p roblem in term s of the linguistic m eans available to the speaker / w riter for exp ressin g that cognitive stru ctu rin g, particularly the them atic organisation of the sentence or clause. T h e distinction is, how ever, d ifficu lt to m aintain for the obvious reason that our only access to the sp eakers / w rite rs cognitive stru ctu rin g is via the language w hich he uses to express that stru ctu rin g. W e have assum ed that the notion of stagin g em braces a much w id er field , facets of w hich w e have only b riefly discussed. It em braces on the one hand the sp eakers / w riters overall rhetorical strategy of presentation w hich m ay be m otivated b y an intention to create suspense, to convince his listener of the truth of what he is sayin g b y ad ding credib le sup p orting details, to persuade his listener to a course of action, or to shock or surp rise. Indeed .1 speaker / w riter m ay sim ultaneously have all of these intentions T h e stagin g w hich the discourse analyst m ight be concerned with is that w hich is m anifest in the language used. It m ust be obvious, ho w ever, that, w hereas the discourse analyst can draw attention to the effect of p articu larly m arked staging, his discussion of tin
148

4 .3 Thematisation and staging e ffe c t of stagin g, or indeed his abstraction of som e p articular linguistic form s rather than others, as con tributin g to that effect, will n ecessarily be fairly unconstrained, in m any w ays akin to traditional literary interpretation or rhetorical discussion. M ost linguists w ill feel uneasy at this soft extension in the discussion of discourse. N otw ith stan d in g, it is clear that discourse analysts can contribute to a description of the staging of the follow in g extract: (jo) B : I think if your physical appearance is em sort of neat and + well-controlled and so on this gives at least a superficial + feeling that ones going to give a neat and well-controlled performance L : thats right + do you know I remember something which er points this up very well something that Gill said + and its now I suppose er + er eight years ago + when + em + whats that Russian chaps name who was here for a while B : Shaumyan L : Shaumyan yeah + when he was here + em I gave a seminar on phonetics and the brain + which I later wrote up in Work in Progress but never did anything + with em + to my slight regret + but I - that was the first time I d ever ventured beyond as it were orthodox phonetics in - in public + em and it was in front of + our department after its first amalgamation I think + and er one or two people from outside were also present + so in very many ways I was before an unknown public + relatively speaking er talking about somewhat unfamiliar territory of a very speculative nature + but claiming expertise + and I remember that one of the things I did was buy a new pair of shoes

Wc could characterise this as a discussion of the beneficial effects of .1 good physical appearance on the confidence of som eone addres sing an audience. W hereas speaker L is certain ly sayin g som ething about this, we can note that he is presenting a detailed, structured orientation from w hich his com m ent on the m atter has to be appreciated. H e first establishes a p articular tim e co-ordinate relating to a place co-ordinate, selecting a m eans of fixin g the tim e which in volves an elem ent fam iliar to his interlocutor (w hen Shaum yan w as h ere) w hich m ay have the effect of rem inding his interlocutor of how m uch youn ger, less experienced or less con fi dent they w ere at that past tim e. T h e im pression is further elaborated by details of w hat the speaker talked about (speculative, 149

n ever-p u b lish ed ) and to w hom (not close colleagues but outsiil I e rs). T h is structured accum ulation of elem ents con tributin g to it 1 lack of confidence is counterbalanced by a single act of confidence- 1 boosting, presented as the final com m ent. W e could point to the 1 fairly com p lex syntactic stru ctu re of the earlier part of the fra^ I m ent, w ith a n um ber of subordinate clauses adding extra detail. Wc could point to the typ ically p olysyllabic lexis of m ost of tin1 fragm ent and the sudden transition to the sim ple, m onosyllabic one o f the things I did was buy a new p a ir o f shoes w ith its parallel 1 transition out of p rofession al life into the everyday life of the High 1 S treet. W e cou ld describe the change of voice quality on that last quoted phrase, the overall raisin g in pitch , the breathy voice, the I effect of sm iling. It is p resu m ab ly the case that these details noted by the discourse 1 ] analyst are relevant to his interpretation o f the fragm ent. T h e 1 problem w ith a com p lex fragm ent o f this sort, w hich has no ] n ear-parallels in m ost of ou r data, is that w e can only brin g the most general notions of reg u larity to bear on it, notions no m ore specific than those found in any general m anual of rhetoric. In the present state of know ledge it seem s to us w ise to restrict the discussion of general stagin g processes, in the analysis of discourse, to data which consists of m ultiple realisations o f strictly com parable data like d escriptions of apartm ent lay-outs (L in d e & L a b o v , 19 7 5 ), rej tellin gs of n arrative events (G rim e s, 19 7 5 ; C h afe, 19 7 9 ; Chafe (ed.) 1980) or in structions to perform a task (G ro sz, 19 7 9 ; Y u le , 19 8 1 ) . O ne form of strictly com parable data w hich is readily available for an alysis in everyd ay language can be found in letter-w riting. T h e stagin g of letters in term s of w hat inform ation is represented and how it is them atised depends on the type of letter and the intentions of the w riter in w ritin g it. In m ost letters the basic elem ents w hich are them atised are those p rim ary (contextual) features of tim e, location and addressee w hich we considered earlier as con stitu tin g part of the topic fram ew ork. T h e m ore form al the letter, the m ore explicit is the inform ation contained in these them atised elem ents. If we com pare the beginnings of two letters, extracts ( 3 1) and (3 2 ), w e can see that the sam e type of inform ation is them atised, but the am ount of explicit inform ation d iffers. T h e inclusion of the ad dressees full address in (32) is not, ob viously,
1 5

 Ir n d e d to in fo rm the addressee w here he him self lives, but murk the letter as one of a filed series in w hich this inform ation |mi nerved, as a form al letter. T h e sp ecificity of inform ation |i'|ticHenting the w riters address is one conventional m eans Him king the startin g p oin t for the discourse w hich follow s. IlD

to is in of

Davies's Educational Services Ltd., 66 Southampton Sow, London W C 15. i+BY 18th August, 1978.

ileorge Yule, Esq., Department of Linguistics, The University, Edinburgh, 8. Dear Mr. Yule, In your letter of the 10th July, 1973, you told me that you were arranging for me to be sent the ELBA material in response to my letter of the 3rd March ...

A s soon as one begins to investigate the differen t form ats used in letter-w riting, general sim ilarities in the type of inform ation them atised are d iscern ible, but the variety of staging considerations to be recognised is extrem ely large. T h e envelope, plain brow n official or light blue personal, airm ail or official p aid , is just one part of the com plex stagin g w hich precedes the reading of the actual contents ! 5I

of a letter. C learly , m uch of this type of stagin g is non-lin^iiiM lI (part of the external context of discou rse), but its effect on nl|fl interpretation of the text of letters and m any other type* n| d iscourse should not be ignored.

5
Information structure

T h e stru ctu re o f in fo rm atio n In the previou s chapters w e have been considering M i< icasin gly restricted view s of the production and interpretation of <1 1 4imrse. In C h ap ter 2 w e considered the effect of situational Hiiilext on discourse and in C h ap ter 3 the effect of differen t |h iHpectives of topic stru ctu re. We devoted C h ap ter 4 to discussing 11ir effect of linearisation in discourse, how w hat is presented first limits the interpretation of w hat follow s and how decisions on |Ik m atisation provid e the overall structure w ithin w hich the mliIressee interprets the discourse. In this chap ter we focus in even fu rth er, to the sm allest units hI discourse stru ctu re: sm all local units at the level of phrase or 1 lause. W e con sider how inform ation is packaged w ithin such sm all ulructures and, p articu larly, w hat resources are available to speak ers and w riters for indicating to their addressees the status of inform ation w hich is introduced into the discourse. 5.1.1

5. i

Information structure and the notion g iven / new in intonation

T h e serious stu d y of inform ation structure w ithin texts was in stituted b y scholars of the Prague School before the Second W orld W ar. T h e y studied w hat they called the com m unicative dyn am ism of the elem ents con tributin g to a sentence, w ithin the fram ew ork of functional sentence p ersp ective . (F o r an overview of this w ork see V ach ek, 19 6 6 ; F irb a s, 19 7 4 .) M an y of the insights developed by the P rague scholars w ere first brought to the attention of W estern scholars b y H alliday in an extrem ely in fluential article p ublished in 19 6 7 . H alliday elaborated and developed those aspects of P rague w ork w hich related directly to his ow n interests in the structure of texts. In particular, he

153