Discourse organisation through Theme position

DISCOURSE ORGANISATION THROUGH THEME POSITION Lydia-Mai Ho-Dac, Uni e si é Toulouse2 le Mi ail, F ance This pape ocuses on he ole o elemen s placed in he ini ial posi ion i.e. elemen s ul illing he ole o Theme in discou se o ganisa ion. The la ge-scale co pus s udy p oposes a new me hodology based on quan i a i e analysis o he discou se oles o sen ence-ini ial elemen s. The heo e ically-based hypo hesis is ha Theme posi ion has an impo an unc ion in discou se o ganisa ion. Theme, de ined as he s a ing poin o he message, is composed o he i s elemen s ha he eade pe cei es. The analysis o he dis ibu ion and he use in discou se o hese elemen s gi es us a g ea o e iew on he ex ual o ganisa ion o di e en ypes o ex . KEYWORDS: discou se o ganisa ion, Theme posi ion, da a-d i en app oach 1 THEME AND DISCOURSE ORGANISATION The objec o his s udy conce ns wha I call ‘discou se o ganisa ion’ o ‘ ex o ganisa ion’. These e ms – conside ed he e as synonymous – con ey he idea ha ex is no a bag o wo ds, a bag o sen ences, a bag o pa ag aphs bu should be seen as a s uc u ed objec . This s uc u e is he esul o he s a egies used by he eade when he p oduces his ex . Two main s a egies a e iden i ied in my s udy: s a egy o con inui y and s a egy o discon inui y. The w i e may wan o link wo uni s by a con inui y ela ionship (s a egy o con inui y) o may wan o indica e a shi be ween wo uni s (s a egy o discon inui y). One o he consequences o his is ha ex is segmen ed in in e ela ed ‘chunks’ which may be embedded in a hie a chical ela ionship. My claim is ha Theme plays a c ucial ole in signalling hese wo s a egies. This claim is based on a posi ional de ini ion o Theme ha co esponds o he beginning o ex ual uni s (sen ences, pa ag aphs, sec ions) and on he idea ha he beginning o ex ual uni s is a good loca ion o indica e i he e is a discon inui y o a con inui y. Mo eo e , du ing eading, we build some assump ions based on he i s elemen s pe cei ed, assump ions ha may o ien he in e p e a ion o he es o he uni . Theme posi ion co esponds he e o he en i e p e e bal zone as s a ed by Be y (1995), F ies (1995) o Ma hiessen (1995). When I apply his de ini ion on pa ag aphs o sec ions, Theme co esponds o he p e e bal zone o he i s sen ence o he ex ual uni . This accep a ion o Theme is necessa y o ake in o accoun all he complexi y o discou se o ganisa ion. 1.1 DISCOURSE ORGANISATION, TEXT SEGMENTATION AND SEQUENTIALITY Discou se o ganisa ion and ex segmen a ion a e seen as he consequence o he ‘linea iza ion p oblem’ (Le el 1981). Al hough he ep esen a ion we ha e in ou mind is no linea (simila o a pic u e, a o m, a scene, e c.), he ex (ei he w i en o o al) mus be linea . Tex is a succession o sen ences. Sen ences mus appea one a e he o he in ime. One canno w i e o speak, ead o unde s and se e al sen ences simul aneously. This lack o isomo phism be ween men al ep esen a ion and wha we mus p oduce is a he oo o discou se o ganisa ion. The s udy o discou se o ganisa ion aims o ind answe s o he ques ion: how speake and w i e go abou p esen ing in o ma ion in a linea o ma . I ake as my s a ing poin he issue o sequen iali y in ex as de ined in Gou sos (1996) who sees ex as a “pe iodic al e na ion o ansi ion and con inua ion spans” (op.ci .:501). Model o Sequen iali y dis inguishes h ee le els o discou se s uc u e. The cogni i e le el sees he w i e 's men al ep esen a ion as s uc u ed by he basic s a egies o con inui y and discon inui y. The linguis ic le el is conce ned wi h he echniques a ailable o ealize hese s a egies. The ex ual le el is he ma e ial esul o hese s a egies and echniques. Tex segmen a ion in o con inua ion and ansi ion spans pe ains o he ex ual le el. Tex segmen a ion can be iewed om he con inui y angle and he discon inui y angle. F om he con inui y angle, linguis ic uni s clus e a ound a speci ic in e p e a ion c i e ion. F om he discon inui y angle, ex is di ided in o segmen s o spans (in Gou sos' e minology). The c i e ia which bind ex uni s oge he may conce n pa s o he subjec ma e o he ex (e.g. e e en ial con inui y, ime e e ence) o he p esen a ion p ocess (e.g. he o ical o documen s uc u e). As long as a c i e ion emains alid, he segmen is open and incoming linguis ic uni s join in o a ‘con inua ion span’. When i is no longe alid, he segmen is closed and he esul ing discon inui y is signalled ia a ‘ ansi ion span’ which indica es a shi in he discou se p ocess. Such a shi may be, o example, a e e en ial b eak, he end o opening o a discou se ame, a he o ical a icula ion o he end o beginning o a documen s uc u e segmen (pa ag aph o sec ion). Con inui y being he de aul , a majo ask in he w i ing p ocess is o signal discon inui y. In he absence o a cue o he con a y, he eade will in e p e incoming sen ences as con inuous. Example 11 cons i u es a good illus a ion o a combina ion o con inua ion and ansi ion spans. In his ex ac , a s ing o cohesion de ices es ablishes con inui y a ound he opic o “deba e be ween specialis s o ansa lan ic ela ions” (I pu hese de ices in bold). All hese de ices cons i u e cues helping he eade unde s and ha he w i e keeps e e ing o he same hing i.e. ha he e is a main con inua ion span cons uc ed a ound a opical con inui y. (1) Since he end o he cold wa , he deba e be ween specialis s o ansa lan ic ela ions has ended o be sa is ied wi h wo hy p onouncemen s and much simpli ica ion. I has no shown su icien conce n o he b ead h o he changes aking place [...]. Mo e ecen ly, he discussion has been ocusing on a supposed gap in social alues 1 The o iginal F ench e sion is gi en in appendix. 2 be ween he wo sho es o he A lan ic, an idea o which he e en s o 9/11 ha e pu an end. This deba e is ongoing, bu i is now limi ed o he domain o social analysis. In o eign policy e ms, his discussion on con inen al shi has u ned in o an opposi ion be ween he unila e alism o Ame ica's policy and he mul ila e alism o i s Eu opean pa ne s. A he same ime, example 1 shows h ee cues o shi opening new ci cums ance ames ( hese cues a e no i alicized). The i s is a empo al ame in oduced ia an ad e bial se ing a ime e e ence (Since he end o he cold wa ). This ime e e ence emains alid all h ough he i s pa ag aph. We can he e o e say ha all he sen ences in his i s pa ag aph clus e a ound a common ime e e ence. The scope o his i s ime ad e bial cons uc s a i s (sub-)con inua ion span. The second pa ag aph begins wi h ano he ime ad e bial (Mo e ecen ly) exp essing a di e en ime e e ence. This ad e bial in oduces a new empo al ame and hen, signal a shi . As o he i s ame, his one also spans he en i e pa ag aph. We can also men ion a hi d ame in oduced wi h In o eign policy e ms. This no ional ame co esponds o he las sen ence and i s inside he empo al ame in oduced wi h Mo e ecen ly wi hou closing he empo al ame. Figu e 1 gi es a ep esen a ion o his complex sequen iali y. I shows ha a con inua ion span may con ain se e al o he (sub-)con inua ion spans associa ed wi h a di e en componen o he p ocess (e.g. ci cums ances). In his example, ansi ion spans a e minimal, consis ing in he space be ween wo sen ences o wo pa ag aphs. To e e o hese di e en componen s o he p ocess, I use he Ges al dis inc ion be ween igu e and g ound. The pa icipan s o he opics o he p ocess plays on igu e whe eas he ci cums ances o he p ocess ha e o do wi h g ound (by se ing he scene in which he igu e appea s). All he con inui y cues (cohesion de ices) and he discon inui y cues (ini ial ad e bials) men ioned occu in Theme posi ion. This posi ioning is no a coincidence. One o my wo k's basic claims is ha elemen s in Theme posi ion play a c ucial unc ion in signalling sequen iali y in discou se i.e. in indica ing whe he he e is discon inui y o con inui y in he discou se low. 3 Figu e 1: Rep esen a ion o sequen iali y in example 1 1.2 THE ROLE OF THEME IN DISCOURSE ORGANISATION Some cogni i e s udies (e.g. Enk is 1989, Gi ón 1995) claim ha wha he w i e exp esses i s co esponds o he ‘c ucial’ in o ma ion. ‘C ucial in o ma ion’ means in o ma ion necessa y o he co ec in e p e a ion o he pu pose o a gi en message. This concep is de i ed om he “C ucial In o ma ion i s ” p inciple (CIF) de ined in Enk is (1989). Enk is opposes his p inciple o he “Old In o ma ion Fi s ” p inciple. Con a y o Enk is , I do no oppose hese wo p agma ic p inciples. I hink ha c ucial in o ma ion can in some cases co espond o old in o ma ion. I depends on he in o ma ion low. Ei he he w i e wan s o indica e ha incoming in o ma ion is in con inua ion wi h p eceding in o ma ion; o he wan s o indica e ha he e is a shi o a b eak. In bo h cases, indica ion is gi en by he i s elemen s o he message. In he case o con inua ion, i s elemen s may co espond o gi en in o ma ion. In he case o shi , i s elemen s exp ess in o ma ion ha o ien s he es o he message by, o example, se ing new ci cums ances ( his concep ion is also suppo ed by Be y 1995, F ies 1995 and Ma hiessen 1995). F om he poin o iew o expe ien ial me a unc ion, elemen s in ini ial posi ion exp ess ci cums ances o en i ies implied in he p ocess desc ibed. F om he poin o iew o ex ual me a unc ion, i s elemen s o ganize discou se by segmen ing i in o chunks i.e. by indica ing i he segmen is s ill opened o i i is closed. Acco ding o my posi ional de ini ion o Theme, I may say ha Theme pa icipa es in he managemen o discou se o ganisa ion, by ul illing a dual unc ion: o ien a ion and connec ion. 1.2.1 Theme as o ien a ion One discou se unc ion o Theme consis s in o wa d-looking o ien a ion: Theme is he beginning o he message and, as a consequence, se s p elimina y c i e ia o in e p e a ion o he es o he message. As s a ed by F ies: “Theme unc ions as an o ien e o he message. I o ien s he lis ene / eade o he message ha is abou o be pe cei ed and p o ides a amewo k o he in e p e a ion o ha message” (F ies 1995:318). In e ms o discou se comp ehension, elemen s ha ha e been ead i s ha e a s onge in luence on he in e p e a ion o he es o he message han la e elemen s (see Thompson 1985, Hasselgå d 1996, Le D aoulec & Pé y-Woodley 2003). I we ocus pa icula ly on ini ial elemen s occu ing be o e he g amma ical subjec (such as ini ial ad e bials), we ind elemen s ha se a discou se ame o he in e p e a ion, as s a ed by Cha e: “[elemen s in ini ial posi ion] limi he domain o applicabili y o he main p edica ion o a ce ain es ic ed domain [...] se [ ing] he spa ial, empo al o indi idual amewo k wi hin which he p edica ion holds” (Cha e 1976:53) This is ypically he case wi h he h ee ini ial ad e bials in he example 1. Cha e's no ion o “ es ic ed domain” is ela ed o he no ion o ‘discou se ame’. Discou se 4 ames ha e eme ged om se e al F ench s udies ini ia ed by Cha olles (Cha olles 1997, Cha olles e al. 2005). These s udies sugges ha ini ial ad e bials ha e an ins uc ional meaning ela i e o segmen a ion by p ojec ing an in e p e a ion c i e ion o wa d and hus de ining he ini ial bounda y o a segmen . In ou iew, wo p ope ies o ad e bials a e called upon in his de ini ion: scope and s uc u ing powe . “Scope” co esponds o he seman ic con inui y o he e e ence exp essed by he ad e bial. “S uc u ing powe ” (c . Le D aoulec & Pé y-Woodley 2003, 2005, Ho-Dac & Pé y-Woodley 2008) co esponds o he capaci y o ini ial ad e bials o di ide in o ma ion in o blocks i.e. segmen s wi hin which sen ences clus e a ound an in e p e a ion c i e ion, o en bu no necessa ily he e e ence exp essed by he ad e bial. G amma ical subjec s may also unc ion as o ien e s o he es o he message. In example 1, he i s subjec se s he main opic o he en i e pa ag aph (and e en o he sec ion) : he discussion be ween ... In his case, Theme may be ela ed o he no ion o abou ness o in Halliday's e ms: “ ha wi h which he clause is conce ned” (Halliday 1985:38). 1.2.2 Theme as connec ion A second discou se unc ion associa ed wi h Theme consis s in backwa d-looking connec ion: Theme connec s he es o he message o he p eceding discou se by exp essing elemen s ha allow he eade o in eg a e incoming in o ma ion in a cohe en way in o he men al model in cons uc ion. As Halliday s a es: “Theme is he peg on which he message is hung” (Halliday, 1970:161) Theme as a peg is seen as a cohesion de ice. The common s a egy ha a w i e can employ o indica e ha incoming in o ma ion is linked o he p eceding in o ma ion is o exp ess gi en in o ma ion in Theme posi ion, and new in o ma ion in Rheme posi ion. In example 1, all g amma ical subjec s (excep he i s ) connec he sen ence o he p eceding one. In o he wo ds, hey c ea e opical con inui y in his con inua ion span. Con e sely, because con inui y is he de aul , he absence o he ini ial ad e bial could be in e p e ed as a cue o ime e e ence con inui y. 2 CONFIGURATIONS OF CUES FOR MARKING DISCOURSE ORGANISATION In p eceding sec ions, I ha e p esen ed my heo e ical poin o iew on he ole o Theme in discou se o ganisa ion. My conce n is mo e speci ically o iden i y he cues ha ha e he capaci y o indica e con inui y o discon inui y. My main in e es is o disco e he kind o cues ha we mus ake in o accoun in o de o desc ibe and unde s and he ex ual le el o discou se o ganisa ion. My claim is ha discou se o ganisa ion is signalled by con igu a ions o cues a he han by single ma ke s. Mo eo e , w i e s and eade s ha e o manage se e al le els o o ganisa ion which include hema ic con inui y bu also ime and space e e ence, he o ical a icula ion and documen s uc u e. As discou se cues may simul aneously con ibu e o se e al o hese in e dependen le els, a global iew o 5 discou se o ganisa ion is needed. A da a-d i en app oach seems o be a ele an way o ha e such a global iew. The con igu a ions o cues ha I aim o disco e can be de ined as ollows: he co-occu ence o elemen A wi h elemen B in a speci ic ex posi ion in a pa icula ex - ype leads he eade o in e p e ei he con inui y o shi be ween he p eceding discou se segmen and incoming in o ma ion. In his de ini ion, I men ion h ee kinds o cues: he lexico-syn ac ic elemen s A and B, ex posi ion and ex - ype. The es o his sec ion desc ibes hese h ee ca ego ies o cues. 2.1 LEXICO-SYNTACTIC ELEMENTS The se o lexico-syn ac ic elemen s aken in o accoun co esponds o all he elemen s ha occu in Theme posi ion. I choose o de ine Theme posi ion as he equi alen o he o e all p e e bal zone like F ies (1995), Ma hiessen (1995) and Be y (1995) do. This choice de i es om he global iew o discou se o ganisa ion adop ed in his s udy. Theme mus be conside ed as simul aneously o ien ing and connec ing in o de o ake in o accoun he complexi y o sequen iali y i.e. he ac ha he e a e di e en le els o o ganisa ion (e.g. igu e, g ound o he o ical s uc u e). In example 1, elemen s occu ing in he p e e bal zone o he i s sen ence o he second pa ag aph (Mo e ecen ly, he discussion) pa icipa e in signalling ha he e is g ound discon inui y (in ime e e ence) and igu e con inui y (in opical e e ence). In example 22, he same phenomenon is obse ed: he sec ion is o ganized a ound a long opical con inui y ma ked by he co- e e en ial g amma ical subjec s and a ound h ee empo al ames in oduced by he ime ad e bials. Each ime ad e bial has a scope and ma ks a discon inui y be ween h ee homogeneous segmen s in e ms o ime e e ence. (2) Flo ence-Milan, 1500 - 1513 [heading] In 1500, Leona do goes o Man o a, whe e he d aws Isabella d'Es e's po ai , [...], o Venice, [...], and o Flo ence, whe e -[...]- he will s ay ill 1506. He sha es his ime be ween pain ing [...], and mili a y enginee ing p ojec s in he A no alley and in Piombino. Leona do esumes wo k on heT a a o s a ed be ween 1487 and 1792, and con inues un il a ound 1513. F om 1506, he di ides his ime be ween Milan whe e [...], and Flo ence whe e [...]. He e u ns o his eques ian s a ue p ojec , [...]. He deploys an in ense scien i ic ac i i y: ana omy, ma hema ics, and p oduces a chi ec u al and deco a ion p ojec s o Cha les d'Amboise. Bu , in 1513, he lea es Milan o good as he ci y is eclaimed by he an i-F ench coali ion. Rome-Amboise, 1513 – 1519 [heading] In Rome, whe e he has his lodgings in he Bel ede e, Leona do inds himsel [...] De ining Theme posi ion as he en i e p e e bal zone means he lexico-syn ac ic elemen s aken in o accoun o he s udy a e hose ep esen ed in igu e 2: 2 The o iginal F ench e sion is gi en in appendix. 6 A i s dis inc ion is made be ween special and canonical cons uc ions. Special cons uc ions (o cons uc ions wi h a ‘special Theme’) co espond o sen ences whe e he g amma ical subjec has no e e en ial meaning. These cons uc ions may be used o ocus on one pa o he sen ence (as in cle cons uc ions o le disloca ions), o in oduce new e e en s (p esen a ional cons uc ions) o o indi ec ly exp ess an e alua ion o he p ocess being exp essed (i -ex aposi ion). In canonical cons uc ions, he g amma ical subjec is he opical Theme. A second dis inc ion ela ed o syn ac ic in eg a ion is made be ween de ached elemen s and in eg a ed elemen s. De ached elemen s a e hose which occu be o e he g amma ical subjec , sepa a e o no wi h a comma (e.g. ad e bials, apposi ions, on ed adjunc s). In eg a ed elemen s co espond o g amma ical subjec s. 2.2 TEXT-TYPE Tex ual a ia ion is a ea u e ha mus be aken in o accoun in he desc ip ion o sequen iali y. I is one o he bases o my me hodology: discou se o ganisa ion (and sequen iali y in discou se) may be di e en in na a i e, desc ip i e, a gumen a i e, e c. ex s. These di e en ypes o discou se o ganisa ion a e ela ed o he no ion o ex - ype. This no ion is de ined he e ollowing Bibe 's iew ha ypes a e de e mined by linguis ic cha ac e is ics in opposi ion o gen es which a e iden i ied wi h espec o ex a-linguis ic pa ame e s such as social unc ion (Bibe 1988). F om his pe spec i e, he e a e s a egies (as in Gou sos 1996) ha a e used mo e in ce ain ypes o ex . These s a egies depend on discou se o ganisa ion i sel a he han on gen e. Fo example, my co pus is composed wi h ex s ha ha e in common mono- e e en ial p ope y ( hey a e cons uc ed a ound a single opic). Con e sely, he e a e plu i- e e en ial ex s s ongly o ganized wi h space and ime e e ences. Sec ion 3.1 desc ibes my co pus and explains his ex - ype cha ac e isa ion. Tex - ype is a cue ha wo ks oge he wi h o he mo e ex ual cues. In o he wo ds, he eade pe cei es (i is al eady in e p e a ion) ha he ex co esponds o a pa icula ype. On he basis o such a ca ego isa ion, he cons uc s assump ions ela i e o he cons uc ion o he ex . Tex - ype has o ien ed his in e p e a ion. 7 Figu e 2: Lexico-Syn ac ic elemen s aken in o accoun 2.3 TEXT POSITION The las ea u e ha o ms he basis o my analysis is ex posi ion. Tex posi ion is linked o he le el o o ganisa ion ha is ma ked by documen s uc u e. O ien a ion and connec ion p ocesses a e likely o a y acco ding o he le el o o ganisa ion: whe he he Theme s a s a new sec ion, a new pa ag aph, o me ely a new sen ence inside a pa ag aph. This hypo hesis is based on he idea ha documen s uc u e s ongly pa icipa es in cons uc ing he meaning o a ex . I conside ha , om he eade 's poin o iew, he beginning o a new sec ion o a new pa ag aph igge s speci ic discou se p ocesses ha o ien he in e p e a ion. These discou se p ocesses a e no o en s udied in linguis ics. In his s udy, I gi e a cen al ole o documen s uc u e. Th ough ‘playing’ wi h he h ee ex posi ions (in h ee ex - ypes), I will disco e con igu a ions o cues ele an o discou se o ganisa ion. The choice o h ee ex posi ions ollows a common (and qui e in ui i e) associa ion be ween discon inui y and sec ion o pa ag aph b eak. The e o e, he h ee ex posi ions aken in o accoun a e: –S1 = sec ion ini ial posi ion; –P1 = pa ag aph ini ial posi ion3; –P2 = pa ag aph in e nal posi ion. 2.4 SOME ASSUMPTIONS ABOUT THESE CUES Acco ding o he mo pho-syn ac ic ca ego ies o g amma ical subjec s, he syn ac ic unc ion o de ached elemen s, and he di e en ex posi ions, po en ial co ela ions can be de i ed be ween hese cues and hei con ibu ion in indica ing con inui y o discon inui y in discou se o ganisa ion. The main assump ions a e p esen ed in able 1. Table 1: Po en ial co ela ions be ween cues and hei con ibu ion in indica ing discou se o ganisa ion Co ela ion wi h discon inui y con inui y a expe ien ial le el igu e G amma ical subjec s wi h a low deg ee o accessibili y (A iel 1990), pa ag aph b eaks (P1), headings and new sec ions (S1) G amma ical subjec s wi h a high deg ee o accessibili y (A iel 1990), pa ag aph in e nal posi ion (P2) g ound ci cums ance ad e bials, apposi ion, on ed adjunc s, 3 P1 is no aken in o accoun when he pa ag aph begins a new sec ion (S1 and P1 a e exclusi e). 8 Co ela ion wi h discon inui y con inui y pa ag aph b eaks (P1), headings and new sec ions (S1) ini ial connec o s a he o ical le el ex ual ad e bials (o ex ual ma ked Theme), ini ial connec o s, pa ag aph b eaks (P1), headings and new sec ions (S1) pa ag aph in e nal posi ion (P2) G amma ical subjec s a e ep esen ed he e in e ms o hei deg ee o accessibili y as in he scale de ised by A iel (1990). Accessibili y models such as hose o A iel (1990), Gundel e al. (1993) o Cen e ing Theo y (Walke e al. 1998) aim o explain cogni i e p ocesses in ol ed in he ac i a ion o discou se e e en s. One o hese p ocesses is conce ned wi h he con inued ac i a ion o a gi en e e en . In A iel's wo k, one way o keep a e e en ac i a ed is o use mo pho-syn ac ic elemen s ha a e co ela ed wi h a high deg ee o accessibili y (e.g. p onouns, demons a i e NPs, e c.). Con e sely, he in oduc ion o a new e e en mus be accomplished ia elemen s co ela ed wi h low accessibili y (inde ini e desc ip ions, new p ope names, e c.). A iel's accessibili y scale enables me o classi y e e en ial exp essions in a way ha may co espond o he no ions o con inui y o discon inui y. The nex sec ion gi es an o e iew o he p oposed co espondences be ween deg ee o accessibili y and g amma ical subjec mo phology. I we look back on example 1, we can see ha all sen ences ha e a opical Theme ( s. a special Theme). In he i s sen ence, he comple e de ini e desc ip ion co ela es wi h a middle-low deg ee o accessibili y (deg ee o accessibili y = 2). This non co- e e en ial exp ession se s he main opic o he sec ion. In subsequen sen ences, se e al cues o opical con inui y occu : an anapho ic p onoun (wi h he highes deg ee o accessibili y = 7), a educed co- e e en ial de ini e desc ip ion (wi h a medium deg ee o accessibili y = 3), a educed co- e e en ial demons a i e desc ip ion (wi h a high deg ee o accessibili y = 6) and inally a comple e demons a i e desc ip ion (wi h a high deg ee o accessibili y = 5). Looking now a se ings (i.e. g ound), he e a e a numbe o discon inui y cues in he o m o ci cums ance ad e bials. These ad e bials indica e o he eade ha he ame has shi ed: om one ime e e ence o ano he , o in o a speci ic domain o knowledge. In example 2 ex ac ed om ex s o ano he ex - ype, anapho ic p onouns a e seen o be mo e equen . This equen use indica es a s ong opical con inui y in his ex . Time ad e bials ha e he same unc ion he e as in example 1. These wo examples a e o ganized, o he igu e, a ound a main con inua ion span and, o he g ound, a ound h ee empo al ames. S angely, he Theme o hese wo documen s uc u e segmen s (he e sec ions) exp esses a ci cums an ial e e ence and he opic o he segmen . In example 1, each pa ag aph begins wi h a ime ad e bial and a e e ing exp ession ela ed o he opic. In example 2, he i s sec ion (no di ided in o pa ag aphs) begins wi h a ime ad e bial and he second wi h a space ad e bial. The 9 Time ad e bials can be seen o occu signi ican ly mo e in he i s sen ence o a sec ion and in he i s sen ence o a pa ag aph while space ad e bials occu signi ican ly mo e in P1 only. Con e sely, he e a e signi ican ly ewe ime ad e bials in in apa ag aphic sen ences. The di e ence be ween space and ime ad e bials may be explained in e ms o local discou se unc ion s. global discou se unc ion. Space ad e bials a e associa ed wi h pa ag aph ini ial posi ion bu no wi h sec ion ini ial posi ion. Mo eo e , he de ia ion as ega ds space ad e bials in P2 is no signi ican . This means ha i is no a e o ind a space ad e bial in pa ag aph in e nal posi ion, in con as wi h ime ad e bials. These esul s may indica e ha space ad e bials ul il a mo e local discou se unc ion han ime ad e bials. The nex esul s will con i m hese obse a ions. Figu e 7 indica es he esul s o he same z-sco e es applied in each sub-co pus in o de o know i hese associa ions wi h ex posi ions a e s able ac oss he h ee ex - ypes. Figu e 7 shows ha all h ee co po a display he de ia ions associa ed wi h ime ad e bials: mo e ime ad e bials in S1 and P1 and ewe in P2. The z-sco e in P1 o GEOPO and PEOPL may be accep ed as signi ican de ia ion, bu no in ATLAS. De ia ions conce ning space ad e bials a e only linked o ATLAS in compa ison wi h ime ad e bials. The discou se unc ion o ci cums ance ad e bials seems o be mo e de ini e in his sub-co pus. Time ad e bials begin sec ions while space ad e bials begin 16 Figu e 6: Time and space ad e bials: de ia ions acc. o ex posi ion Figu e 7: Time and space ad e bials: de ia ions acc. o ex posi ion in each sub-co po a Time Ad Space Ad -10 -5 0 5 10 15 S1 P1 P2 ATLAS GEOPO PEOPL -5 -2,5 0 2,5 5 S1 P1 P2 Time Ad ATLAS GEOPO PEOPL -5 -2,5 0 2,5 5 S1 P1 P2 Space Ad pa ag aphs. This ole dis ibu ion is acili a ed by he ac ha , in ATLAS, sec ions a e e y sho and hie a chically embedded compa ed wi h PEOPL o GEOPO. These i s esul s lead us o conclude ha ad e bials may cons i u e good discon inui y ma ke s because o hei s ong associa ion wi h he s a ing poin o documen s uc u e segmen s. I mus howe e be ca e ul in my in e p e a ion o da a. Fi s ly, his associa ion has o ake in o accoun ex - ypes ( ime ad e bials appea o be less speci ic han space ad e bials). Secondly, his associa ion does no mean ha ad e bials signal discon inui y on hei own o example, when hey appea in non sec ion o pa ag aph ini ial posi ion. To cla i y his las poin , le us obse e he lexico- syn ac ic en i onmen o ad e bials. 4.2.2 S ep 2: a ia ions in g amma ical subjec p eceded by an ad e bial This analysis co esponds o he second s age o my me hodology. I measu ed a ia ions in subjec posi ion which can be associa ed wi h he p esence o an ad e bial in de ached posi ion. All he esul s a e gi en in Ho-Dac & Pé y-Woodley (2008). In his pape , I will only p esen he in e p e a ion o hese esul s. Fi s ly, he discou se unc ion o space/ ime ad e bials seems o be e y sensi i e o ex posi ion. I seems ha space/ ime ad e bials a e good segmen a ion ma ke s when hey occu in S1 o P1. In P2, i seems ha he discou se unc ion o space/ ime ad e bials depends o he ex ual s a egy used. Tex may be o ganized a ound a dominan opical con inui y o a dominan space/ ime s uc u e. Va ia ions measu ed o g amma ical subjec s acco ding o ex - ype and ex posi ion indica e ha PEOPL is o ganized a ound a s ong opical con inui y unlike ATLAS and GEOPO, as was seen in he p e ious sub-sec ion. The powe o his opical con inui y is also ele an in a ia ions obse ed in hos and ollowing sen ences o space/ ime ad e bials and may explain he s ong di e ence which opposes ATLAS and GEOPO o PEOPL. In ATLAS and GEOPO, space/ ime ad e bials may indica e discon inui y bu only in speci ic con igu a ions. Space/ ime ad e bials colloca e signi ican ly mo e wi h ei e a ions ha co ela e wi h medium accessibili y. This kind o subjec may be used o emphasise a opical con inui y when he e is a shi in he se ing (i.e. g ound) o he he o ical s uc u e bu no in hema ic s uc u e (i.e. igu e). Space/ ime ad e bials also colloca e signi ican ly mo e wi h new p ope names ha co ela e wi h lowe accessibili y. This colloca ion may indica e ha he e is simul aneously a g ound and a igu e discon inui y. Bu a ia ions in he ollowing sen ences do no suppo his sugges ion. I he opening o a new ime o space con inua ion span co esponds o he opening o a new hema ic con inua ion span, he subjec in he ollowing sen ence should co ela e wi h high accessibili y. Bu my da a shows ha i is no he case. G amma ical subjec s o sen ences ha ollow a P2 sen ence in oduced wi h a space/ ime ad e bial a e signi ican ly mo e associa ed wi h he bo om o he middle o he accessibili y scale. We can also no ice a signi ican associa ion wi h demons a i e NPs. Demons a i e NPs co ela e wi h high accessibili y, bu hey mean mo e han jus e e en ial accessibili y. The p e e en ial use o demons a i e NPs in compa ison o he use o p onouns is o en associa ed, in F ench, wi h “ eclassi ica ion”. Reclassi ica ion consis s in exp essing a known e e en s ipped o i s ini ial ci cums ances (De Mulde 1997). The e e en 's eclassi ica ion nega es he possibili y 17 o an ex ension o he ad e bial's scope (see Ho-Dac & Pé y-Woodley 2008 o mo e de ails). The signi ican posi i e a ia ions obse ed in he sen ence ollowing a space/ ime ad e bial's hos sen ence may be a sign ha space/ ime ad e bials do no open a new con inua ion span a g ound le el. They me ely loca e he p ocess o he hos -sen ence. Con igu a ions whe e he hos sen ence's subjec is a new p ope name and he ollowing sen ence's subjec is a demons a i e NP may indica e a discon inui y o do wi h he igu e bu no wi h he g ound. The case o PEOPL is e y di e en . The opical con inui y is so s ong in his ex - ype ha ime ad e bials seem o align hei beha iou wi h he o ganisa ion es ablished by opical con inui y. In PEOPL, ime ad e bials co-occu signi ican ly mo e wi h high accessibili y co- e e en ial exp essions such as p onouns, possessi e NPs and epea ed p ope names. These associa ions in PEOPL a e in ag eemen wi h he gene al model: in P2 subjec e e en s p esen a ema kably high deg ee o accessibili y, indica ing opical con inui y. This con inui y is absolu ely no dis u bed by he p esence o a ime ad e bial in ini ial posi ion. The powe o opical con inui y is so s ong ha i is possible o ha e such associa ions in sec ion ini ial o pa ag aph ini ial posi ions. This esul ag ees wi h obse a ions p esen ed in Le D aoulec & Pé y-Woodley (2003) whe e au ho s show ha , in na a i e ex s, ime ad e bials do no open a discou se ame bu a he loca e he ch onological s a ing poin o a succession o e en s. This is exac ly wha we ha e wi h he i s ime ad e bial in example 2 whe e In 1500 does no eally ex end i s seman ic scope un il he second ime ad e bial. The seman ic c i e ion o he i s empo al ame is om 1500 o 1506 ins ead. Ne e heless, we may s a e ha in example 2, ime ad e bials s uc u e he ex by indica ing he bounda ies o he h ee pe iods o Leona do's li e be ween 1500 and 1513. Bu his s uc u ing powe would ce ainly be less s ong wi hou his heading and i he sec ion did no begin wi h a ime ad e bial p edic ing a ime o ganisa ion o he es o he documen s uc u e segmen . 5 CONCLUSION The e a e wo aspec s o my conclusion: he i s conce ns me hodology while he second conce ns ad ances in he s udy o discou se o ganisa ion. Conce ning me hodology, I ha e shown he capaci y o he da a-d i en app oach o p o ide new insigh s. Fi s ly, i has p o ed o be a good ool o manipula ing da a. The z-sco e es is e y simple o manipula e and enables us o es he s uc u ing powe o each ea u e ha may in e ac in he signalling o discou se o ganisa ion. Secondly, i o e s new pe spec i es o he s udy o discou se o ganisa ion. I enables us o iden i y he ex ual cha ac e is ics o global o ganisa ion. Fo example, ATLAS seems o ha e a s ong spa io- empo al o ganisa ion while PEOPL has a mo e opical o ganisa ion. Thi dly, his me hodology makes i possible o es he discou se unc ion o speci ic lexico-syn ac ic elemen s such as ime ad e bials. Finally, his kind o s udy is e y mo i a ing o u he discou se s udies. As we ha e seen, he e a e o he elemen s ha a e associa ed wi h documen s uc u e segmen s. I would be in e es ing 18 o desc ibe hei discou se unc ion in a same way as we ha e jus done o space/ ime ad e bials. Conce ning now he ad ances in he s udy o discou se o ganisa ion, wo poin s a e essen ial. Fi s o all, he hypo hesis conce ning he ma king o discou se o ganisa ion has been pa ially alida ed. I is clea ha we canno speak abou he s uc u ing powe o a lexical ma ke by i sel . I is a he a ma e o con igu a ions o cues whe e lexico-syn ac ic elemen s play a ole. These con igu a ions seem o be qui e complex. The case o ime ad e bials is a good illus a ion o his complexi y. This alida ion shows also ha discou se o ganisa ion is s ongly sensi i e o ex - ype. In my s udy, conside ing ex - ype co esponds o aking in o accoun he shape o a documen and he ex ual s a egies used in he documen . A p omising u u e di ec ion would be o es he use o he con igu a ions o cues disco e ed in his s udy in au oma ic ex - ype p o iling. REFERENCES A iel, Mi a. 1990. Accessing noun ph ase an eceden s. London: Rou ledge. Be y Ma ga e . 1995: Thema ic op ions and success in w i ing. In M. Ghadessy (ed) Thema icDe elopmen in English Tex s 55-84. London: Pin e , pp. 55-84. Bibe , Douglas. 1998. Va ia ion ac oss speech and w i ing. Camb idge: Camb idge Uni e si y P ess. Cha e, Wallace L. 1976. Gi eness Con as i eness de ini eness subjec s opics and poin o iew. In C.N.Li (ed) Subjec and Topic 25-55. New-Yo k: New Yo k: Academic P ess. Cha olles, Michel. 1997. L'encad emen du discou s; uni e s champs domaines e espaces. Cahie de Reche che Linguis ique 6. Nancy: LanDisCo uni e si é Nancy2. Cha olles, Michel, Le D aoulec Anne, Pé y-Woodley Ma ie-Paule & Sa da Lau e. 2005. Tempo al and spa ial dimensions o discou se o ganisa ion. Jou nal o F ench Language S udies 15(2). 115-130. De Mulde , Wal e . 1997. Les démons a i s : des indices de changemen de con ex e. In N.Flaux, D.Van de Velde & W.de Mulde (eds) En e géné al e pa iculie : les Dé e minan s. 137-200. Besançaon: A ois P ess Uni e si é. Enk is , Niels E. 1989. Connexi y, In e p e abili y, Uni e ses o Discou se, and Tex Wo lds. In J. Allén (ed) Possible Wo lds in Humani ies, A s and Sciences. 162-186. Be lin/New-Yo k: Wal e de G uy e . F ies, Pe e H. 1995. Themes, Me hods o De elopmen , and Tex s. In R.Hasan & P.H. F ies (eds) On Subjec and Theme. A Discou se Func ional Pe spec i e. 317-360. Ams e dam: John Benjamins. Gi ón, Talmy. 1995. Func ionalism and G amma . Ams e dam/Philadelphia: John Benjamins. Gou sos, Dyonisis. 1996 A model o sequen ial ela ions in exposi o y es . Tex 16(4). 501-533. Gundel, Jeanne e K., Hedbe g N. & Zacha ski R. 1993 Cogni i e s a us and he o m o e e ing exp essions in discou se. Language 69(2). 274-307. Halliday, Michael A.K. 1985. An in oduc ion o Func ional G amma . London : Edwa d A nold. 19 Halliday, Michael A.K. 1970. Language s uc u e and language unc ion. In R.Hasan & P.F ies (eds) New ho izons in Linguis ics. 140-164. Ha mondswo h: Penguin. Hasselgå d Hilde. 1996. Whe e and When: Posi ional and unc ional con en ions o sequences o ime and space ad e bials in p esen -day English. Oslo: Scandina ian Uni e si y P ess, Ac a Humanio a. Ho-Dac, Lydia-Mai. 2007. Explo a ion en co pus de la posi ion ini iale dans l'o ganisa ion du discou s. Thèse de doc o a en sciences du langage. Uni e si é de Toulouse 2. Ho-Dac, Lydia-Mai & Pé y-Woodley Ma ie-Paule. 2008. Tempo al ad e bials and discou se segmen a ion e isi ed. Linea isa ion and Segmen a ion in Discou se. Mul idisciplina y App oaches o Discou se 2008 (MAD 08), Feb 20-23 2008, Lysebu, Oslo. Le D aoulec, Anne & Pé y-Woodley Ma ie-Paule. 2003. Time a el in ex : Tempo al aming in na a i es and non-na a i es. In L.Lage we , W.Spoo en & L.Degand (eds) De e mina ion o In o ma ion and Teno in Tex s : MAD 2003, 267-275. Ams e dam & Müns e : S ich ing Nee landis iek & Nodus Publika ionen. Le D aoulec, Anne & Pé y-Woodley Ma ie-Paule. 2005. Encad emen empo el e ela ions de discou s. Langue F ançaise 148. 45-60. Le el , Willem J.M. 1981. The speake 's linea iza ion p oblem. Philosophical T ansac ions o he Royal Socie y o London B295. 305-315. Ma hiessen Ch is ian. 1995. THEME as an enabling esou ce in idea ional 'knowledge' cons uc ion. In M.Ghadessy (ed) Thema ic De elopmen in English Tex s. 20-54. London: Pin e Rayson Paul E. 2002. Ma ix : a s a is ical me hod and so wa e ool o linguis ic analysis h ough co pus compa ison. Ph.D. in Compu e Science, Lancas e Uni e si y, UK. Schnedecke , Ca he ine. 2005. Les chaînes de é é ence dans les po ai s jou nalis iques: élémen s de desc ip ion. T a aux de Linguis ique 5. 85-133. Sinclai John McH. 2004. T us he Tex : Language Co pus and Discou se. London: Rou ledge. Thompson Sand a. 1985. G amma and W i en Discou se: Ini ial s. Final Pu pose Clauses in English. Tex 5. 55-84. Walke , Ma ilyn A., Joshi A a ind & P ince Ellen. 1998. Cen e ing in Na u ally Occu ing Discou se:An O e iew. In M.Walke , A.Joshi & E.P ince (eds) Cen e ing Theo y o Discou se 1-28. Ox o d: Calend on P ess. APPENDIX O iginal F ench e sion o example 1. (1) Depuis la in de la gue e oide, le déba en e spécialis es des ela ions ansa lan iques s'es op sou en con en é d'oscille en e les bons sen imen s e la simpli ica ion. Il ne s'es pas su isammen po é su l'ampleu des changemen s de ond endus [...] Plus écemmen , la discussion s'é ai po ée su un éloignemen supposé des aleu s sociales en e les deux i es de l'A lan ique, auquel les é énemen s du 11 sep emb e 2001 on au moins p o isoi emen mis in. Ce déba se pou sui , mais il es main enan limi é à la sphè e de l'analyse sociale. En e mes de poli ique é angè e, ce e discussion 20 su la dé i e des con inen s a p is la o me d'une opposi ion en e l'unila é alisme de la poli ique amé icaine e le mul ila é alisme de leu s pa enai es eu opéens. O iginal F ench e sion o example 2. (2) Flo ence-Milan, 1500 - 1513 [heading] En 1500, Léona d se end à Man oue, où il dessine le po ai d'Isabelle d'Es e, [...], à Venise, [...], e à Flo ence, où - [...]- il a es e jusqu'en 1506. Son ac i i é se pa age en e des a aux de pein u e : [...], e des a aux d'ingénieu mili ai e dans le al d'A no e à Piombino. Léona d eme en chan ie le T a a o commencé en e 1487 e 1492, e y a aille jusque e s 1513. À pa i de 1506, il pa age son emps en e Milan où [...], e Flo ence, où [...]. Il e ien au p oje de s a ue éques e, [...]. Il déploie une g ande ac i i é scien i ique : ana omie, ma héma ique, e ou ni des p oje s d'a chi ec u e, de déco s pou Cha les d'Amboise. Mais, en 1513, il qui e dé ini i emen Milan econquis pa la coali ion an i ançaise. Rome-Amboise, 1513 – 1519 [heading] À Rome, où il loge au Bel édè e, Léona d se ou e [...]. 21