ne ‘A smmacTic APPROACH 10 DISCOURSE SEMANTICS 2 * Tivia Polanyi and Penkio Scha English Department University of Ansterdan fe é Ansterdan ‘ ‘The Netherlands anorencr A correct structural analysis of a discourse 5 a prerequisite for understanding it. this paper etches the outline of a discourse grammar which cknowledges several different levels of structure. his gramar, the "Dynamic Discourse Hedel", uses fn hugmented ‘transition Network parsing mechanisn ‘0 build a representation of the semantics of a Ascourse in a stepwise fashion, from left to right, ‘nthe basis of the semantic representations of the dividual clauses which constitute the discourse. hhe intermediate states of the parser model the in ermadiate states of the social situation which ge- erates the discourse. ‘The paper attempts to demonstrate that a dis~ curse may indeed be viewed as constructed by means, f sequencing. and recursive nesting of Aiscourse constituents. It gives rather detailed examples \f Aiscourse structures at various levels, and hows how these structures are described in he franevork proposed here, DISCOURSE STRUCTURES AT DIFFERENT LEVELS Tf a discourse understanding system is to be ble to assenble the meaning of a complex discourse ragment (such as a story or an elaborate descrip don) out of the meanings of the utterances consti~ wuting the fragment, it needs a correct structural cnalysis of it. Such an analysis is also necessary ‘0 assign a correct semantic interpretation to lauses as they occur in the discourse; this is ‘een most casily in cases where this interpretation ‘epends on phenorena such as the discourse scope of fenporal and locative adverbials, the movenent of he reference time ina narrative, or the interpre- ation of discourse anaphora. ‘The Dynamic Discourse Model, outlined in this vaper, is a discourse granmar under development ‘hich analyses the structure of a @iscourse in or er to be able to deal adequately with its senantic spects. It should be emphasized at the outset hhat this system is a formal model of discourse yncax and gonantics, but not a computer implenen— ation of such = model. For a systen to be able to understand a dis- course, it must be able toanalyse ic at several, Afferent levels. + Any piace of talk mst be agsigned to one Inter~ action -- i.e., £6 a socially constructed verbal ‘exchange which has, at any noment, «well-defined set of participants. « Virtually every interaction is views by ita particivants a belonging to a particular pre- defined genre =~ be it a doctor-patient interaction, a religious ceremony, or a casval chat. Depending fon the genre, certain participants may have specif— ic roles in the verbal exchange, and there may be a predefined agenda specifying consecutive parts of the Interaction. an interaction which is sccially ""in- terpreted" in such & fachion is called a speech Beent, (Hiymes,1967,1972) « ‘stretch of talk within one Speech Event nay be characterized as dealing with one Topic. 4, within a Topic, we may find one or nore Dis~ course Unite (DU's) -- socially acknowledged y units of talk which have a recognizable “point” Or purpose, while at tho sane tine displaying a specific syntactic/senantic structure. Clear examples are’ stories, procedures, descriptions, and Jokes. : ‘5. When consecutive clauses are combined into one syntactic/senantic unit, we call this unit a Giscourse constituent unit (deu). Examples are: lists, narrative structures, and various binary structures ("a but B", "A because 3", etc.). 6. Aajaconcy ‘Structures nay well be viewsd as a Kind of dou, but they deserve special mention. ‘They are two or three Bart conversational rou- tines involving speaker change. The clearest examples are question-ancver pairs and exchanges. of greetings. 7. The smallest units which we shall deal with at the discourse level are clauses and o: Operators include "connectors" 1ike, “an “because”, ag well as "discourse markers’ wwell", "go", “incidentally” ‘The levels of discourse structure just dist cussed are hierarchically ordered. For instance, any DU mst be part of a Speech Event, while it must be built up out of dcu's. The levels may this be Viewed as an expansion of the familiar linguis tic hierarchy of phonene, norphome, word and clause. This does not mean, however, that every Giscourse is to be analysed in terns of 2 five level tree structure, with levels corresponding to aca, DU, Topic, Speech Bvent and Interaction. To be able to describe discourse as it actual- ly occurs, discourse constituents of varicus types must be allowed to be enbedded in constituents of ‘the same and other types. Ha shall see various ex- anples of this in later sections. Tt § phasizing here alzeady thet "high Lev. fonts” may be enbedded in “low level co! Por instance, a déu may be interrupted by 2 cleuse which initiates another Interaction. Thus, 2 str tural description of the unfolding i include an interaction as entagded inhis way, we can describe "intrusions", "asides to hird parties", and other interrustions of one In~ eraction by another. In the description of discourse semantics, the evel of the deu's (including the adjacency struc~ lures) plays the most central role: at this level he system defines how the senantic representation £ a complex discourse constituent is constructed ut of the semantic representations of its parts. ‘he other lavels of atructure are alo of some ro- fevance, however: ‘The Discourse Unit establishes higher level se mantic coherence. For instance, the senantics of diggerent episodes of one story are integrated at this level. ‘The Topic provides a frame, which determines the interpretation of many lexical items and dascrap- ‘thone ‘The Speech Event provides a seript which describes the conventional development of the discourse, and justifies assumptions about the purposes of dis~ course participants. ‘The Interaction specifies xeferents for indexicals Like "3", "you", "here", "now", 1 tHe DeWMTC DISCOURSE MODE, Dealing with Linguistic structures above the Janse level is an enterprise which differs in an ssential way from the more connon variant of: Lin uustic activity which tries to describe the. inter~ al structure of the verbal sysbols people exchange: sogurse Linguistics doas not study static vernal dieses, but must be involved with the social pro ‘eas! which produces the discourse -- with the ways rn which the digcourse participants manipulate the DLigations end possibilities of the diecourse sit- ation, and with the ways in which their talk is constrained and framed by the structure of this Sscourse situation which they thenselves creatoa. ‘ne structure one may assign to the text of a dis~ ourse is but a reflection of the structure of ‘the rocess which produced it. Because of this, the Dynamic Discourse Model hat we are developing is only indirectly involved rn trying to account for the a posteriori structure £ a'finished discourse: instead, it tries to trace he relevant states of the social space in tems of hhich the discourse is constructed. This capability s obviously of crucial importance if the model is 9 be applied in the construction of computer sy=~ ens which can enter into actual dialogs. ‘The Dynamic Discourse Model, therefore, mst construct the senantic interpretation of a a: ‘ourge on a clause by clause basis, from left to ight, yielding intermediate semantic representa~ ions of unfinished constituents, as well as set~ ing the semantic parameters whose values influence he interpretation of subsequent constituent: A syntactic/senantic system of this sort may mulated as an Augronted Transition letnork granmar (Hoods, 1970), a non-deterministic larsing systom specified by a set of transition etworks which may call each other recursively. wery Speech Event type, DU type and deu type is ssaciated with a transition network specifying its ntornal structure. As a transition network pro~ sass aes Polanyi & Scha cesses the consecutive constituents of a discourse « segnent, it builds up, step by step, a representa- tion of the meaning of the segment. ‘This represen= tation is stored in 2 register associated wien tne network. At any stage of the process, this resister contains a representation of the meaning of the di: course segment so Zar. Ap AON parsex of this sort nodels inportant aspects of the discourse process. After each clause, the systen is in a well-defined state, characterized by the stack of active transition networks and, for each of then, the values in its registers and the place where it vas interrupted. when ve say that discourse participants know "wheze they are” in 2 complicated discouree, ve mean that they know which, discourse constituent is being initiated or contin— ued, ag well as which discourse constituents have been interruptes where and in what order ~~ inother words, they aze aware of the enbedding structure and other information captured by the ATN configuration ‘The meaning of most clause utterances cannot be determined on the basis of the claugo alone, Dut involves register values of the enbedding acu —- as when’ a question sets up a frane in terns of whichits ansver is interpreted (cf. Scha,-1983) or when, co determine the temporal reference of a clause in a narrative, one needs a "reference time" which is established by the foregoing part of the narrative (section TIT B 2)- From such examples, ve see that ‘the discource constituent unit serves as 2 framework for the semantic imterpretation of the clauses which constitute the text. By the sane token, we see that ‘the genantics of an utterance is not exhaustively described by indfeating ies iliccutionary force and its propositional content. an utterance may also cause an update im one or more senentic resisters. of the du, and thereby influence the senantic. in= terpretation of the following utterances. ‘This phenomenon also gives us a useful per- spective on the notion of interruption which was mentioned before. For instance, we can nov Gigference between the case of # story rupted by a discussion, and the super? lar case of a story. followed by a alécussio: is, in its turn, Followed by another story. first case, the same dey is resumed and 211 register values are still available; in the second case, the firct story has been finishe4 Sefore the Aiscuesion and the re-entry into a storvsorlé via a different story. The first story has closed off and, its register values are no loncer avilable for re-activation; the teller of the sec- ond story must re~initialize the variables of ©: place and charactar, even if the events ©) fond story concern exactly the same chara- situations as the first. ‘this, the notions of interruption ent ress tion have fot only a social reality whi: rienced by the interactants ifvolved. 1) hhave senantic consequences for the building ani interpretation of texts. Interruption and resumption ari y signalled by the occurrence of markers". Interruption is signalled by a USi- marker auch as "imeidentally", "py ¢! 2 know" of “Like”. Resumption is signalled =such’ as %0.K.", *well",\"so" or “anyeay". yer Lists of discourse marking devices, and, ‘nore discussion of their functioning, see (2981) and Polanyi and Scha (1983p) .) terms of our AM description 6f discourse 8) the PUSH and PoP-marker's do alnost ex~ at their nanes suggest. A PUSii-narker sig- creation of a new enbedded discourse con— + while a PoP-marker signals a return to an J constituent (though not necessarily the . Ply embedding one), closing off the cur- stituent and all.the intermediate ones. the tone Pop-marker may thus create a whole Df discourse-POPS was one of Reichnen's gunents for rejecting the ATM model of dis- fructure. We have indicated before, however, mmmodating this phenomenon is at worst a F minor technical extensions of the ATM for~ Polanyi and Scha, 1983); in the present 2 shall from now on, ignore it. DISCOURSE CONSTITUENT UNITS Introduction. This section reviews sone important ways in auses (being our elementary aiscourse con~ units) can be combined to form complex svconstituent units (uhich, in most cases, anther’ conbindtt t6° forn larger aou's by! appligation ‘of the same mechanisns} . For reve are thas focussing on the basic dis- mtactic patterns which make it possible to E complex‘ discourses, and on the semantic tation of these patterns. Sections IV and V } discuss the higher level structures, where ractional perspective on discourse cones the fore. % be able to fous on discourse level phe- ve will assune that the material to be dealt the discourse gramar Ls a sequence con~ of clauses and operators (connectors and > markers). Tt is assuned that every clause the value it has for features such as speak- se topic, propositional content (represented! mula of 4 suitable logic), preposed coneti-” sith thematic yole and semantics), tense, Jality. (the syntactic features we mist, here have semantic consequences which can ye be dealt with within the meaning of the ince they may involve discourse issues.) the semantics of a deu is biilt up in par~ th its syntactic analysis, -by the:sene re nechanism. When clauses of deu's are con- form a larger dcu, their meanings are com- form the meaning of this dcu. Along with 5 for storing syntactic features and seman— neters, each acu has a register which io guild up the logical representation of ite Since the syntactic and semantic rules op- parallel, the syntactic rules have the ity of referring: to the senantics of the ents they work on. This possibility ie in Ain certain casas. We shall see an example on TIT 1 complex discourse constituent units can be into four structurally different tyses: e Polany3. @ Scha i senna asses eit tae See Ynated. unit waich "expands" on it? } A eT eae EE “A because B", “If A then 3”. | ghee moma Bre Se sfsseeed cect, toeirinn rate at a one ae cael tame ues cn srzanaton In ne deca oe onnrel ons 9, sea. La ts mg ere elronay cospeces of Se an eens {propositions expressing the meanings of their con Efopetsione waitesg Ue soenige of aa os consider further now, ¢artainly.comgaspend to id- Senet, Srshes Da SRT aie | jample,)) } Since we will not discuss adjacency struc- | dos geet TL og cians Stacy See i taser oy oath th ce pega ot left for another occasion. " B. Sequential structure Wo shall aiscuss three kinds of sequential te structures: lists, narratives, and topie chaining, e 1. Lists. Perhaps the paradigmatic sequential oa structure is the list: a series of clauses C,, which have a semantic structure of the fm. Play) = Vy) sesy FQ) & Vyr i.e, thd claubes expresspropofitions which con— vay the values which one function has for a series of alternative argunents. For instance, when asked to describe the interior of = room, someone may give an answer structured Like this: “when I come into the door, then I see, ‘co the left of me on,the wall, a large win- dow (.--) Bh, the wall across fron me, there is a eh backet chair (.--) On the right wall is a mm chair (...) In the middle of the roon there is, from lefe to right, an oblong table, next to that = round table, and next to that a tall cabinet. Now I think I'got everything.” (transcript by Phrich and Koster (1983), transiated from Dutch; the constituents we left out, indicated by parenthesized dots, are subordinated consti. tuents appended to the NP they follow.) ‘The List here occurs enbedded under the phrase "T see", and is closed off by the phrase "Nov I think I got everything". 3 Often, the successive argunents in a + List are mentioned ina non-random order ~~ in the above case, for instance, we first get the 1 tions successively encountered in a "glance tour: from left to right along the walls; then the rast.avec onc aU RA NR RONEN RRR ‘The ATW description of Lists is very simple*: clause: next” clause: foxe. . a ema rege kee) Both the iret and the next arc parse clauses which must havg the semantic structure F(a) = v. (Whether f clause can be analysed in this fashion, depends fon surface properties such ag stress pattern and Preposing of constituents.) lWarious registers, ara’ bet by the first clause and checked whon nos clauses are parsed, inorésr ts Gn features such ac tense, soo’; modality. The se~ mantics of a new clause being parsed is simply conjoined with the senantics of the list so far. 2. Narratives: Narratives nay be seen as a special case of Lists ~~ successive event clauses specify what happens at successive timepoints in the world de- scribed by the narrative. Narratives are subdivided into different genres, marked by different tense and/or peraon orientation of their main line Clauses: specific past tine narratives (marked by clauses in the simple past, though clauses in the “néetorical present" may also occur), generic past tine narratives ( marked by the use Of "would" and “used to"), procedural narratives (present tense) , simultaneous reporting (present tense), plans (use of "wilt" and "shall"; present tense also occurs). We shall from now on focus on specific past -narra— tives. The properties of other narratives turn out to be largely analogous. (Cf. Longacre (1979) who suggest treating the internal structure of a dis~ course constituent and ite “genre specification" as ‘two independent dimensions.) clause: clause: ‘elachback specific pact narrative [ALL clause-processing arcs in this network for "specific past narratives” require that the tense of the clause be present or simple past. The evant arc and the event, are process clauses with a on-durative aspact. Tht circunstance arc processes clauses with a durative aspect. (The aepectual ca~ tegory of @ clause is daterminad by the samantic categories of its constituents. Cf. Verkuyl, 1972.) ‘he avent, are is distinguished because it initial~ izes the Pogister settings. All diagrams in this paper have one in tial state (the leftmost one) and one final state (the ost one). The nano of the diagran indi- cates the category Of the constituent it parses. Ares have labels of the forn "a:b" (or sonetines fe ""), where A indicates the category of the constituent which must be parsed to traverse the arc, and Bis A labol identifying additional con~ @itions and/or actions. ‘Polanyi & Scha he specific past narrative network has a ‘fine register containing a formula representing the current reference tine in the progression of the narrative. Uhen the time register has a value €, an incoming cixeunstance clause is evaluated at €, and it does not change the value of the tine re~ gister. An event clause, however, is evaluated vith Fespect to a later but adjacent interval t', and resets the tine register to an interval t", later than but adjacent to t'. (Cf. Polanyi and scha, 1982) To show that this gives us the desired semantic consequences, we consider an abbreviated version of a detective story fragment, quoted by Hinrichs (1981) (el) He went to the window (2) and pulled aside the soft drapes. (C1) Zt was a casement window (C2) and both panels were cranked dovn to let in the night aix. (g2) "You should keep this window locked," he said. (G4) “rt'e dangerous this way. The @ clauses are events, the C clauses are circun- stances. The events are evaluated at disjoint, suc~ Sessively later intervals. ‘The circunstances are evaluated at the same interval, between £2 and 23. ‘To appreciate that the simultaneity of subsequent circumstance clauses in fact is a con sequence of aspectual, class rather than a matter of worl knowledge", one may conpare the sequence lie went to the window and pulled aside the soft drapes" to the corresponding sequence of circun= stance clauses: “He vas going to the window and Was pulling aside the soft drapes". Horlé knowledge ose cone in, however, when one has to decide how much the validity of a circumstance clause extends beyond the interval in the narrative sequence where it is explicitly asserted. specific past narratives may also co: tain other constituents than clauses. An inportant case in point is the “flashback” -- an exbedded nar~ rative which relates events taking place in a peri~ od before the reference tine of the main narrative 1B flashback is introduced by a clause in the plu- perfect; the clanses vhich continue it may be ii the pluperfect or the sinple past. clause: f-event clause: fninit Spon, CyB sittintance Alachback ° The first clause in a flashback ig an event clause; it initializes tings. The reference time within a flashback according to the same meachanism sketched above fo: the main narrative line. After the completion of 2 fias! ain narrative Line continugs »! fe.) 3€ peoceeds from the reforence &: pain narrative. A simple example Peter and Mary left the party ins Mary had ran into John and she had insulted hi so they got into the car and drove doun Avenue C.opie chaining Doother sequential structure 4s the ic chatning steuctuve, where’m series of Stee tt predications about the’ tere ergenene’ act ted. A topic shain conslete of cerlos sf Se0nCy,crvy Gyr with a senaneie stretics of form ¥, (ai .-8, Dy (a), whore a" eranclates the te ersten Ske Saabs’ nthe Elem cima of hain, the topic te expressed. Bhs’ sheaee ther’ a G11 'No of a pronoun) whey eoesee ak ject position of at & proposed const ituens, tn thar clauses, At is tavelly a pronoun eeeen mibject poeition: an example, Wilbur's book I really Liked. © was on relativity theory nd talks mostly about quarks. got it while I was working on the initial part \fomy research. (Based on Sidner (1983), example 26.) ‘The topic chain may he defined by a very de transition network. clause: ton clause! clauses oe teh GH ten topic chain Retwork has @ topic register, which is ser by First clause (parsed by the tol arc), Ghich'Hc. S88 various other 'registefs’ she’ ten arc teata’T snt-dn the'usual way. ag for the topic regis We require that the clause being parsed ‘constituent: which ig interpreted as co= f rdWeial with the value of this register. ‘The weles of @ topic chain is created by simple metion of the semantics of subsequent constit= 3, as in the case of the list. Lists, narratives and topic chains aif- is to their internal structure, but are distr: mally indistinguishable —- they may occur in Heal slots within larger @iscourse constito: For an elegant formation of the grammar, it lerefore advantageous to bring them under a 'n denominator: we define the notion sequence the union of list, narrative and topic chain. Under the heading "expansions" wo describe onstructions in which a clause is followed by € which expands on it, either by elaborating ontent (elaborations") or py describing Prepé 3 of a referent introduced by the clause ic-dominant. chaining"). 1, Elaborats A clause, may be followed by a deu (a or clause sequonce) which expands on ite nt, i.e. redescribes it in more detail. For nee, an event clause may be expanded by a narrative which recounts the details of the Binod at fadane Gilbert's. rst there was an hors d'ecuvre. on the fish bee that the butler Brought a glazed © repast ended with a flaming dessert.. Me é f Polanyi and scha 5 i ‘The Gtscourse syntax perspective suggests that in a case like this, the whole little narrative mist “be viewed ag subordinated to the clause which pro= codes ic. Wf trotorereonstrct ongideu hich cage pists of the first Clause-plue theskoriowing sev favence!? Ay A2iastration of the semantic necessi2 ty of such structural analyses is provided by the Rovenont of the reference tine in-nareatives’ sho above example (by i. Kane) appeared in the context Of the discussion about that phenomenon, (cf bos tyr 1982) “Along with other, oiniler ones, St wee brought up as complicating the idea that very event clause in a narrative moves tho reference tine toe ster interval, Ne would like to suggest that ie de no coincidence that such "problenatic” chase invalee clause sequences belonging to known parasranh oses, and standing in an elaboration relation to the yest ceding clause. the reason why they imeersope She flov of narrative tina de simple enough thee fLiauses arent direct constituenes of the aarativg ‘he ett, sbut consti tutaisheir om eabeaaegedce 0 describe elaborations, we redefine the notion of a clause to be either an elementary one Or an elaborated one (where the elaboration can ‘be constituted by a sequence or by a single clause). o-Scelausg eens sume So . So evclause clause If a clause C is followed by a dou D, D may be parsed as an elaboration of C, if Cand D may be plausibly viewed as describing the same situstion. (Note that this is s relation not between the gurface forns of C and D, but between their nean~ ings C' and D'.) when constructing the senaneice for the conplex clause, this senantic coherence must als 80 be made explicit. 2. Nopie-pominant Chaining. Another phenomenon which gives ris. similar structure is “topic-doninant chaining”. Within a clause with a given topic, certain otter constituents may be identifies as possibly donenant’. A dominant constituent may becone the topic af the next clause or sequence of clauses. He suggest that such a continuation with a new topic be seen as ox Panding on the clause before the topic-swisch, ent 5 syntactically euharainated to this clause. sis Subordinated constituent may either be a single clause or another topic chain sequence. toa Sinlarly, a clause may be foll: clause, tho relative pronoun istituont of the ding clazz. in this case, the relative clause may be the £1 Clause of an embedded topic chain. e-clause, topic chain 9 SoPhe She clause * The notion of dominance Links discourse pi with oxtraction phenomena within the sentence 2-9.) Eeteschik-Shir and Lappin (1379) 5 bWe thus introduce an alternative! network for claus: nto the gramar, in addition to the one given be- ore.) ‘The dominant constituents of the e~clause re stored ina register; the topic of the topic hain, as vell as the relative pronoun of the rel. Laue must be interpreted as coreferential with one if those constituents. the topic. of topic tail a “headless” ‘topic chain) mst in its turn corefer ith the relative pronoun. ‘The senantics consists of simple conjunction. Both variants of topic-dominant chaining Lower by the ‘above network are exemplified in hhe following text (Sidner, 1983), example 026): 1) Wilbur is a fine scientist and a choushtéul my. 2) fe gave ne a book a while back 2)" Sntch 3 really Liked, 3), Yat was on selativicy “esoy 2") Hoatanke“wetly about quarks, sy | ‘ghey are hard to imagine 3 Because they indicate the need for elenentary field theories of a con; plex nature: n ‘hese theories are absolutely es- sential to all relativity research. 8) anyway 8) r got tt 8") while I was working on the initial part of my research. 9). He's a really helpful colleague to have thought lof giving it to me. inddntation indicates subordination with respect to he most recent less indented clause.) This enbed- ing of constituents by means of topic~dominant) haining vould explain the "focus~stack" which sdnex (1983) postulates to describe the pronominal ‘eference phenomena in examples like this. IV DISCOURSE UNITS We now leave the discussion of the basic syn- actic/semantic mechanisms for building discourse ut of clauses, and turn to the higher levels of 3, where considerations involving the goals £ the interaction start to come in. First of all, © shall discuss the entities which Wala (1978) alls Discourse Units#, corresponding closely to he entities which Longacre (1983) sinply calls Discourses". Discourse Units (DU's) are socially cknowlegged units of talk, vhich have a recogniza- Le point or purrose, and which are built around ne of the sequential deu's discussed in section rE. Discourse Unit types which have been inves~ Hgates include stories (Labov, 1972; Wald, 1978; olanyi, 1978b), descriptions of various sorts (Linde, 1979; Ehrich and Koster, 1983), procedural ‘scourse and hortatory discourse (see various re~ ferences in Longacre (1983). Wald restricts his notion te monologic discourse ragnents, It seems reasonable to genoralize it to ‘ages where more than one speaker may be involved. Polanyi and Scha Because of the pragnatic relation between the Dis- course Unit and the surrounding talk (specifical- ly, the need to appear “locally opcasioned" (Jef— ferson, 1979) and £0 make a "point" (Polanyi, 1978b), the central part of the Discourse Unit ugually is not a piece of talk standing completely on its ovn feet, Sut is supported by one or more stages of preparatory and introductory talk on one fend, and by an explicit closure and/or conclusion at the other. This may be illustrated by taking a closer look at comversationally embedded stories the paradigmatic, and most widely studied, DU type. specific past @AWiance setting, narrative, dowrexit oH BES, Seeting, jamzative, story A typical story is initiated with éntrance talk which sets the topic and establishes the rela- tion with the preceding talk. Often we find an ab- stract, and sone Kind of negotiation about the ac~ tual telling of the story. Then follows the "setting" which gives the necessary background material for the story*. Than, fehlows-the “core” cisié past narratives x67 sting @ sequence ‘of everite. The story is concluded with "exit talk” which may formulate the point of the story quite-explicitly, connecting the story world with more general discourse topics. For instwnce, one story in tabov's (1972) collectson nas ag SE enstance tak an explicie f Welicitation and ifs response 0 it: 1 What was Ee most important fight thet jou fenenber, one that sticks in your mind.- A: Well, one 1 think) was with agiri. ‘here is an extemsive section describing the set— fing! "Like Twas a KL@ you know. And she was the baddest gixi, the baddest girl in the holgh= borhood. EF you didn't being her condy to School, sine would punch you in the noweh: tnd you hadi to Kise herunen she'a tell you This gitl was only twelve yours 016, man, tut. she wae a killer. She dsén't cake no junk; she whipped all her brothers. thon, the event chain starts, and finally “jnd i cams to school one aay and I didn't have any money. {..+-) An@ ¥ Hit the girl: owen aid 7 put something on it. Twin fhe fight. ‘the story is explicitly closed off: hat ves ome of the mest importon Not every specific past narvative may be the core of A story. Because of the interactional status of the story (its reauirenent to be "point= ful") theve are other properties which are notice- able inthe Linguistic surface structure -- notably the occurrence of “evaluation” (Polanyi, 1978b) and of 4 "peak" in time narrative line (Zoneacre,1993)._//” + That the necessary background material must be given before the actual event sequence, is attested by a slightly complicated storytelling strate described in Polanyi (1979a) az the "Truc Start repair: the storyteller first plunges right into the event sequence, then. breaks off the narrat: Line and xestarta the telling of the story, nov. with the insertion of the proper background data.‘the structural description of stories, above, should probably be further elaborated ‘sunt. for the phenomenon of episodes: Astoryy Wibuile by consecutive pieces of talk which 24 7 ‘cute separate’nacrative acu's. At’ the level Metory DU, the meanings"of thege narratives je integrated to form a description of one world rather than many. tn English and other Western Buropean lan~ 5, the Discourse Unit seems to be a largely ictional notion. Its constituents are pieces Uk defined by the independently motivated Bey TR.