A framework for the description of Latin discourse markers

A framework for the description of Latin discourse markers

Jmtrml~ ELSEVIER Journal of Pragmatics 30 (1998) 205-223 A framework for the description of Latin discourse markers Caroline Kroon* Department of Cl...

1MB Sizes 19 Downloads 66 Views

Jmtrml~ ELSEVIER

Journal of Pragmatics 30 (1998) 205-223

A framework for the description of Latin discourse markers Caroline Kroon* Department of Classics, Vrije Universiteit Amsterdam, De Boelelaan 1105, NL-1081 HV Amsterdam, The Netherlands

Abstract The article sketches the outlines of a theoretical framework for the analysis and description of discourse markers, based on an evaluation of earlier attempts in this direction, and on the results of an empirical study of a number of Latin connective particles. Central to this framework is the recent insight that connective particles somehow play a role in signalling or maintaining the discourse coherence, and, even more important, that discourse coherence obtains on various 'levels'. In the model proposed, three levels of discourse are distinguished, which are called the representational, the presentational and the interactional level of discourse. It is argued that the functions of discourse markers should be defined in terms of these levels of discourse, and that, in the analyses of individual markers, one should differentiate between discourse function, basic meaning, actual use and side-effects. Application of the framework to a number of so-called coordinating conjunctions in Latin enables us to give more adequate and unified accounts of these particles than those which are usually found in Latin grammars. It also helps to explain seemingly deviant uses and to draw clearer distinctions between nearsynonyms.

1. Introduction

From antiquity onwards, Latin linguists have described particles such as nam, enim, igitur, ergo, autem, vero and at in terms of the expression of semantic relations between adjacent main clauses, that is, as the coordinating counterparts of subordinating conjunctions: nam and enim are, for instance, considered to express causal relations between independent clauses ('for'), igitur and ergo to express consecutive relations ('so'), and autem, vero and at to express adversative relations ('but'). Consider the overview in Table 1.

This article is a condensed version of Kroon (1995), which contains further details and argumentation. I am very grateful to Rodie Risselada for her comments on a previous version of this paper, and to Shannon Leslie for correcting my English. * E-mail: [email protected] 0378-2166/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved PII S0378-2166(98)00025-3

206

C. Kroon / Journal of Pragmatics 30 (1998) 205-223

Table 1 Traditional account of Latin connectiveparticles

Causal Consecutive Adversative

Coordinating

Subordinating

nam, enim ('for') igitur, ergo ('so') autem, vero, at ('but')

e.g. quia ('because') e.g. ut ('so that') e.g. etsi ('although')

This traditional, semantic-syntactic account of Latin so-called coordinating conjunctions leaves us with a number of problems. For one thing, all 'coordinating conjunctions' mentioned in Table 1 appear to have 'deviant' uses, that is, uses in which the particle concerned cannot reasonably be regarded as indicating a causal, consecutive or adversative relation between two adjacent main clauses. This can be illustrated with examples (1), (2) and (3), which all contain an instance of Latin enim. In (1) enim indeed seems to signal some sort of causal relation; this is, however, less likely in (2), where the causal relation between both clauses is already expressed unambiguously by the subordinator quia; in (3), where the clause containing enim occurs in isolation, a causal interpretation of enim is even outright impossible. A popular, though quite unsatisfactory, solution for this problem is the assumption of two or even more different functions of enim. (1) Iam eum, ut puto, videbo; misit enim puerum se ad me venire 'I think I shall see him soon; he has sent a servant to announce his coming' (enim not translated; Cic. Att. 10.16.5) (2) (Charinus is thinking about going into exile) Eu: Cur istuc coeptas consilium? Ch: Quia enim me adflictat amor Eu: 'What makes you think of taking such a step as that?' Ch: 'Because (quia) I suffer so from love' (enim not translated; PI. Mer. 648) (3) (Mercury in an aside, interrupting Sosia's soliloquy:) Certe enim hic nescioquis loquitur 'Surely (certe), someone is speaking here' (enim not translated; P1. A m . 331) A second problem with the semantic-syntactic account summarized in Table 1 is that it does not give an explanation for differences in distribution that I have observed between alleged synonyms, that is, between items that end up in the same semantic-syntactic category (such as e.g. nam and enim). To mention only one distributional difference between nam and enim: enim can occur in proposals and pieces of advice, whereas there are no attestations of nam in such an environment. On account of differences like this, it would be difficult to maintain that nam and enim are more or less synonymous devices. The same holds for the other groups of markers categorized together in Table 1. These problems, which mutatis mutandis hold also for semantic-syntactic accounts of connective particles in languages other than Latin, can be solved if one approaches connective particles from a more discourse-oriented viewpoint. This has

c. Kroon / Journal of Pragmatics 30 (1998) 205-223

207

been done quite successfully for English connective particles by e.g. Schiffrin (1987), and for French connective particles by the Geneva school of Roulet and others (see e.g. Roulet et al., 1985). Essential assumptions underlying such a discourse approach are: (i) that connective particles are somehow involved in marking the coherence structure of a discourse, and (ii) that discourse structure is a complex system of hierarchical and linear relationships between various types of discourse units, rather than a monolithic sequence of semantically connected grammatical clauses. In this article I propose a theoretical framework for a more adequate description of the Latin 'coordinating conjunctions' mentioned in Table 1, a framework which is in line with these two assumptions and aims to be applicable to the analysis of connective particles in other languages as well.

2. Theoretical framework The theoretical framework on which my description and subclassification of Latin connective particles is based combines elements from the Geneva discourse model (as it was presented in Roulet et al., 1985) with views that are akin to Halliday's meta-functions theory (cf. e.g. Halliday, 1973, 1985) and Schiffrin's discourseplanes approach (Schiffrin, 1987). Characteristic of this framework is its strongly empirical basis and the attention it gives to the linguistic implementation of the theoretical concepts used. Central is the view that coherence relations obtain on three interrelated levels of discourse, for which I use the terms representational, presentational and interactional level of discourse. The characteristic nature of each of these levels is summarized in (4). In 2..1-2.3 I will comment on them and on the role that particles play in signalling coherence on these various levels. (4) Discourse coherence on various levels: 1. representational level of discourse accounts for relations between states of affairs in the represented world 2. presentational level of discourse accounts for a. linguistic action structure: rhetorical relations between communicative acts or moves within a monological stretch of text b. thematic structure 3. interactional level of discourse accounts for linguistic action structure: relations between the constituent moves of a communicative exchange 2.1. Particles on the representational level of discourse The representational level of discourse (in other approaches called also 'ideational level' or 'content level') is concerned with the representation of some real or imagined world outside the language itself. Connective particles with a function on this level usually signal semantic relations between the states of affairs that make up the

208

C. Kroon / Journal of Pragmatics 30 (1998) 205-223

represented world. T h e y often have the form o f subordinating conjunctions which provide a link b e t w e e n g r a m m a t i c a l clauses. In the present article I will leave this category out m y discussion. 2.2. Particles on the presentational level o f discourse The presentational level o f discourse is c o n c e r n e d with the particular w a y in which a s p e a k e r or author ' s t a g e s ' and organizes the represented content. On this level, two interrelated types o f structure can be distinguished: a linguistic action structure and a thematic structure. Particles with a c o h e r e n c e m a r k i n g role on the presentational level indicate h o w discourse units are functionally related to other discourse units within the s a m e m o n o l o g i c a l stretch o f text. S o m e o f them are p r i m a r ily i n v o l v e d in m a r k i n g the linguistic action structure (i.e. they m a r k what I call rhetorical relations b e t w e e n c o m m u n i c a t i v e acts and clusters o f acts), others are prim a r i l y used to m a r k the thematic structure (i.e. the relations b e t w e e n thematic or information units). To illustrate what I m e a n by linguistic action structure on the presentational level o f discourse, c o n s i d e r (5): (5) I ' v e got an extra ticket for the Santa Fe C h a m b e r Orchestra tonight

act --

(So) A r e you interested?

move

act - -

This e x a m p l e contains a m o n o l o g i c a l stretch o f text w h i c h counts as one c o m m u nicative m o v e . This m o v e consists o f two discourse acts w h i c h m a i n t a i n a certain rhetorical relation. 1 The first d i s c o u r s e act in (5) has a s u b s i d i a r y status with r e g a r d to the second, m o r e central, discourse act and can be said to have the m o r e specific rhetorical function o f p r e p a r a t i o n or orientation. This p a r t i c u l a r rhetorical relation m a y be e x p r e s s e d e x p l i c i t l y by m e a n s o f a m a r k e r (e.g. so) or it m a y r e m a i n implicit. It is to be noted that rhetorical relations obtain not only on a relatively local text level (i.e. b e t w e e n simple acts, as is the case in 5), but also on a m o r e g l o b a l text level, b e t w e e n m o r e c o m p l e x discourse units. A hypothetical e x a m p l e o f the latter type o f structure is given in (6):

The terms (discourse) act and move are used in more or less the same way as the terms acte de langage and intervention that are known from the Geneva discourse model (cf. Roulet et al., 1985). Cf. also Edmondson (1981). A (discourse) act can be defined as the smallest identifiable unit of communicative behaviour, while a move is defined as the minimal free unit of discourse that is able to enter into a communicative exchange structure. For more observations on the concepts discourse act (not to be confused with illocutionary act), move and rhetorical relation, I refer to Kroon (1995: 64~57, 73-78, and 1997).

C. Kroon / Journal of Pragmatics 30 (1998) 205-223

209

(6) central act subsidiary move~ - subsidiary act I subsidiary a c t 2 subsidiary act 3 - -

~ - --

extended monologue

central move 2

(so) What we see in (6) is a structure in which a virtually complete movel (which happens to be internally complex) retrospectively turns out to have a subsidiary status with regard to a more central move2 (which, by the way, may be intemally complex as well). Although rhetorical relations on a global text level are usually rather vague, the relation between movel and move2 in (6) could still be described, like in (5), as an orientation or preparation relation. This explains why the discourse marker so can be used in both cases. Note that in the more global construction illustrated in (6), the discourse marker not only signals a particular rhetorical relation between two discourse units (viz. orientation or preparation), but also serves to indicate the macrostructure of the text. As such it resembles to a certain extent markers of the thematic structure of the discourse (about which more will be said below). Typical markers of rhetorical relations in Latin are the particles nam and igitur. (7) and (8) contain examples of igitur, the Latin equivalent of English so and French alors. Igitur indicates that the upcoming discourse unit has a central status with regard to a preceding unit, a unit which has a preparatory (or otherwise subsidiary) function. In (7) igitur signals a local rhetorical relation between two discourse acts: the first sentence functions as a preparation (in this case a justification or motivation) for the request uttered in the second sentence, which counts as communicatively more central: (7) (Beginning of a letter:) Et mihi discendi et tibi docendi facultatem otium praebet. Igitur perquam velim scire, esse phantasmata et habere propriam figuram ... putes ... an ... 'The present recess from business affords you to give, and me to receive, instruction. So, I am extremely desirous to know your sentiments concerning spectres, whether you believe they actually exist and have their own proper shapes .... or ...' (Plin. Ep. 7.27.1) In (8) igitur signals a more global rhetorical relation, of the type illustrated in (6). The 16 paragraphs that precede the segment quoted here function as a preliminary to the narrative proper that the author Cicero restarts here in section 17 of the third volume of his work on oratory, de Oratore. The story that starts in the passage in (8) thus continues the narrative that stopped at the end of the second volume. Igitur forms an explicit signal for the transition from the prologue of the third volume to the continued narrative and hence counts as an important indication for the macrostructure of the entire work: it signals that after some intermediate, preparatory information we are back at a more central part of the text. Thus, what we have

210

c. Kroon / Journal of Pragmatics 30 (1998) 205-223

here is a marker of a rhetorical relation which functions at the sarBe time as a marker of the overall organization of the text.

(8) Ut

igitur ante meridiem discesserunt paulumque requierunt ... 'So, after they had separated before noon to take a brief siesta .... ' (Cic. de Orat.

3.17) The second type of structure that can be distinguished on the presentational level of discourse (cf. 4) is a thematic structure. A monological stretch of text is commonly composed of several hierarchically ordered segments, which are each characterized by their internal thematic coherence. Thematic coherence, in turn, is based on the continuity (or recurrence) of certain information elements. Within a thematically coherent stretch of text various thematic chains may be discerned, along which the discourse evolves. The most pervasive of these chains is referent continuity. In addition one can often distinguish one or more other thematic strands which are operative at the same time, usually of minor importance. In narrative discourse, for instance, thematic coherence usually does not depend on referent continuity only, but also on location continuity, temporal continuity, action-event continuity, and the like. Each interruption of one of these thematic chains (regardless of whether it concerns a referential discontinuity or a discontinuity of one of the other possible thematic strands) may in essence count as a thematic discontinuity and hence - in terms of the organization and segmentation of the discourse - as a discourse boundary. One of the conclusions of my study of the Latin so-called coordinating conjunctions is that one of the three Latin but-equivalents listed in Table 1, namely autem, functions primarily as a marker of the thematic structure of a discourse. Autem is a typical marker of thematic discontinuity, as is illustrated in (9) and (10): (9)

(lo)

(Beginning of a chapter on the events of the year 405 BC:) Fuere autem tribuni T. Quinctius Capitolinus, were on the other hand tribunes T. Quinctius Capitolinus, Q. Quinctius Cincinnatus .... Q. Quinctius Cincinnatus .... 'Now the tribunes were T. Quinctius Capitolinus, Q. Quinctius Cincinnatus, ...' (Liv. 4.61.1) ... nautae coacti fame radices palmarum agrestium ... colligebant et iis miseri perditique alebantur; Cleomenes autem ... totos dies in litore tabemaculo posito perpotabat. Ecee autem repente ebrio Cleomene esurientibus ceteris nuntiatur piratarum esse navis in portu Odysseae; nam ita is locus nominatur; nostra autem elassis erat in portu Pachyni. Cleomenes autem ... speravit ... ' T h e sailors (nautae), as food was so short, had to set about collecting the roots of wild palms ... and they tried to keep themselves alive on these. Now Cleomenes (Cleomenes autem) ... spent whole days drinking in a tent pitched on the shore. And now (ecce autem), while he was drunk and his men starving, the news suddenly arrived that there were pirate ships in the harbour of

C. Kroon / Journal of Pragmatics 30 (1998) 205-223

211

Odyssea (because that was what the place was called). Now o u r fleet (nostra autem classis) was in the harbour of Pachynus. Now Cleomenes (Cleomenes autem) ... counted on ...' (Cic. Ver. 5.87) In (9) autem signals a transition to a new episode in the narrative, which coincides with the introduction of a new temporal frame and of new referents. The text fragment in (10) contains a series of referent shifts plus one instance of a sudden change in the action-event continuity (the shifted elements are in bold). The attention is first on the Sicilian sailors, then on the naval commander Cleomenes. Next there is a sudden change of circumstances (a message is brought that there are pirates in the neighbourhood), which is also marked by autem (ecce autem). Then there is again a shift in attention to the fleet, and finally we are back with the commander Cleomenes. Autem's function of indicating the thematic structure of a stretch of text is not shared by the other two Latin but-equivalents listed in Table 1, which turn out to have coherence-indicating roles on the interactional rather than on the presentational level of discourse. 2.3. Particles on the interactional level of discourse Whereas the presentational level of discourse accounts for the internal structure of a monological stretch of text (so for instance for the ways in which discourse acts are combined to form complex moves), the interactional level of discourse accounts for the coherence of a unit of discourse in terms of the communicative interaction taking place. We can say that the interactional level of discourse pertains to the language in its function of establishing and maintaining interactional relationships between interlocutors. A typical coherence structure on the interactional level of discourse is the exchange structure. Particles with a function on this level of discourse signal for instance how a communicative move (or, more informally speaking, some monological stretch of text) fits into an interactional exchange. Example (11) illustrates a linguistic action structure on the interactional level of discourse. The discourse segment as a whole has the form of an exchange which consists of an initiating move by speaker A and a challenging reactive move by speaker B, the latter of which is explicitly marked as such by but. (11) A: I've got an extra ticket for the Santa Fe Chamber - Orchestra tonight Are you interested? --

act - - -

initiating move - ~

act - exchange

B:

But I thought you didn't like classical music!

--

act - -

reactive move

Among the particles investigated (cf. Table 1) there are two which function primarily as markers of reactive moves within an interactional exchange, namely the

212

C. Kroon / Journal of Pragmatics 30 (1998) 205-223

but-equivalent at and the so-equivalent ergo. 2 These particles operate, in other words, not as markers of the linguistic action structure on the presentational level of discourse (i.e. not as markers of rhetorical relations), but as markers of the linguistic action structure on the interactional level of discourse. Example (12) contains an instance of Latin at in the same function as English but in example (11), that is, as a marker of a challenging reactive move.

(12) A: eloquere B: at pudet A: 'Go on, tell me' B: 'But I'm ashamed to' (P1. Cas. 911) Other Latin equivalents of English but (for instance autem and vero, the ones cited in Table 1 above) do not have this particular function.

3. Distinguishing between 'presentational' and 'interactional' particles: Diaphony Summarizing thus far, we can say that the Latin particles investigated do not all have functions on the same discourse level: nam, igitur and autem have for instance primary functions on the presentational level of discourse, whereas ergo and at turn out to be primarily interactional particles. An important criterion for distinguishing between typically 'presentational' and typically 'interactional' particles is the factor discourse type: presentational particles tend to occur in a monological discourse type, interactional particles in a dialogical discourse type, so for instance in conversations. This is, however, an oversimplified account of the matter, given that a considerable number of instances of ergo and at in Latin literature occur in monologue, conversation being altogether an underrepresented genre in Latin literature. How are we to account for these non-conversational instances within the theory sketched? As to this point the concept of diaphony (or polyphony), as developed especially by the Paris-Geneva school of linguistics, proves to be useful (see e.g. Roulet et al., 1985: 69-84, and Kroon, 1995: 108-115). The integration of this concept in the model proposed is meant to capture the fact that in the essentially monological contribution of a speaker, 'voices' may be integrated of a real or fictitious conversation partner. Although ergo and at occur often, indeed, in a monological environment, the monologues concerned can be demonstrated to have clear dialogical 'traits'. This means that a number of conversational features can be found within the essentially monological environment of a move, without all the formal characteristics of a dialogical discourse being present. In other words, in such a monological discourse type it is suggested (rather than formally expressed) that an interactional exchange is taking place, either between the speaker/author and some implied addressee, or between 2 In contrast to the other Latin so-equivalentigitur, ergo can for instance be used in requests for confirmation. See Kroon (1995: 92-93, 369-370).

c. Kroon / Journal of Pragmatics 30 (1998) 205-223

213

the internal characters of a narrative. This can be illustrated with an example like (13): (13) Tu fortasse orationem, ut soles, et flagitas et expectas. At ego quasi ex aliqua peregrina delicataque merce lusos meos tibi prodo. 'You expect and demand, perhaps, as usual, an oration; but I am going to put into your hands, as if they were some choice bits of foreign merchandise, some of my poetical amusements' (Plin. Ep. 4.14.1) In (13), quoted from a letter by Pliny, we have an interesting phenomenon in that we can describe the succession of the two clauses quoted in terms of all three levels of discourse coherence. First, we can describe the combination of the two clauses in terms of a semantic relation on the representational level of discourse, given that the states of affairs referred to (viz. 'demanding' and 'offering') are somehow related in the represented world. This semantic relationship is, however, not marked explicitly by a particular particle. Second, we can describe the stretch of text in (13) in terms of a rhetorical relation on the presentational level of discourse. From this perspective the clauses are to be described as discourse acts. The first discourse act, which happens to coincide with the first grammatical clause, functions as an orientation or preparation for the second, more central discourse act. Together they may be regarded as one single communicative move. Like the semantic relation involved, this rhetorical relation is not expressed explicitly by a marker either. Finally, we can describe the combination of clauses in (13) in terms of a relation on the interactional level of discourse. That is, the text quoted (which, being phrased by the speaker/author only, is essentially monological) can be regarded also as an interactional exchange consisting of an initiating move (by the addressee of the letter) and a corresponding reactive move by the author: the presence of the particle at (instead of e.g. one of the other ]Latin but-equivalents, cf. Table 1), whose prototypical function is to mark a challenging reactive move in an interactional exchange (cf. 12), can be explained only when the essentially interactional nature of this text segment is taken into account. A comparable example is (14), which is closer to real dialogue in that the speaker/author explicitly quotes the words of some implied interlocutor. More such examples can be found in Kroon (1995: 340-350). (14) sunt hic inter se quos nunc credo dicere: '... novum attulerunt, quod fit nusquam gentium.' at ego aiio id fieri in Graecia ... 'There are some here who, I suppose, are now saying to each other: '... something new this - something that happens nowhere on earth?' But I say it does happen in Greece' (P1. Cas. 67-71) I would contend, thus, that also in examples like (13) at signals a relation, not on the representational or presentational level of discourse, but rather on what I call the interactional level of discourse. An overwhelming majority of the 'monological'

214

c. Kroon /Journal of Pragmatics 30 (1998) 205-223

examples of at in my corpus 3 turned out to occur in a monological context with a clearly diaphonic character, that is, in a context with one or more of the features listed in (15): (15) Features typical of diaphonic monological discourse: the presence (in the host unit or immediate context) of first and second-person pronouns and verb forms the use of present tense verb forms in a passage that is otherwise presented in the past the presence of metadiscursive expressions, i.e. reflections or manifestations of the interactional rather than of the ideational function associated with the discourse involved: performatives (such as I tell you that .... I ask you w h e t h e r ...), metadirectives (tell me, be sure that, r e m e m b e r that .... believe me), and evaluative or procedural expressions of the type I have to admit, it -

-

-

should be stressed, n o w that we have c o m e to this p o i n t in our discussion, as stated above) - the immediate presence of subjective evaluation verbs like arbitror, opinor, credo, puto, mihi videor ('I mean', 'I think', 'I believe', and the like), and of -

-

t

other expressions of subjective mood the use in the immediate context of questions (both rhetorical and non-rhetorical) or of directives, either of which presupposes the involvement of an addressee in the speech event (more than e.g. assertions do) presence of extraclausal interactional elements such as interjections, swear words and vocatives

h

e

Example (16) is more or less comparable with (13) and (14): (16) si tu in legione bellator clues, at ego in culina clueo 'You may be a well known hero in the army, but that's what I am in the kitchen' (PI. Truc. 615) The example represents a small group of remarkable at-instances, occurring in what could be called the apodosis of a conditional clause (an ' t f ... but'-construction, so to speak). The examples have in common that the /f-clause contains, in an indirect, quotative way, the words of an 'embedded voice' against which the speaker himself raises, in the main clause, an objection. Thus, what we have here, is an exchange structure (consisting of two moves) within the limits of one grammatical sentence. The specific interactional relation involved is explicitly marked by at. 4 3 The research reported in this article is based on Latin literary texts from about 200 BC to about 200 AD, with special reference to Latin comedy (Plautus and Terence), the complete works of Cicero (letters, orations, philosophical discussions, rhetoric), historiography (notably Caesar, Sallust, Livy, Curtius and Tacitus), the "novel' Satyricon by Petronius, and Seneca's Letters to Lucilius. Within this corpus, about 400 instances of each of the particles in Table 1 were studied in detail. 4 Note, incidentally, that the remarkable si ... at-constellation might form support for the view that the communicative (i.e. linguistic action) structure of a discourse may be quite independent from its

C. Kroon / Journal of Pragmatics 30 (1998) 205-223

215

4. 'Situating' particles on the interactional level of discourse (extratextual coherence) So far we have been concerned with a classification and description of Latin connective particles that is based on the criterion 'level of discourse'. A further parameter for the description and subclassification of these particles is concerned with whether the relationship indicated by them is strictly textual (i.e. between two verbally expressed units of text), or rather extratextual (i.e. between a unit of text and some non-verbal, extratextual element). The particles discussed so far (nam, igitur, ergo, autem and at) indicate primarily textual relationships (we can say that they are connective particles in a strict sense). The two remaining particles in the lefthand row in Table 1, enim and vero, turn out, however, to be less strictly connective devices. For this extratextual type of coherence particles I use the term 'situating particles': they situate (or perhaps better: evaluate) their host unit against the background of (some element of) the extratextual reality. As such, these particles strongly resemble what are usually called 'modal particles'. This extratextual reality may be for instance the communicative situation in which the text is embedded, including the social relationships between the interlocutors, their interactional goals and intentions, their general and situation-bound knowledge, the views of the interlocutors about the knowledge, attitudes and communicative intentions of their discourse partners, and so on. Although situating particles are not connective in a strict sense, they can still be seen as coherence devices, since they fit their host unit, in one way or another, in its non-verbal context. For the Latin particle enim this extratextual, 'situating' function appears, for instance, from examples such ;as (3), in which the particle occurs in an isolated clause. In such cases it is impossible to describe the function of enim in terms of indicating a connection between two verbally expressed text segments. As a matter of fact, such a description is not warranted in (1) and (2) either. From a thorough investigation of enim-instances in Latin literature it appears (Kroon, 1995: ch. 8) that enim functions primarily as an appeal to the involvement and cooperation of the addressee in the speech event. By using enim the speaker or author confirms or suggests that there is a certain consensus between himself and his addressee or audience. The speaker pictures the addres,;ee, so to speak, as being on the same 'wavelength'. As such, enim can be said to play a role in the management of the interaction, and to be more or less comparable with German ja and English y'know. In many cases the precise function of enim appears to be rendered best in the form of a tag-question grammatical structure: in terms of communicative structure, so-called 'apodotical' at can be said to connect two alternating moves within an interactional exchange; in terms of grammatical structure, however, we are dealing with a connection between a subordinate and a main clause, which are uttered by one single person. In other words, the use of at in pseudo-apodotical clauses seems to demonstrate that grammatical sentences may have an internal communicative structure. This observation would form a major complication for discourse grammars that take the grammatical clause or sentence as a starting point on the basis of which larger communicativestructures (such as moves and exchanges) are built. Cf. Kroon (1997) for a discussion of this complication within the context of Functional Grammar.

216

c. Kroon /Journal of Pragmatics 30 (1998) 205-223

('isn't it?'). An important clue to this particular interactional function of enim in Latin discoure is the type of discourse it tends to be associated with. In contrast with its alleged synonym nam (which is a more genuine equivalent of English 'for' or 'cos') enim occurs predominantly in dialogical discourse or in monological discourse with clear diaphonic features (i.e. in a monological context which has one or more of the features listed in 15). The causal connective connotation enim seems to have in instances like (1) should be explained as a regularly occurring side-effect of the particle's proper function of marking interpersonal consensus. It is quite understandable that consensus markers occur often in discourse units that have a subsidiary status with regard to a preceding discourse unit (i.e. in units that function as justification, explanation or evidence with regard to a preceding text segment), because such units usually function to make a preceding, potentially challengeable utterance more acceptable. The use of a consensus marker like enim can be regarded as an additional device for toning down the challengeability of a preceding utterance in that it lends the subsidiary text unit the character of a common sense argument. In other words, enim is very compatible with a specific, rhetorical type of causal relation (motivation, explanation, justification, and the like), but does not, by itself, signal such a relationship. Rather enim lends subsidiary discourse units of this 'causal' type the additional feature of a common sense argument. Or, in more general terms, rather than having a strictly connective function on the presentational level of discourse (as nam turns out to have, as is illustrated in 18), enim is involved in fitting its host unit into a particular communicative setting. I therefore classify enim as a marker of extratextual coherence relations on the interactional level of discourse.

(17) Quae res (sc. tempestas) magnas difficultates exercitui Caesaris attulit. Castra enim, ut supra demonstratum est, cum essent inter flumina duo ... neutrum horum transiri poterat ... 'This (i.e. the storm and the consequent breaking down of the bridges) caused serious difficulties to Caesar's army. For (enim) the camp being situated, as has been explained above (ut supra demonstratum est), between two rivers ... neither of these could be crossed ...' (Caes. Civ. 1.48) (18) Is pagus appellabatur Tigurinus; nam omnis civitas Helvetia in quattuor pagos divisa est 'The name of the canton was the Tigurine; f o r the whole state of Helvetia is divided into four cantons' (Caes. Gal. 1.12.4) In (17), by using enim, the author/narrator makes an appeal to the reader, thus bringing to the surface, in his narrative, the interactional frame in which the text is 'embedded'. This interactional frame is made explicit also by the procedural remark ut supra demonstratum est ('as has been explained above'). In (18), which has an identical rhetorical structure consisting of a central and a subsidiary act, this interactional frame of speaker and addressee is, so to speak, out of sight. In (18) the rhetorical relation of explanation between the second and the first clause is expressed explicitly by nam. In (17), on the other hand, this relation is left implicit.

C. Kroon / Journal of Pragmatics 30 (1998) 205-223

217

A comparable line of reasoning applies to the alleged adversative coordinator vero. Like enim it is not a connective particle in a strict sense, but rather a situating particle with a strongly interactional nature. By means of vero a speaker or author

indicates his personal commitment with regard to one or more aspects of the communicative act he is performing, thus enhancing the trustworthiness or credibility of the utterance. The particle could be rendered in English by 'really' or 'you can take my word for that'. It is found especially in contexts with a highly surprising, salient or controversial import, that is, in the immediate environment of a strong focus element or informational climax. An example is (19): (19) (On the necessity of avoiding crowds:) Nihil vero tam damnosum bonis moribus quam in aliquo spectaculo desidere 'Nothing, really, is so damaging to good character as the habit of lounging at the games' (Sen. Ep. 7.2) It may be noted that, in a sense, vero and enim are complementary means: enim is used to solicit commitment on the part of the addressee, vero to indicate commitment on the part of the speaker himself (this explains, incidentally, why they are often combined within the same clause). As such, both can be said to be involved in the management of the interaction, and thus to have a function on the interactional level of discourse. The adversative connotation that vero may seem to have in monological texts is to be described, in my view, as a side-effect of its proper, interaction-monitoring discourse function. This can be illustrated with (20), which in the authoritative Oxford Latin Dictionary is mentioned under the heading 'vero with stronger adversative force', with the translation 'however' or 'yet'. In my opinion the occurrence of vero in (20) is not to be explained by the presence of some adversative relationship with the immediately preceding context, but rather by the highly remarkable and salient content of the vero-unit itself, which pushes the use of an interactional, 'speakerauthority' marker. Indications in this example for such an interpretation are the presence of a cumulative string ('nine months, six months, four months, three months, only a few hours'), and of the superlative form paucissimis, which means 'very few'. Vero in this example should, in my opinion, not be rendered by 'but', but rather by something like 'really' or 'believe it or not'. (20) ceteros (sc. consulatus) aut novem aut sex aut quattuor aut tribus mensibus (sc. gessit), secundum vero paucissimis horis 'The rest (viz. of the consulships) he held for nine, six, four, or three months, (but?) the second (secundum vero) lasted only a few hours' (vero: 'believe it or not', 'really'; Suet. Aug. 26.4)

5. Summary: Revised account of Latin so-called coordinating conjunctions In summary, by application of two main parameters (viz. level of discourse and intratextual/extratextual coherence), the rather unsatisfactory picture given in Table

218

C. Kroon / Journal of Pragmatics 30 (1998) 205-223

1 can be replaced by a much more sophisticated system of Latin discourse markers, a simplified version of which is given in Fig. 1.

Extratextual

Intratextual Thematic structure

Linguistic action structure

Presentational level

Thematic chains

I

Rhetorical relations

i

So-called adversative

autem

Interactional level

Interactional level

Interactional relations

Situating

at

vero

I

So-called causal So-called consecutive

j I I I I

enim

ham

igitur

ergo

I I

Fig. 1. Revised account of Latin so-called coordinating conjunctions.

Instead of three semantic categories of coordinating conjunctions (the situation in Table 1), we end up with at least three function-groups of particles: (i) first, a group including autem, nam and igitur, which all have a primary function on the presentational level of discourse, and usually behave as connective particles in a strict, textual sense; this function-group can be subcategorized further according to whether the particle involved marks the thematic structure of a discourse (autem), or rather the rhetorical structure (nam and igitur); (ii) then, there is a second group including at and ergo, which behave also as connective particles in a strict sense, but which have a primary function on the interactional level of discourse; and (iii) a group including enim and veto, which are both primarily 'situating' particles with a function on the interactional level of discourse (the adversative and causal connotation that these latter particles undeniably have is to be described in terms of a regularly occurring side-effect of their use in a particular context), s From the overview in Fig. 1 it is clear, furthermore, that all alleged synonyms are in fact clearly distinct items. The three Latin but-equivalents (autem, vero and at), 5 For the sake of completeness I add here that in addition to the category 'situating (i.e. extratextual) particles with a function on the interactional level of discourse', we should distinguish also a category 'situating particles with a function on the representational level of discourse'. For an example of a situating particle that functions (also) at the representational level, see Risselada (this issue). As this latter category is less relevant for the markers discussed here, I have not incorporated it in the picture. I leave undiscussed also the issue of whether we are to distinguish a category 'situating particles with a function on the presentational level of discourse'.

C. Kroon / Journal of Pragmatics 30 (1998) 205-223

219

for instance, end up in three different function-groups, which is in accordance with the observation that they have a clearly distinct distributional behaviour. It should be emphasized here that classification within the same function-group does not mean that the particles concerned have identical discourse functions. In addition to the function-group to which it belongs, the more specific discourse function of an individual particle is determined by an abstract basic meaning, which differs from particle to particle. For instance, enim and vero are both classified as 'situating particles with a function on the interactional level of discourse', in the category 'extratextual relationships'. However, whereas vero indicates the commitment of the speaker, enim is used to solicit the commitment of the addressee. This difference in discourse function is connected with a difference in their basic meanings: 'actuality' or 'reality' in the case of vero, 'consensus' in the case of enim. Furthermore, the interplay of the discourse function of a particle (which is the pivotal element in the analysis and description of a discourse particle in that it determines the more distinctive features of a particle in contradistinction to other particles) with properties of its context may lead to a number of different actual uses. For enim, for instance, at least three different actual uses can be distinguished: (i) toning down the challengeability of a (polemic, controversial, etc.) preceding utterance (as is the case in example 1, in which the challengeability of the preceding utterance is already clear from the addition of the hedge ut puto ('I think'); (ii) displaying irony in the case of a patently false statement; (iii) soliciting empathy, e.g. in lively reports or eyewitness accounts, as is for instance the case in example (3). Actual use (i) may give rise to a side-effect of causality (cf. the discussion above), actual use (iii) to a side-effect of reinforcement of (part of) the utterance. In Latin grammars and dictionaries these side-effects of enim have often been mistaken for the particle's proper function. This system,6 in which a distinction is made between basic meaning, discourse function, actual use, and side-effect, may count as a more sophisticated explanation of a situation that has often been described as mere chaos or an extreme form of polysemy. It is summarized in the flow chart in Fig. 2. 1. basic meaning <---> levels of discourse

$

2. discourse function <---> properties of context

$

3. actual use

$

[4. side effects] Fig. 2. Descriptive format.

6 Cf. Foolen (1989: 309) for a broadly comparable proposal for a multi-stage description of modal particles in German, and Rissetada's adaptation of this proposal for the description of the Latin modal particles modo and sane (Risselada, 1994, this issue).

220

C. Kroon / Journal of Pragmatics 30 (1998) 205-223

As an illustration of the type of fully-fledged functional analyses I propose for the Latin discourse markers concerned, consider the analysis of the alleged but-equivalent autem, summarized in (21). A broadly comparable analysis is given by Risselada in this issue for the Latin discourse marker sane. (21) Illustration: Functional analysis of Latin autem (i) basic meaning: 'distinctiveness'. (ii) discourse function: indication of the discrete status of a text segment in relation to its preceding context (presentational level of discourse) (iii) actual use: a. local: marker of (contrastive or parallel) focus b. global: marker of the organization of the text (more specifically: marker of thematic discontinuity) (iv) side effects: a. reinforcement of the illocutionary intention of the speaker/writer, or indication of his specific emotional attitude b. (semantic or rhetorical) adversativity, causality In (21) we can read that there are two actual uses of the particle autem which both emanate from one and the same discourse function. This discourse function is associated with the presentational level of discourse. The basic meaning of 'contrast' that is intuitively felt to be present in all cases (and which led Latin grammars to rank autem among the adversative conjunctions) is to be associated with the discoursepragmatic notions of distinctiveness and demarcation rather than with semantic adversativity. Depending on whether it is applied locally or with a more global scope, the particle is used as a focus marker (see example 22, in which autem is involved in signalling contrastive focus), or as a marker of thematic discontinuity (see examples 9 and 10 above). (22) Tu magnus amator mulierum es, Messenio, ego autem homo iracundus 'You, Messenio, are a great lover of the ladies, while I, on the other hand, am a choleric man' (P1. Men. 268-269) Alleged functions of autem as a marker of adversative or even causal relations, or as a modal particle indicating the emotions (indignation, surprise, impatience, etc.) of the speaker, can be shown to be no more than incidental side-effects of the use of autem in a particular context. More details on this functional analysis of autem can be found in Kroon (1995: ch. 10).

6. Heuristics As to the analysis of particles in general and discourse markers in particular the question remains, of course, how in practice the various components of their functional definitions are to be determined. This question is all the more relevant for a

C. Kroon /Journal of Pragmatics 30 (1998) 205-223

221

language like Latin, for which we cannot rely on the intuitions of native speakers. The obvious method is to start the analysis of a particle by surveying its recurrent, different uses in a sufficiently large and heterogeneous corpus, on the evidence of which one or perhaps two discourse functions can be deduced and, eventually, one basic meaning. This inventory of uses of a particle is not a random affair, and should make appeal to observable criteria. Hence, in addition to the interpretation of the content of the host unit and its immediate context (which is to a certain extent subjective and ad hoc), the inventory of uses of a discourse particle should rely particularly on objective linguistic clues. (23) contains a (non-exhaustive) list of possible syntactic, discoursepragmatic and other properties of the immediate context of a discourse particle. These properties can be used as heuristic clues for determining the discourse function of a particular particle, in contradistinction to other particles of the same semantic-syntactic category. Usually the discourse function of an individual particle is not revealed by one single clue from the list, but by the interplay of a number of clues, in contrast with other particles involving different sets of distributional properties. It is to be noted moreover that not all of the clues are equally relevant for each particle. (23) Heuristic clues for determining the function of discourse markers: discourse type: dialogical, monological or diaphonic - collocations with other particles - illocutionary force of the clause in which the particle occurs communicative structure (i.e. position of the host unit in the hierarchically organized, communicative structure) information structure a. global: thematic structure b. local: focus marking - syntactic status of the clause in which the particle occurs (subordinate or main) - tense in narrative discourse genre and text type -

-

-

-

For explanations of the items in this list and for more details I refer to Kroon (1995: 115-125, forthc., on the criterion tense in narrative texts). From the discussion above it will be clear now that discourse markers from the same function-group (see Fig. 1) have more distributional properties in common than items that belong to different function-groups. For instance, the alleged but-equivalent vero shares more distributional properties with the alleged for-equivalent enim, than with the alleged but-equivalent autem. Consider (24) and (25). (24) Distributional similarities between vero and enim: (i) vero and enim both combine readily with unequivocal connectives. This may imply that they do not themselves mark sequential discourse relationships, i.e. that they are not connective in a strict, textual sense. This is also suggested by the fact that both are able to occur in relative clauses;

222

C. Kroon / Journal of Pragmatics 30 (1998) 205-223

(ii)

v e r o and e n i m both occur typically in dialogical or diaphonic discourse. This is indicated, among other things, by the fact that they readily cluster with other interactional elements such as interjections, subjective evaluation verbs and first and second person pronouns; both are, moreover, quite common in rhetorical questions. In narrative texts v e r o a n d e n i m typically occur in (in)direct speech, or in passages with a subjective or emotional colouring (contemplation by the author, narrative climaxes in the historic present, and the like); (iii) v e r o and e n i m both tend to occur in challengeable (i.e. surprising, controversial, polemic, etc.) contexts. This is indicated e.g. by the frequent combination of v e r o and e n i m with a t (which has been described above as a typical introducer of challenging reactive moves), and with negation particles. (25) Distributional differences between v e r o and a u t e m : (i) a u t e m tends to cluster with full nominal phrases which usually have the status of shifted discourse topics. This tendency cannot be observed for v e r o , which follows more or less equally all kinds of elements, with a slight preference for collocation with (resumptive) pronouns. The combination of a u t e m with a resumptive pronoun is rare; (ii) interrelated with (i): in contrast to a u t e m , v e r o is not systematically used at major thematic transition points (paragraph boundaries); (iii) in contrast to a u t e m , v e r o is not systematically used in the minor premiss of a syllogism; (iv) v e r o and a u t e m differ significantly in their possibilities of combining with other particles. A u t e m is practically limited to combinations with enumerative particles or adverbs. V e r o , on the other hand, tends to cluster with 'focus' particles or adverbs, with negation particles, with challenging reaction particles such as at, and with other interactional particles such as e n i m . These combinations are excluded or extremely rare with a u t e m ; (v) v e r o has a preference for dialogical or diaphonic discourse, a u t e m for monological discourse.

7. Concluding remarks In this article I have sketched the outlines of a model for the analysis and description of discourse markers, based on an empirical study of a number of Latin connective particles. The theoretical framework proposed (which combines ideas of the Geneva discourse model with those of Schiffrin) has the advantage over Schiffrin's discourse planes approach of being more homogeneous: whereas Schiffrin fails, in my opinion, to make sufficiently clear how the five discourse planes she distinguishes interact, 7 the three levels of discourse that form the basis of my framework 7 For the same criticism, cf. Fraser (1990) and Redeker (1991). The five discourseplanes in Schiffrin's model are called 'ideational structure', 'action structure', 'exchange structure', 'participation structure' and 'information state'. The exact theoreticalstatus and definitionsof these planes remain unclear.

C. Kroon / Journal o f Pragmatics 30 (1998) 205-223

223

are m o r e clearly interrelated and motivated. M o r e o v e r , m u c h attention is given to the linguistic criteria u n d e r l y i n g the distinctions m a d e . A s for Latin, the a p p r o a c h a d o p t e d has the a d v a n t a g e o v e r traditional, semanticsyntactic accounts o f L a t i n connective particles o f p r o v i d i n g a m o r e unified picture o f the particles i n v o l v e d , in w h i c h the interrelatedness b e t w e e n the various uses o f a particular particle is m a d e explicit, and in w h i c h ostensibly d e v i a n t instances can be e x p l a i n e d rather easily. It enables us, m o r e o v e r , to distinguish m o r e clearly b e t w e e n near-synonyms.

References Edmondson, Willis J., 1981. Spoken discourse: A model for analysis. London: Longman. Foolen, Ad, 1989. Beschreibungsebenen ftir Partikelbedeutungen. In: H. Weydt, ed., Sprechen mit Partikeln, 305-317. Berlin: De Gruyter. Fraser, Bruce, 1990. An approach to discourse markers. Journal of Pragmatics 14: 383-395. Halliday, M.A.K., 1973. Explorations in the functions of language. London: Arnold. Halliday, M.A.K., 1985. An introduction to Functional Grammar. London: Arnold. Kroon, Caroline, H.M., 1995. Discourse particles in Latin. A study of nam, enim, autem, vero and at. Amsterdam: Gieben. Kroon, Caroline H.M., 1997. Discourse markers, discourse structure and Functional Grammar. In: John H. Connolly, Roel M. Vismans, Christopher S. Butler and Richard A. Gatward, eds., Discourse and pragmatics in Functional Grammar, 17-32. Berlin: De Gruyter. Kroon, Caroline H.M., forthc. Particles, tense, and the structure of Latin narrative texts. In: R. Risselada, ed., Latin in use. Amsterdam studies in the pragmatics of Latin. Amsterdam: Gieben. Redeker, Gisela, 1991. Linguistic markers of discourse structure: Review article of Discourse markers, by D. Schiffrin. Linguistics 29:1139-1172. Risselada, Rodie, 1994. Modo and sane, or what to do with particles in Latin directives. In: J. Herman, ed., Linguistic studies on Latin, 319--343. Amsterdam: Benjamins. Risselada, Rodie, 1998. Latin sane: Marker of agreement in description, interaction and concession. Journal of Pragmatics 30:225-244 (this issue). Roulet, Eddy, A. Auchlin, J. Moeschler, C. Rubattel and M. Schelling, 1985. L'articulation du discours en franqais contemporain. Bern: Lang. Schiffrin, Deborah, 1987. Discourse markers. Cambridge: Cambridge University Press.