7/28/2019 p.97

1/30

GLoSSoLoGA |6 (2ou-25) 97-128

MLCATYE DScoURsE MARKERS :A coMPARATvE NQURY NTo GREEK AtD ENGLstLEXS ALOERNOS & BRUCE FRASERThis paper proposes an analysis of discourse markers from the point of viev of argumen-tatin thery, on the basis of the categry of discourse modality. Within this frameork itdefines by the notion of justification an area of implicative discourse markers, and consist-ing of causals, inferentials, telics and resultatives. The application of the theory t Englishand Greek data shovs that causal and inferential markers form the two strong poles of thisdmain, vhile there is a marked lack of resultative markers; it also explains the paramet-ric variation of Greek and English. The wide scope f this paper allows a tendency t-wards functional symmetry to become apparent; this affects the procedural semantics ofmarkers and thus determines the features of an interface between the linguistic and thecognitive system, rvhich could be designated as discourse syntax.1. ntroductionWe rill dvell here neither on the nature of discourse markers (DMs) nor n ageneral discription of their field. Fr discussion on these issues, see Fraser (1987,1996, 1999), alokerinos (2004).n this paper we will focus on a subfield f discourse markers, vhich will be de-fined by a notion of communicatie action vhich all members of the categorysubserve: justification From our pint f view mplicative DMs (DMs) articu-late discourse Segments into justificatory communicative acts. n what fllows vevill attempt to put to the fore the procedural means by which DMs steer dis-course segments tovards justification. Cross-linguistic (Greek - English) evidencewill be advanced for DM classification. Bringing to light shared as well as diver-gent features in mechanisms of discourse constructin will help us to assess theshape of the interaction betveen discourse Strategies and language systems.2. On the nature of DM articulationWe take as a defining feature f DMs their moda character. DMs take into theirscope aspects of the indicated ar presented, not merely the said or represented

7/28/2019 p.97

2/30

98 Aexis aokerinos & Bruce Frasermeaning of the discourse segments (DSs) they cnnect and integrate into upperlevel discourse units1. The distinction /e trace is not coextensive vith the stan-dard semantic - pragmatic distinction that is adopted by Sanders (1997, Sanderset al. 1992): our modal level differs frm Sanders' pragmatic level in that the fr-mer is not confined to a focus on illocutionary force but also encompasses propo-sitional attitudes such as epistemicity. We believe that though epistemicity affectsillocutionary frce there is theoretical gain in keeping the tw distinct2, as relatedaspects of discourse modaify. We also want to avoid mapping the Semantic-pragmatic dichotomy onto the propositional content- illocutionary force dis-tinction, since we take as vell established that on the one hand there is an (at leastpartial) pragmatic determination of propositional content, and on the other handthere is an (at least partial) semantic determination of illocutionary force (and,more generally, of the presented aspects of meaning).The need for modality-internal distinctin between propositinal attitude andillcutinary force is put into light by the non-equivalence in paraphrses f thefollowing pair of discourse units, both articulated by a causa DM:1. Buy your supplies nov/, because care for you.2. John oves her, because he came back.1a. advise you to buy your supplies now and the reason want yu to is that care for yu.1b. state that John lves her and the reason believe s is that he came back.a. advise you to buy your supplies now and the reason do so is that carefor you.2b. #l State that John loves her and the reason do so is that he came back.

The first set of paraphrases (1a, 1b), focusing on propsitional attitudes is un-problematic. From the second set though, focusing on illocutionary forces, thenn epistemic sequence (2a) but not the epistemic one (2b) constitutes a possibleparaphrase of (1) and (2) respectively. t appears that the causal DM providesreasons for wanting and advising in (1); whereas in (2), it provides rea.sons for be-lieving but nt (adequate) reasons for stating3.Furthermore, there is linguistic motivation for postulating a metamoda do-1. The distinction beteen "what is said" and "hat is indicated" belongs t Grice. Thehomologous distinction beteen vhat is represented and vhat is presented is evoked byalokerinos (1999), based n Ducrot (1984) and Rdcanati (1981), who traces it back toGardiner (1932).2. For discussion of this pint, see Kalokerinos 2m. Kalokerinos proposes the term inten-tiona attitudes as a cover term for illocutionary forces and propositional attitudes. Thedescriptive gain of this approach will appear in our cross-linguistic discussion later on.3. For extensive discussion of the issue, see Kalorerinos 2ru.

7/28/2019 p.97

3/30

mpicative Discourse Markers: a comparative inquiry into Greek and Engish 99main, where the justifying function of the DM focuses on higher-order speech actaspects of a discourse segment. Here are two examples focusing respectively onthe "saying what" and "saying how" aspects of communicative acts:3. Go! ecause you didn't hear it.4. Go! ecause we have to shout in this place.

There als exist domain-interface cases, which will not be treated here (but seealokerinos 204).DMs so cnstrued may affect the illocutionary meaning of at least one of theDSs they connect. For this to happen, thugh, DMs must link nt simply sen-tences but at least "embryonic" utterances. This requirement has a syntactic sideto it: no complement sentences may enter the focus of DMs as such, because theycannot be cnceived as utterances. The Same iS true for the sentences t vhichcomplements are attached: they cannot be conceived of as utterances independ-ently of their complements. Fr the same resons, complex sentences ith loverlevel attachments are excuded from consideration. At the other extreme, syntac-tically self-contained sentences are in principle suitable to enter under the scopeof DMs as utterances, and are generally the obiect of coherence relations. Themiddle grund is occupied by complex sentences with highest-level dependencies.These are cases of loose subordination ("hypotaxis"), which may be discoursivelymodal.DMs articulate a justifying and a justified discourse segment. The first will benamed Source Segment (SS) and the secondTarget Segment (TS).First Segment (DSl) Discourse Marker Second Segment (DS2)Target Segment (TS) because Source Segment (SS)Surce Segrnent (SS) so Target Segment (S)Table 1DMs operate by modally intereting the DSs they connect and integratingthem into upper level discurse units. This process v/e name integration by inter-pretation t may be anaysed as a three step procedure, as follws:Step 1: DM initiates a process of integration of the tro segments. This requiresan interetation of DSs.Step 2: DSs are being intereted inside the process of integration. ntereta-tion is steered by the aim of integration.Step 3: ntegration is cmpleted in light of the discourse modality f the inter-preted DSs.

7/28/2019 p.97

4/30

100 Aexis alokerinos & Bruce Fraserere are two examples:

5. am your commander. so, go out.6. Are you going to clse the vindow, because it's cold outside.Step 1: So and because initiate an integrative process of justification. DSs willbe interpreted in the light of this process.Step 2: n (5) SS is an argument for TS; therefore it is an assertion. Given theforce and content f SS and the DM, TS will be intereted as an order. n (6)given the impossibility to interet SS at a lower level (as a V-bar attachment,see below), SS wil be interpreted as a quasi-assertion. Given the status of SSand the DM, TS will be interpreted as a request. Therefore SS is cnfirmed as astatement. TS is interpreted in the light f DM and SS as an indirect speech actof request (for 6) and a direct speech act of order (for 5).Step 3: n light of the output meaning and the meaning of DM the lvhole se-quence(s) get their full meaning as acts of justification: a justified direct orderand a justified indirect request, respectively.

3. The class of DMsn our view, most theorists have wrongly taken the notin of cause S a definingfeature fr at least Sme subcategries f DMs.For Sanders et al. "[o]ne of the more organizing [text] types is Causation" (ibid,p. 95). ndeed, the authors opt for the pair of "ba^sic operations" causal vs. additiveas the first of their four "prototypes" in a taxonomy of coherence relations (ibid, p.98-99, als Sanders et al. 1992: 6-7).ln a similar vein, Knott & Dale (1994) consid-er the pair causal vs. non-causal relations to be {mong the main features of theirtaxnmy of DMs ("cue phrases"). They split the category in tvro: cause phrases areintrduced by cue phrases such as because, for, considering that, given that, on thegrounds that, while result phrases are introduced by cue phrases such as therefore,consequenty, as a resut, aS a consequence' so, so that, in order that, this way.The catgory of crsality is subsumed under a more general category of "implica-tion" by van Dijk (1977), in his taxonomy of "natura language connectives", /herethis category is contrasted to the category of "conjunction". e refers to a single cat-egory of "causality (cnsequence)" DMs which includes becalse, So, for, therefore,since, due to, given, thus, as a resut, consequenty, the reason why, hence, whie,whist, and as (but not in order ro, which appears as the nly member of the categoryof "finality'' - ibid, p. 15). n Halliday (1985: 213-214) the "causa _ conditional" ap-pears as the most extended of the four categories of "markers of enhancement claus-es" (the other three being the temporal, the spatial, and the one of manner). Lastly,Lascarides and Asher (1991) pstulate a result category for coherence relations.

7/28/2019 p.97

5/30

7/28/2019 p.97

6/30

l02 Aexis aokerinos & ruce Fraserresn clause" (ibid). Puose usually is indicated by the subjunctive (e.g. in Ngiz-im), r non fully inflected verb forms (e.g. in anuri _ see ibid, p.185-186) 4.Much the same happens in ndoeurpean languages such as Greek and, t alesser extent, French and English, which use the Same mrpheme (pour, ja andfo respectively) ith distinct configurations of mod and tense to express causeand puose (see Kalkerinos 1999, }ru).This extended cross-linguistic parallelism in discourse - grammatical phenome-na Strongly suggests a common conceptual basiss. What is more, if TELC is acmposite categorY, &S re have argued, ne wuld als expect some DMs tspread from the categry of RESUI to TELC. ndeed, this prediction is bornout, as attested by evidence of Greek and English, to be presented in section 4.26.Again, that spreading is subject to similar modal constraints in the verb of DS2 tothose applying to causals7.n summary, the sentential/pre-modal domain of implicatives is dividedamong CAUSALS which are marked fr cause, RESULTATIVES, which aremarked for result, and TELC, vhich are marked for both8. This partition reflectsa conceptual landscape that is linguistically depicted with the concuffence of bthlexical (DM) and grammatical (verbal mod) means.Table 2 does not feature a most basic class of DMs, namely inferentias. Theseare pure DMs; they do not assume sentential (pre-modal) functions. nferentialsintrduce the result of a mental process as such, i.e. they epistemically qualify thepropositinal content of the TS and thus subserve the higher order speech act of4. Cf . Longacre (2996:72): "llianen Manobo teaches us that what have often been calledcause and puose ire sufficiently similar that they may rightly be grouped under causa-tion. t teaches us at the Same time that ... natural languages consistently distinguish ef-ficient from final cause, even in a relatively similar surface structure encoding as in lia-nen Manobo".5. We acknowedge the existence of functional motivation in discourse - grammatica|phenmena, because we believe that these are among of the most prominent interfacesbetween language and cognition. oweve }ve do not need to take a psition as to thenature of sentence - grammatical phenomena, i.e. t take a positin in the controversyof "functionasts" vS. "formalists" (for a recent constructive account, See Newmeyer1998).6. For convergent crosslinguistic evidence, see Palmer (1986: 180-182), vho concludes"there is a god case for believing that puose and result are in a variety f languagesclosely related, and even indistinguishable, cncepts" (ibid, p. 182).7. Though in English the indicative des not exclude purpose readings (see Palmer, ibid, p.180).8. n this paper we are conserned only with "positive" relations betveen the DSs that arebeing connected. We do not take into account "negative" relations that give "con-trastive" interetations (see Sanders at al. 1992:10-11, Sanders et al. |993:101-13,also van Dijk 1977).

7/28/2019 p.97

7/30

mpicative Discourse Markers: a compaative inquiry into Greek and Engish 103concluding. n this way, no "sheer" propositional content may be introduced byan DM. This can be shown with the aid of the fbllowing examples:10.The metal is being heated. tAs a resut / ?Therefore} it is expanding.1 1.The metal is expanding. { *As a resut / Therefore| it is being heated.12.The metal is being heated. {As a resut / *Therefore} have a perception of itexpanding.13.The metal is being heated. { *s a resut/ *Therefore} have a perceptin of itgetting heated.

The distribution of acceptability of therefore Suggests that its fitness to (1)and (11) is a matter f intellectual processing, not of the fact described by TS be-ing the result of the fact described by SS. ndeed, when this is "objectively" thecase, there is some difficulty in acceptig therefore (ex. 10), due to the need toimagine a context in which the content of TS is not perceived but inferred. Asperception can never be a matter of inference, therefore is nt acceptable in (12)and (13)9. These niceties leave unaffected the resultative expression as a resut,which in every case indicates factual relations that either hold (exs. 1, 12) ordon't (exs. 11, 13)10.f now we introduce the modality character into the aforementioned definingfeatures we get a rather different picture:

Table 3This is a minimum table. The question v/e will have to answer is whether telicsand resultatives may enter the picture (i.e. may function as DM-prper). Thisdouble question can be schematized as follows:

9. Judgments and evidentials are cnsidered as frming the tvo main sub-systems ofepistemic moda|ity (Palmer 1986: 53). Many languages grammaticalize both. "]hevisuas are the preferred evidentiall' (ibid, p. 68-69).

10. more accurate wording would e that the utteran ce presents its content as being inthe vorld. Nothing we say in this artice is abut ontology. Some is about h'wontolo8y is spken. Fr the notin of the linguistic (selD presenting as (se presentercomme), see Ducrot 1984. The same argument should be extended to modal readings.Cf. Tsohatzidis (1994:222): "[O]ne must present oneself as having those beliefs,desires and intentins, vhether or not one actuayhas them".

Du introduces a:Type of DMs justifying DS (modal "cause") justified DS (moda "effect")CAUSALS Yes NoNFERENTLS No Yes

7/28/2019 p.97

8/30

106 Aexis aokerinos & Bruce Fraserthe possibility of commenting on non-foregrounded aspects of what has just beensaid. on the contrary, jati has access to that domain. The pa of this domain thatappears inaccessible t epii bears to the quasi-analytic justification of a sayingby rendering explicit a presuppsed part of its content. ati appears to feel com-foable in this role:4.etaniosa pu ton skotosa_ Uati/??epii]l ton skotosa." regret killing him - forkill him did l cos did kil him".25.a to fantasma tu iryu _ Uati / ??epidi| ipar1un fantasmata." saw the Pyrgos ghost _ forghosts do exist"l1.

ere jati but nt epl'l turns out to fullfil the function f justifying a way ofsaying whatever is said in TS by rendering explicit in SS presuppositions car-ried by a factive verb (24), or existential presuppositions of definite descrip-tions (25).on the contrary, epii is almst acceptable, along with jati, in justificatorycommentaries of a previous saying that are presented, in one v/ay or anothe asinformative by themselves:26.o filos mu _ { jati / ?epii} mono ena filo eo _ eryotan kaimerina na me i."My friend _ [cos/ for} 've only got one friend _ came to see me every day".27.o Makis i Killer _ Uati/?epii} prokite ja to iio prosopo _perimene ta imata tu [...]."Makis or the killer - {cos/for} he is one and the same person -stalked his victims [...]".

t thus appears that only jati can access and comment on the non-explicit as-pects that constitute the quasi-analytic background of what is said. nterestingly,this metacommunicative function may be assumed in written English als by foro On a different (oral) registe by cos. Here are some Darwinian examples withfor,that may sound "dated" to contemporary speakers:28.t excited the liveliest admiration that , a perfect strange should knovr theroad (for direction and road are synonymous in this open country) to placeswhere had never been.29.heard one of his mad buffoons (forhe keeps tw, like the barons of old) re-late the fllowing anecdte.30.t night experienced an ttack (for it deserves no less a name) of the"Benchuca", o species of Reduvius, the great bug of the Pampas.11. Translations reveal an interlingual affinity betryeen jati on the one hand and for andcos on the other hand. Here cos and for appear as variants in register. The question ofthis interingual affinity will be discussed in the next section.

1:tI

itai

I

I

7/28/2019 p.97

9/30

mpicative Discourse Markers: a comparative inquiry into Greek and ngish t0731. one which caught at tquique (forthey are found in Chile and Peru) was veryempty.32. [T]he numerous specimens shot either on Chatham or Charles sland (for thetwo sets were mingled together) all belonged to the tw ther speciesl2.

t should be reminded that for as a preposition has similar functions as the MGja in its prepositional uses. Prominent among them is the meaning of puose.Nevertheless, English DM for doesn't spread towards telics, contrary to MG ja (+subjunctive, see below). "ta + subjunctive may also assume epistemic readingsvithout commitment to the factuaity of the SS content and without order restric-tions:33.o Janis (a) ine eo, ja na ine t aftokinito tu apekso."John {is/must be} here, since his car is outside'' l"f his car is there, then Johnis outside".34.la na ine to aftokinito tu apekso _ops anaferun anepiveveotes plirfories- ojerusiastis a ine eo."For his car to be outside, as claimed by unconfirmed reports, the senatormust be there".

The situation for the MG equivalents of because can be summarized as fol-lws: in the premodal domain jati and epii are interchangeable (given restric-tions of position for jati, which can never have SS paced in first position). nthe modal domain though, it appears that jati and epii are complementaries.n "epistemic" readings, where no other order than TS-first is acceptable,epii requires at least some grammatical clues for the epistemic meaning ofDS1. n the rest of the modal cases, the epii-introduced segment (SS) exhibitsa strong preference for first position, vrhereas similar jati-introduced segmentsappear always secnd, in compliance with a general requirement f that DM.This complementarity gives rise to different functional perpectives, and re-flects different strategies f argumentation. DM-SS initial (epi) construc-tions give rise to "entrenched" and formal argumentations that may be felt asmore objective than exposed and informa argumentations expressed by DM-SS final (jat[) constructions. s a matter of fact, the former aim twards orbetter pretend to be objective. Judging from the filling of speakers, this aim isachieved.Finaly, in metamodal uses epii appears rather restricted. A whoe area ofmetacommunicative uses (quasi-analytic commentaries on "saying how") is occu-pied solely by iati. The situation is summarized in the Table 6.12. Examples are taken from the electronic edition, published 1997 by the ProjectG u t e nb e rg Offi ci al Web S te (htt p ://prom o. netlpg/).

7/28/2019 p.97

10/30

NoN-MoD MODAL META-MoDLEPSTEMc ON-EP ST. "Saying that" "Saying ho\"

epii {TS, eSS} ()jati ja+subj {TS, jSS}because {TS, bSS} {bss, TS} [TS, bSS] ()for

* Explicitly epistemic.

l08 Aexis alokerinos & Bruce Fraser

Table 6Beyond complementarity inside the Greek modal domain there is anothercross-categorial and cross-inguistic complementarity to be noticed, one that con-cerns the epistemic domain. As a matter of fact, this is the work of inferential

DMs, to which e vill turn in next chapter. nferentials are basically epistemic.They present the content of the discurse Segment they introduce as the object ofa beief produced within the same discurse. Since they are paratactic, they pres-ent a strict word order. Schematicaly: . Within the same perspec-tive, the epistemic use of becalse, jati and epii in (2), (14) and (16-18) respec-tively may be schematized as < Bp, causDM-q>l3. t is also a noticeabe fact thatpre-position of DM-SS seems impossible in the epistemic use of because, as wellas the use of epii:35.John came back, so he loved her.36. *So he loved he John came back.7'. Because he oved he John came back.2'. #ecause he came back, John loved her.

The symmetry f the two forms may be considered as uncovering the motiva-tion for causals to enter the domain of epistemicity. The field exhibits comple-mentarity in the formal means to linguisticaly retrace mental effect.Apart fr English for and because and G jati and epii, there is a third DMusually described as causal, namely since and afu, respectively, which have also atemporal meaning. Afu and since, as ell as French puisque exhibit a prpensityfor epistemic interetations (ex. 37, below). They seem dd in the classic exam-ple of direct nn-modal (sentential) etilogy (Sweetser 1990: 77, itis 1996:43o13. For detail argumentation for this meaning form, vhich is different from Sweetser's(1990) reading as far as the cntent of TS is cncerned, see Kalkerinos 20.

7/28/2019 p.97

11/30

mpicative Discourse Markers: a comparative inquiry into Greek and Engish 109_ see eX. 38). f the Same sequencing is epistemically modalized (exs. 39, 4o),things get better. Put in dialogue (ex. 41), everything is fine. ndeed, afu, morethan since, shows also a propensity for dialogue:

37G. O Janis tin aapuse, afljirise.37E. John loved he stnce he came back.38G. ?o Janis jirise aftin aapuse.38E. ?John came back since he ved her.39G. o Janis a jirise afutin aapuse.39E. John must have cme back sincehe loved her.40G. o Janis a jirisi afuti aapa.40. John will come back since he loves her.41tG. A: O Janis jirise.B: E, vevea, afutin aapuse.41. A: Jhn came back.B: Of course he did, since he loved her.The prpensity fr dialogic environments and epistemic interetation must betaken into account when defining the prcedural meaning of slnce and afu as dis-course markers. According to einmki (l975)"'both the speaker and the hear-er assume the proposition i a since-clause to be true". Wickboldt (1997: 134) onthe other hand, claims that "[t]he necessary and sufficient cndition for a causalmeaning [f since] is that the content of the clauses [i.e. TS, SS] allows inferring acausal relation, with the since-proposition as the reason". Of course this cannotbe the hole thing, since as Wickboldt notes, since introduces a "Secondary asser-tion''. s a matter of fact, since-clauses pass Rutherford's (1970) tests for non-re-strictive clauses. alokerinos (21) contends that"a function of backgrounding,tgether with a function of free epistemic anchoring, is at the kernel of the proce_

dural meaning of afu, since, and [frenchf puisque",obviusy since and afu stem from a temporal origin, They still are ambiguusbeteen a tempral nn_mdal meaning (V-bar attachment) and a causal ("justifi-catry") mdal meaning (-bar attachment, see Wickbldt 1997, chapter 3, alsoabve, $ection 2). ccordlng to Kalokerinos (ibtd) the latter mning is generatedby the frmer ia a grammat1calization proces$ whlch turns representatins ofpast events into presentations f assumed events, Events that are presented as as_sumed (i.e. as the object f a belief) form a backgrund. The default eplstemlc an_choring for a background is the "mbtual cognltive enlronment" (see Srber &Wilson 1986, chapter 1) f the participants in communicatin. Therefore, it basi_cally lies beyond the sole speaker.This tendency of since and afu should materialize in constraints on mnlogi-cal uses of these markers and lead to a propensity for dialogue. s a matter offact, this is rvhat the folloving examples illustrate:

7/28/2019 p.97

12/30

110 Aexis aokerinos & ruce Fraser42G. Ea, { epii/?afu} to eo.4'. Come, |because/?since| want you to.43G. Ela, {afu/?epii} to elis.438,. Come, lsince/?becausel you want to.44G. : Ela.

B1: a ertho, {afu/??epii} to elis.2: aerth, |epii /??afu} to elo.44'. A: Come.B1: 'll come, {since/??because| you want me to.82: 'll come, {because/??since| want to.Nevertheless, the epistemic anchoring of srnce and afu may remain unspecified.This may happen to afu in DM-SS second position only, contrary to since vhichdoes not require specification of the epistemic anchoring in either positin, andmay even prefer pre-position (see Wickboldt, chapter 4). n the later positin,since may even not exit the modality domain and convey only a propositional

content in which human agency (that is, intentinality) is involved. For the fl-lowing Sentences, tvhich are equivalent in prpsitional meaning, to be uttered,only the Greek one requires that the epistemic anchoring constraint stated abovebe obeyed:45E. Since the diners fell ill,boldt ibid, p. 66)45G. Afuipelates arostisan,

the restaurant's licence was suspended. (Wick-aia tu estiatoriu anakliike.

Contrary to (45E), example (45G) presupposes that the content of SS is not as-sumed exclusively by the speaker (and it might not be assumend by the speaker atall). The Greek utterance may receive a (45)-like reading only if the order of con-stituents is reversed:

45'G. aia tu estiatoriu anakliike, afuipelates arostisanl4.This is an "etilated'' moda use, vhere the epistemic anchoring, by remainingunspecified, is fading out. This use has recently become pervasive in journalisticwriting and speech. The reNon for this must be the particular functional perspec-tive that afu endovs speech: backgrounded reasons are presented as undisputedinformation which is out of focud and rather prvides a frame for focal infrma-tion (see also note 15, below).There is also another brderline case which involves "reported intentionality'',operating a transfer from the modal to the non-mdal domain. Examples (46GlE)

14. Fr fuher discussion of distributinal properties of afu, see Kalokerins 2l.

7/28/2019 p.97

13/30

mpicative Discourse Markers: a comparative inquiry into Greek and Engish 111below illustrate the case. Neveheess, core reasons introducing uses Such as(47) with slhce, are impossible with afu, in either order configuration (althoughagain, (47Gb) may marginally receive a reading of reported intentionality):

46G. re, afuaton plironan.468. Since they would pay him, he came.47'. Since John loves Mary, he gave her flowers (Wickboldt, ibid, p. 55).47Ga. #Afu Janis a.apa ti Meri, tis 1arise luluia.47Gb. #o Janis 1arise luluia sti Meri, afutin aapa.n any case, both since and afu assume a backgrounding function, which pre-vents the segment they introduce from entering the focus of what is communicat-ed. This is an important option fr discourse organization, since both epii andbecause may introduce either focus or topic infrmation, and both jati and for arerestricted to non-topic introductionls. Therefore the use of since and afu is also astaightfonvard way of conveying nn-foca information.The prcedural features of the since and afu render them apprpriate for re-trieving and bringing to the fore parts f the nn explicit meaning of the previousspeaker utterance. These implicit meanings range frm presuppositions t con-versational implicatures and to figures f speech. n doing this in MG, afu appearsto be the dialogica equivalent of jati (see above). The following examples in Eng-lish illustrate the point:

48.A: regret killing him.: Since/afu yu killed him, there is no salvation.49.A: Where des Harry live?B: Somewhere in France.A: Since/afu you don't know where he lives, how are we going t find him?50.A: She has an irn will.: Since/afu she is so persistent, she'll succeed.

15. For is restricted in peripheral commentary functions; it can assume neither topic orfocus positions. See the distribution of acceptability in the following examples:aG. Jirise, {epidi, jati, afu} tin agapuse.aE. e came back, {because, for, since} he oved her.bG. Jirise, akris {epidifati l*afui tin agapuse.bE. He came back, precisely {because/*for/*since} he loved her.cG. {Epidi, *Jati, #?fu } tin agapuse, jirise.cE. (Because/*For/Since } he lved he he came back.dG. krivos {epidi, *jati, *afu} tin agapuse jirise.dE. recisely {because/*for/*sinceI he loved he he came back.

7/28/2019 p.97

14/30

tl2 Aexis alokerinos & Bruce FraserThe same Semantic features render Since and afu prompt fr the speech actsof concession and prvocation. This depends on the respective endorsement ornon-endorsement by the speaker of whatever she attributes (and projects) tothe other side of the communicative exchange. n this W&y, the marker helps inelucidating the modality of speech, within the familiar DM process, vhich wenamed "interpretation by integration". The following examples illustrate thepoint:

51.: Throw it away.: Since/afu yu want me to...5.A: have an award in mathematics.: Since/afu you are so good (as you say), what is the root of 4594?But only Greek afu can react to silent events:

[n a car accident, one driver to the othe at the beginning f the verbal en-counter:l53G. fu i1a to flas anameno, kopane!53E. *Since had the turn signal n, jurk!Moreover afu but not since can be conveyed in dialogue in order to underminethe first speaker's speech, by denying some implicit information upon rvhich herargumentation is based. Smevhat surprisingly, this possibility is realizedby afu,which appears in interjective utterances vrithout TS, exposing a background thatcontradicts the one on which the former speaker's speech appears to have beenbased. How is it possible to deal ith a sheer background (that is a backgroundwithout foreground)? Kalokerinos Q4) argues that, due to the interjective char-acter of the utterance, this contradicting background is nt posited but exposed,

so that its character as a background is being respected. n this ro|e afu may beheaded by ma, which is a dialogic ppsitive marker in Greek. n the circum-stances, English simply uses buf.54G. A: O Lakis epapse na kapnizi meta apo 251ronia.B: [Ma] afu en kapnize potd!54. : Lakis gave up smoking after 25 years.B: But he never did smoke!55G. : a to fantasma.: Afu en iparyun fantasmata!55E. A: saw the ghst.B: But ghost do not exist!on this evidence alokerinos (2001 ,2) concludes that (strictly speaking)afu is not a causal DM. Contrary to since it has not undergne grammatica\iza-

7/28/2019 p.97

15/30

mpicative Discourse Markers: a comparative inquiry into Greek and ngish 113tion so aS to incorporate a basic causal relation to its core-meaning. lthough thismarker is preferentially invested with a causal meaning, this comes from a prag-matic processing that is still external to its meaning. On the other hand, when in-vested with causal meaning, since may enter the non-modal domain of reasons.The following table illustrate a template of functions for stnce, which emergeverall as less modal than thos e of afu16 (Table 7).

Table 7our discussion of other causal DMs vill be cursory and only selective.English given that and Greek eomenu oti are excusively modal:56E. He loves her, given thathe came back.56G. Tin aapa ' eomenu oti jirise.57'. #He came back, given thathe loves her.57G. #Jirise, eomenu oti tin aapa.58E. ??Buy your supplies now' given that care for you.58G. ??eomenu oti se niazome, kane tis promiies su tora.n order to be accepted, examples (57) require a context providing an epis-temic relation between the segments. Such relation is provided by the default con-text of (56). Though both given that and eomenu oti are grammaticalized as

DMs, their rigin of formation is fairly transparent and seems to govern theirfunctin and limit them to the epistemic domain.English as and Greek kaos which are temporal and signal that the event de-noted in TS is situated ithin the temporal interval of the event of SS, seembund to their temporal origin and unable to go beyond the post hoc ergopropter hc, which gave rise to them. They d not enter the modal domain,and therefre are not DMs. Nevertheless, they introduce backgrounded (i.e.non-focal information only) in a parallel way, just as since and afu do. As amatter of fact, there is a tendency for afu, in the "etiolated" modal uses, tocompete with kaos in its ovn domain. aos will probably lose the fight,since as a causal it appears archaic to many MG speakers today. Nevertheless16. Greek locution mia ke, which introduces a "circumstantiadistributin. reason", has a similar

Non-modal- atency Non-modal+ agency ModalEpistemic ModaNon-epistemb MetamodaLinguistic MetamodalSituationaSince NO YES YES YES tYEsl NOfu NO 0imited) YES YES YES YES

7/28/2019 p.97

16/30

ll4 Aexis aokerinos & Bruce Fraserit is still the only to assume a non-modal exclusively backgrounding function infirst position:

59Ga. athos tin agapuse, jirise.59Ea. s he lved he he cme back.59Gb. (?)Jirise, kathos tin agapuse.59Eb. He came back, as he loved her.6G. #Tin agapuse, kathosjirise.60E. #He lved he as he cme back.61G. *Pijene, kathos ime o diikitis su.61E'. *Go, as 'm your commander.The situation for Greek and English causal DMs is summarized in the followingconcentrating table 8.

. Table 8We are now in a positin to cmplete the Greek part of table 6 vith afu andkathos (able 9a).

on-moda- aSency

Non-moda+ atency

Modalepistemic

Modanon-epistemic

Metamodal"saying that"

Metamoda"sayinghow'because YES YES YES YES Es (YES)

epii YES YES YES YES YES NOjati YES YES YES Es YES YESsmce NO YES Es YES YES ESafu NO (limited) Es YES YES YESgiven that NO NO YES (YES) o NOeomenuoti NO NO YES (YES) NO NO

s YES YES NO NO NO NOkaos YES YES NO NO NO NO

7/28/2019 p.97

17/30

NON-MODAL MODAL MTA-MoDALESTEMC NON-E ST. "Saying that" "Saying how"

epii {TS, eSS} ()jati ja + subj {TS, jSS}afu [] {TS, aSS} {TS, aSS} {TS, aSS) ()kaos {TS, kSS}

mpicative Discourse Markers: a comparative inquiry into Greek and Engish 115

Table 9a4.2. Telicss we said abve, telics positively combine the defining features of the categoryof implicatives: + CUSE, + RESUI. SS caTies the motivation for an actionpresented in TS, whose result will be the accomplishment of the propsition e-pressed in SS. one would expect this intricate meaning relatinship between twoDSs to be reflected in a tight syntactic link. As a matter of fact, telic relationsmay be expressed with strong subordination ("embeddment") resulting in "desen-tentialization" (see Lehmann 1988: 193ff.) of the subordinate clause and may notneed a specific marker to introduce them. Such desententialized puose clausessuffer also restrictions in moodl7 linfinitive for English, subjunctive for Greek):

628. He came (in order) /o see.62G. re a) na di18.Since these puose clauses have a reduced Sentential status they could nt

possibty have an independent illocutionary frce. Therefore they cannot pssiblyreclaim a status of utterance. s a consequence, they cannot be members f dis-course marker relations.t is to be expected frm the semantically compsite nature of purpose clausesthat both cause and result particles will enter the field if cmbined with non factu-17. "]he absence of the grammatica marking in non_finite clauses is possible with littleloss of relevant information, because the relevant informatin is mostly indicated inthe main clause" (almer 1986: 156). almer (ibid, p. 162) reports that according tGlon (19s0) "the degree of reduction lof verbs] is related to the degree to ryhich theeven[ described in the subordinate clause is "bound" to the agent or experiencer interms of his lnfluence over lt",18. Greek nais averbal prefix f subjunctive mood (see Veloudis & Philippaki-Warburton

1 983, Phiippaki-Warburto n |92).

7/28/2019 p.97

18/30

116 Aexis aokerinos & Bruce Fraseral moods. of the causa domain prepositions for and ja, in English and Greek mayserve that functionle:

63E. You must win the all stars game, foreeryone to call you a hero.63'. For eeryone to cal you a hero, you must vin the all stars game.63G. Prepi na kerisis ston teliko, jana se lene oli ira.63'G. Jana se lene oli iroa, prepi na kerisis ston teliko.Note that for als has a causal DM function; as for ja, it is the prepositional ba-sis for the basic Greek causal D jati, frmed in compsition with an indicativecomplementizer (for further discussion on these issues, see alokerinos 2f4).The other possibility, i.e. from the side of result markers is exemplified byGreek oste (+ subjunctive) and English so as (+ infinitive).64. He sits in the first row, So aS to be filmed by [television] camers.64G. aete stin proti esi osre na ton pernun kameres.n having recourse to subjunctive2o, Greek puose expression allolvs for aSyntactic connectin between main and subordinate clause less tight than English,which has recourse to infinitival constructions2l. As a matter of fact, puosesclauses introduced by ja may target both the illocutionary or the higher levelspeech act status of the main (S) clause. They therefre ssume both mdal andmetamodal functions.65G. ana mi nomizis oti e se niazome: kane tis prmiies su tora!65'G. Kane tis promiies su tora! anami nomizis oti e se niazome.65E. *n order that you don't think don't care for you: buy your suppliesnw.66G. ana kseris ti inete, o Janis en tin aapa pia.66' G. o Janis en tin aapa pia. ana kseris ti inete.67. ??John desn't love her anymore, so that you be aware f the facts.68G. tana mi to kselaso: tilefonise o Janis.68E. *So that don't forget it: John has called.The above three groups of examples are instances of modal, "hybrid" (inter-face) modal - metamodal, and metamodal DM functions respectively.

19. See also section 3 abve for reported evidence from other languages.20. MG lacks infinitive. For a historical account of the loss of infinitive in Greek, seerrcks (|997).21. Nevertheless, Engish can indirectly express purpose with inflected clausesintroduced so follorved by complementizer (that) and future tensed verb:Do it, so that everyone will call you a hero.

7/28/2019 p.97

19/30

mpicative Discourse Markers: a compaative inquiry into Greek and ,ngish |7t therefore appears that in Greek, contrary to Engish, purpose has a directDM expression. As a matter of fact, the same mohem, jo, combines with amarker of indicative t introduce cause and with a marker of subjunctive to intro-duce puose. Combination vith the indicative marker ofi is grammaticalizedinto a connective (jati) that assumes DM functions. a (+ subjunctive) als ex-

hibits across-the-board DM behavior. nterestingly, the only sot that cannot beoccupied by a telic function22, finds a causal (but nn-factual) vocation. This wehave reviewed in the previous section (see exs. 33, 34, and related discussion).The two-f old jacomplementarity is schematizedin Table 9b.

Table 9b4.3. nferentialsThe case with inferentials (DMs) Seems relatively clear. DMs intrduce TS as aconclusion o more broadly, as a speech act justified by the content of SS. SSprovides the premises or pa of them. n the latter case the rest of the premisesare to be retrieved in the context23.The inputs to natural inferences are bearers of truth values, i.e. propositionslinguistically expressed by assertions. Nevertheless, a conclusion in natural lan-guage may take the form of a non-assertive utterance. Beynd epistemic utter-ances, deontically modalized ones are the possible outcome of a linguistic infer-ential prcess2a:

69. e is an Englishman; he is, therefore' brave. (Grice 1989:25)Z.This is so on conceptual reisons: "the contents of puose as a goal to achieve, and ofbelief as an achieved mental basis for inference, are conceptually incompatible.Therefore, we should not expect ja na, as a marker of purpose, to carry epistemicmeaning'' (Kalkerinos 2004).23. "mplicated premises", see Sperber & Wilsn 1986, ch.2.24.n case of a dentic DS2, the context usually provides as (implicated) premises thepreparatory conditions of the speech act that is being performed in that discoursesegment.

NoN_MoDL MODAL META-MODALEPsTEMC NONEPsTEMc.ja (+na:subjunctive) TELc "cAUSL"t- nAcTwEl TELc TELcjati (+indicative)

..CAUSAL" "cUsAL',[+ IACTwE]

'cUsL" ..CAUSAL"

7/28/2019 p.97

20/30

118 Aexis aokerinos & Brace Fraser70E. The weather is cludy. |So/Therefore/Hence/Thus| it will rain.70G. o keros ine sinefiasmenos. |Ara/pomenos/Sinepos| a vreksi.71'. You have many debts. {So/Therefore/Hence/Thus|, sell your car.71G. E1is pola rei. |Ara/E'pomenos/Sinepos}, pula to aftokinito su.Despite the apparent simplicity f the case, there is the zz|ing problem ofdelimiting the semantic import of DMs in cases where the cntent of an assertiveTS seems nt to be the product of an inferential process but to be independentlystated as a fact. The above possibility of interpretation is exemplified in the fol-lowing utterances2s:72'. Yesterday it was very hot, thereforewe went to the beach.72G. Xes ixe poi Zesti. Piame epomenos stin paralia.73'. We were unable to get funding and therefore had to abandon the project.73G. en bresame na vroume 1rimatotisi ke epomenos anangastikame naengatalipsume to sxeio.74'. He is retiring in March and thus not able to take on the project.74G. Sintaksiootite to Martio ke sinepos en bori na analavi to eryo.We think that the solution to the puzzle is to be found in Blakemore's (1988:192-193) comparison of the folowing utterances (examples renumbered):

75.Tm ate the condemned meat andhe fell ill thirteen and a half hurs later.76.Tom ate the condemned meat and so he fell ill thieen and a half hours later.As Blakemore (ibid) remarks (76) "would... be acceptabe to a hearer wh be-lieved that anyone who ate the cndemned meat would fall ill thieen and a halfhours later. n contrast, the causal interpretation of [75] is not dependent on suchan assumption"26.As a matter of fact, it is significant that examples (72-74) are conjoined sen-tences (explicitly,by and, or tacitly, with comma or "comma intnation"). What ismore, the position of the DM in Greek is revealing: though in "pure'' inferentialreadings, the Greek DMs ccupy an external (eftmst) position, heading the TS, inthe utterances in question it is embedded in the TS, as a parenthetical commentiry.

25. x. (70) is taken from van Dijk (1977: 41). The rest is taken from the Cambridge Dic-tionary of English. Examples have.also been translated int Greek. nterestingly lexicog-raphers' intuitions diverge on the meaning af therefore. According to the CDE: it meansthe same as "as a result, because f that; for that reason". According t Collins CobuildEnglish Dictionary, "[y]ou use therefore to introduce a logical result or conclusion".26.n osvald Ducrot's tens the speaker should be able to convey a topos relating con-sumpin of bad food and a very particular illness (for the notion of topoi, a tol of ar-gumentation theory, see Ducrot 1988).

7/28/2019 p.97

21/30

7/28/2019 p.97

22/30

120 Aexis Kaokerinos & Bruce Fraser79G. A: Vjeno me kapion alo.: {Tote/osfe} ola teliosan metaksi mas.79E'. : 'm seeing someone else now.: {Then/So} it's all over beteen us.Speaker B cannot introduce his utterance with oste if he thereby wants to an-nounce to B that, as a consequence of what he just has heard, "everything isover". Here osfe may only serve to ascertain "everything is over" as a fait ac-compi rhich the speaker nor comes to acknowledge. our description predictsthat tote but not osfe headed TS may have a performative value. The prediction isborn out as the distribution if acceptability in the folloving example attests:80G. A: eteria apofasise na se metaesi.: |Tote/*oste| paretume.8OE. A: The company has decided to remove you.: |Then/So} quit.Oste on the other hand is also a means for bringing presuppsitions of the for-mer speaker's speech to the fore in dialogue:81G. A: o Janis epapse na erni ti jineka tu.: |oste / *Tote| tin eerne!81E. : John stopped beating his wife.: {So/ *Then| he was beating her!n a wrd, with oste (+ indicative) the turn taking (B) speaker presents himselfas infering a content frm what the former speaker said, which he finds (that is,presents himself as having found) in the background (either presuppsitional or

inferential) of the former speaker's speech. n this way osre is a backward lookinginferential, whereas the other Greek dialogica DM, tote (then), is forward look-ing: it presents the result of B's inferential prcessing of A's speech, as informa-tive also for 28.The above discussion may help t outine the situation from a contrastivepoint of vievr. Markers in different languages extend in different v/ays over theSame functins in discourse. From this perspective, the most interesting are themore wide ranging. n what follows we vill overviev the discourse functions ofso, which seem to assume all non properly causal functions (i.e. t introduce re-sult, inference, and purpose), and try to map the Greek implicatives that match

28. Backward looking inferences are factive; hence they convey a commitment f thespeaker to their truth. Frward loking inferences are unmarked as t these features.

7/28/2019 p.97

23/30

mpicative Discurse Markers: a comparative inquiry into Greek and ngish |2lthe same functions. t first sight, there are tv/ wide ranging Greek particles: e/sland ipon.Let us first note that so has a deictic origin and still serves synchronically (ntaS a DM, of course) an ostensive function2g. t is nt alone in this: both Englishthus and Greek efsi have the same active origins. reasonable speculation wuldbe that, in a process of grammaticalization, English so and Greek efsi have ac-quired an anaphoric function, then a discourse marker function. (82), below, ex-emplifies deixis, (83) exemplifies anaphra. (84) contain inferential DMs:

82lG. D it so! / Kanto e/si!83E. think that Celtics will win. ll my friends say so. (Fraser 1999b:399)83G. Pistevo oti a kerisun i Seltiks. Efsi lene oli i fili mu.84E. The water didn't boil. So we can't have tea. (Fraser 1999b: 407)84G. To nero en evrase. Etsi, en borume na ftiaksume tsai.These are not the only functins of so. As noted above so may assume func-tions f forward or backward inferencing in dialgue (see examples 77-80 and dis-cussion, above). Greek ersi cannot do so in either cse. For backward dialogicalinferencing Greek has DM oste. The case of forward dialogical inferencing willbe examined later on.Fr the moment, we might bear in mind that so in compsition (so as, and indi-rectly so that) serves as a telic connective. Here it parallels again osfe introducingsubjunctive (osfe na). obviously the resultative lcution so that is derivative oerthe sentential function so x that, which is a degenerated deictic/anaphoric func-tion. There is an intermediate step between them, namely anaphoric so that, astep from lexical conceptual meaning to lexical procedural meaning:85E. Put the instructions down so that everyone can understand them.85G. Grapse tis oijies etsi oste kaenas na tis katalaveni.s a matter of fact, this is a hybrid case, betveen conceptual and proceduralfunction. nterestingly, Greek here combines deictic efsi with telic/resultativeoste (+ subjunctive). All these instances appertain to the non-modal domain.Nevertheless, bth so and ersi enter the modal dmain, aS example (84) abovesuggests. So, however appears to have a vider modal range than ersi. Let us re-turn to dialogical forward inferential cases (see examples 77-80). ere so ap-pears to alternate vith non-commitl then, Fr the atter function Greek has tote(see above). The Greek equivalent of so in the afrementioned psition is ipon,which strongly prefers second (at least after the head of first XP) or final posi-tion:

29,Fot an overyiev f the mutifarious functions of ,so, see Frer 1999b,

7/28/2019 p.97

24/30

1'22 Aexis akerinos & ruce Fraser77' G. A: pofasisame na metakomisume Sto L.B: Puate to spiti sas on.78'G. : To mats a ine poli diskol.B: Proponisu lon sklira.Efsi cannot possibly find a place in the above replies.Lipon may also alternate with e/si, osre and tote and even combine with them, ,with n change in meaning, in replies like (79G-B), (80B) and (8lG-B). The possi-ble options ire as follows:79a. Tote ipon ola telisan metaksi mas.79b. oste ipon ola teliosan metaksi mas.79Bc. Ola telisan metaksi mas ipon.80Ba. Tote ipon paretume.8Bb. Paretume ipon.81Ba. oste lipontin eerne!81Bb. Tin eerne iponl.84Ba. Etsi ipon, den brume na ftiaksume tsai.84Bb. en borume na ftiaksume tsa iponThere is no difference in meaning between the (8B), or the (81B), r the(84B) options. s fr (79), v/ere both tote and osre are possible, (79Bc) may beintereted either W&y, i.e. as intrducing a forward or a backward inference.As a matter f fact, ipon may also take the place of every inferential DM inmonological discourse. ts versatility in the domain parallels the behaviur of so.Nevertheless, the origin of ipon lies on the other side of the pre-modal _ modal _meta-mdal cline. Where so, aS well as ersi, have a pre-modal origin, the use ofipon spreads to the modal dmain from the meta_modal side.Basically, ipon is a marker f the sequencing of discourse. This upper level in-tegrative function is achieved through a marking operatin of a previous utter-ance element as a theme (topic) f the following utterance3. This is a very gener-al coherence function. n this respect, the examples below are telling:

86.once upon a time there v/s a bad wolf. The wolf ipon v/as very lonely. Oneday ipon he decided...3. Notice als that ipon may combine not only with every inferential DM (osre ipon,etsi ipon, sinepos ipon, etc.) but also vrith causals (epidi ipon, afu ipon) with nchange f meaning, beyond fixation of thematic perspective. t cannot do so nly withjati, for a good reason: jati is confined to non-thematic positins (see KalokerinoszoM).n the light of above remarks, Brevster's (1'992) attempt to draw a strict paral-lelism betrveen so and lipon appears t miss the point.

7/28/2019 p.97

25/30

mpicative Discourse Markers: a comparative inquiry into Greek and 'ngish |2387. saw ostas. He told me ipon that his sister got maTied.n the same vein, ipon may serve as a tpic-change operator and as a weapn fora candidate speaker to take the floor:88.4: Blah, blah, blah,...: Lipon, ...

English Seems not to dispose of such a functor of discurse Sequencing and al-ternation. he overall picture fr Greek presents one discourse-communicationfiller (ipon) and a deictic particle (er t converge, combine and overlap insidethe modal domain. Though each of them has a naTov/er range than so, they to-gether cover an even vider range of functin, from conceptual word t procedur-al communicatin operator.The situation may be paially schematized in the following table 10.

Table 104.4. DMs f resutThe above conceptualization of the uses of so in discourse together with Blake-more'S (19s8) remarks also alluded to abve, leave no Space for a "resultatiYe''function beyond the non-mdal (sentential) domain. t appears that so acquiresits prcedural inferential meaning in its way out of the deictic/anaphorical do-main of the sentence, in the way of becoming a discourse marker. The same lineof thought applies to Greek ersi which, s we saw, covers a part of the functionadomain f so (and as a matter of fact, the relevant one, from sentence to dis-curse). But then what is left in the domain of result?The obvious candidates remaining are that's why, Greek j'afto, as a reslt,Greek os apoteesma' and as a consequence, Greek os sinepia.We should notice the transparent character of these locutions: clearly they car-ry a conceptual meaning, presumably together with their alleged procedural one.English that's why and Greek j'afto are self contained anaphorical expressions

Sen f ence DiscourseCONCEPTUALNON- MODAL PRoC'DURALNON- MODAL MODAL MTA-MODAL

DE'IxS/ANAPHORA TLICANAPHORA RSULT &T'Lc NFERElT Dscot/RsESEQUENCEENGLs{ so so that so a.slthat so soGREEK etsi etsi oste (etsi) oste na etsi,lipon oste,lipon lipon

7/28/2019 p.97

26/30

24 Aexis Kaokerinos & Bruce Fraserwith transparent composition. s a result/os apoteesma and as a consequence/os sinepia are no less transparent. lndeed, they are elliptical expressions. Theystand for and are equivalents to as aresult of..., as aconsequence of... (Greekequialents complemented with NP genitive). Notice that no other candidate tothe DM status has such characteristics: because is not interchangeable in the samepositions with because of ... Moreove it seems that once similar lcutions getgrammaticalized they slip into the inferential domain. This is what happens inGreek rvith os sinepia("as a consequence", an expression of result) vs. sinepos(" consequentt'', an inferential DM)31 :

89. John loves her. { So/As a resut| he came back.90. John came back. So/#As a resut| he loves her.91. 'm your king. {So/*As a resut| g and get ly Grail!92.The cost of maintenance is very high. |As consequence/Consequenty| asignificant part of the budget goes to it.qzc.To kostos sintirisis ine ipsilo. {os sinepia/Sinepos} mealo meros tuproiploismu pai eki.93E. The cost of maintenance is very high. |Consequenty/?As a consequence|we will apply for supplementary funds.93G. To kostos sintirisis ine ipsilo. {Sinepos/?os sinepia| a anazitisumeproseti xrimatotisi.94E. The cst f maintenance is very high. |Consequenty/ ?As a consequence|go and raise founds.94G. To kostos sintirisis ine ipsilo . lSinepos / ?Os sinepial pijene ke vres porus.

t appears that os sinepia, contrary to the adverbial sinepos, cannot go beyondits litteral compositional meaning to assume a function that is beyond merepropositional content relations. s a matter of fact, markers of result seem tostick to the sentential level. So, they do not seem to be able to get into the do-main of discurse modaity, which is the domain of discourse markers.But why this should be so? A possible functional answer to that question is thatmarkers of result tend not to exist because they need not exist. Their wrk vouldbe to lexiealize a coherence relation that is the most basic of all: in terms of San-ders et al. (1992) it is causal, "semantic" (that is, non-modal) has "basic order"and is "positive". This shuld be the default textual relation. s such it doesn'tneed lexical supprt for processing. ero can do, and this should be the most ec-nomical slution. The appearance of any bearer of procedural meaning instead,should be taken aS an indication to g beyond the basic level relation, and so toenrich discourse with modal meaning.31. An additional indicatin of the grammaticalizatin of this adverbial is that the -osadverbializing moheme is no longer productive in Modern Greek.

7/28/2019 p.97

27/30

Impicative Discourse Markers: a comparative inquiry into Greek and ngish |25 corollary of this study is that we shoud not Speak of discourse markers sabout mphological or syntactic formatins. There are no DMs mohemes orexpressions but merely DM functions.

5. Sme conclusie remarksn this study, we have set out a definition f DMs as linguistic expressions, whichServe the integration of meaning units (MU1) t upper level meaning units(MU2). Typically, DMs integrate two MU1 int one MU2. MU1 are illocutions,at least "embryonic" utterances. Prior to the process f integration, MU1 must beendowed vrith a discourse modaity independenty of the other MUl. f this is ntthe case then integration takes place at the sentential level, which precedes thelevel of discourse. During the process of integration in discourse, the modality ofone or of both MU1 may get specified. This part of the integrative process /ehave named "interetation by integration''.In our study of the broad category of implicative DMs, lve did not take s atheoretical basis for meaning unification the notion of cause. This notion des ntpertain to discourse units; it is a non-modal ntion. Moreover, the prductinand interetation of discourse cannot receive a inguistic explanation in terms fcause. The notion pertaining to discourse is the one of justification, which lsmodal, i.e. it characterizes the speaker's ding as a speaker.From that perspective, we have confined markers f result to the pre-modaldomain. We have also looked into the reasons why expressions of puose aremosty confined to the sentential (pre-mdal) level and put to light cross-linguis-tic variation: Greek though nt English has one telic DM, spreading from the theproto-etiological domain lexicalized by preposition ja. s a resut of our study, itappears that "causas" and "inferentials'' constitute the two strong poles of theimplicative DM domain.nside the modal domain though, these two classes appear as discourse - syntacticvariations of the sime relation, which they serve by the different means, namely hy-potaxis and parataxis, resctively. The basic scheme of their function is as follovrs:

{TS, causDM-SS }Both schemes are instantiatins of the justification relation:JUST (SS, TS)Nevertheless, the tro instantiations f the same relatin focus on different

parts of that relation, namely arguments and conclsions, and thus give rise to dif-ferent, and indeed complementary, perspectives fr the unflding f discourse.

7/28/2019 p.97

28/30

|26 Aexis aokerinos & ruce FraserThese interact with ordering possibilities which are related to functional (topic -focus) perspective. more comprehensive account of the semantics f these "conventional impli-cature" particles should take into accunt all these factors. That is to say, it shoulduncover the instructions they carry for discurse modality constitution and inte-gration, argumentation structure, and functional perspective. These aspects ofdiscurse are not "encapsulated'' but influence one another. Nevertheless, such astudy is beyond the puose of the present inquiry.Aexis KaokerinosUniversity of Cretee-mai : aexis@ phLuogr

ruce Fraseroston Universitye-m ai : bfraser@ bu.e du

REFERENCESBlakemore, D. L. 1988: "So as a constraint on relevance". n R. empson (ed.),Menta Representations. The ntefface betveen Language and Reaity. Cam-bridge: Cambridge University Press, 183-195.Brewster, C. A. 1992: "Discourse and relevance: the case of Modern Greekipon''. Studies in Greek Linguistics 12,357-376.Dijk, T. van |977: "Connectives in text grammar and text logic". n T. van Dijk &J. Petofi (eds.), Grammars and Descriptions. Berlin: $/alter de Gruer, t1-63.Ducrt, o. 1984: Le dire et e dit. Paris: Les Editions de Minuit.Ducrot, o. 1988: "Topoi et formes topiques''. letin d'efudes de inguistiquefranisezz, 1-14.Frase B. 1987: "Pragmatic formatives". n J. Vershueren & M. Bertuccelli-Papi

(eds. ), The Pragmatic P erspective. Amsterdam : Benjam ins, 17 9 - L94 .Frase .-1996: "ragmatic markers''. Pragmatics 6' 167-190.Frase . 1999a: "What are discourse markers?''. oma of Pragmatics31,931-952.Frase B. 1999b: "The particle so in English". Rask 9l1o,397-413.Gardine A. . t932: The Theory of Speech and Language. Oxfrd: ClarendonPress.Grice, . 1989: Studies in the Way of Words. Cambridge, Mass.: Harvard Univer-sity Press.Halliday, M. A. K. 1985: An ntroduction to Functiona] Grammar. London:Arnold.Heinmki, o. L975: " ecause and Since" . Linguistica Silesian a 1, 135-142.

7/28/2019 p.97

29/30

mpicative Discourse Markers: a comparative inquiry into Greek and ngish t27Horrocks, G. 1997: Greek. A History of the Language and its Speakers. London:Longman.alokerinos, A. 1999: "Aetiolgies''. n A. Moser (ed.), Proceedings of the 3rd n-ternationa Conference in Greek Linguistics. thens: Elinika Gramata, 141-150.

[n Greek]Kalokerinos, A. 200/ms.: "Domains f discourse and the levels of explicitness".Boston University, Applied Linguistics Program.Kalokerinos, A. 21; " Af]''. n Y. gguraki et al. (eds.), Proceedings of the 4thnternationa Conference in Greek Linguistics. Thessaloniki: University StudioPress, 31,5-322. ltn Greek]alokerinos, . 4: "The justifying connection, mostly in Greek''. ourna ofGreek Linguistics 5' 7 -8o.itis, . 1994: "Some preliminary remarks on the Greek causative conjunctionsy[,, and 6L". Proceedings of the 8th nternationa Symposium onngish and Greek, Thesssaloniki, 306-318.itis, E. |996: "Supplementary observations on the causal connectives inMdern Greek". Studies in Greek Linguistics 17,424-437 [in Greek].itis, E. (fohcoming): "Causal connectives: the evidence from Greek''.nott, A. & Dale, R. 1994: "Using linguistic phenomena to motivate cherencerelations". Discourse Processes 18, 35-62.Lascarides, A. & Ashe N. 1991: "Discourse relations and defeasible knowledge".Proceedings of the 29th Annua Meeting of the Association for ComputationaLinguistics,52-55.Lehmann, C. 1988: "Twards a typlogy of clause linkage''. n J. Haiman & S. A.Thompsn (eds.), Cause Combining in Grammar and Discourse. Amsterdam:Benjamins, 181 -225.Lngacre, R. E. t9962: The Grammar of Discourse. New York: Plenum ress.Newmeyer, F. J. 1998: Language Form and Language Function Cambridge,

Mass.: MT.Palmer, F. R. 1986: Mood and Modaity. Cambridge: Cambridge UniversityPress.Philippaki_Warburton, . 1992: "On mood and cmplementizers in ModernGreek". n . Philippaki_t'arburton & R. ngham (eds.), Working Papers: Oc-casiona Papers in Genera and Appted Linguistics /. Reading: The Universityof Reading,5-4t,Rdcanati, F. 1981 : Les cnnccs performarifs. aris: Les Editins de Minuit.Rutherford, W. E. 1970: "Some obServations concerning subordinate clauses inEnglish". Language 46, 97 -115.Sanders, T. 1997: "Semantic and pragmatic sources f coherence: on the categ-nzation of coherence relations in context". Discourse Processes24,ll9-147.Sanders, T., Spoore, W. & Noordm, L. l992z "ovard a taxonomy f coher_ence relations". Discourse Processes 15, 1-35.

7/28/2019 p.97

30/30

|28 Aexis aokerinos & ruce FraserSanders, T., Spoore, W. & Noordman, L. 1993: "Coherence relations in a cogni-tive theory of discourse representation". Cognitive Linguistics 4,93-133.Sperbe D. & Wilson, D. 1986: Reevance. oxford: Blackvell.Sveetser, E. 199: From tymoogy to Pragmatics. Cambridge: CambridgeUniversity Press.Thmpsn, S. & Longacre, R. 1985: "Adverbial phrases". n T. Shopen (ed.), Lan-guage Typoogy and Syntactic Description. Cambridge: Cambridge Universityress, 171-34.Tshatzidis, S. L. 1,994: "The gap between speech acts and mental states". n S. L.Tsohatzidis (ed.), Foundations of Speech Act Theory. London: Routledge,22O-233. :Veloudis, Y. & Philippaki-Warbuon, . 1983: "Subjunctive in Modern Greek".Studies in Greek Linguistics 4,151: 168 [n Greek].Wickboldt, J. M. |997: The Semantics of "Since''. Doctoral Dissertation, ndianaUniversity.