the italian modal dovere in the conditional: future reference, evidentiality and argumentation

19
© Cahiers Chronos The Italian modal dovere in the conditional: future reference, evidentiality and argumentation. Andrea ROCCI University of Lugano (Switzerland) 1. Broader context of the research The present paper is at the same time a development of an ongoing research on epistemic modality in Italian and the very first step into a research on argumentation, modality and evidentiality in economic-financial news discourse that the author is currently developing. 1.1 Relative modality and the semantics of the modals Here, we will be drawing on previous work on the epistemic readings of the Italian modals potere and dovere (Rocci 2005a and 2005b), of the Italian future tense (Rocci 2000 and 2005a), as well as of other Italian modal constructions (Rocci 2007a). This work explained how these linguistic structures can function as markers of inferential evidentiality and can establish argumentative discourse relations, using a version of the relative modality approach initiated by Kratzer (1981). According to this approach modals are to be analysed as two-place relations, of the form R (B, p), taking as their arguments a set of propositions – their conversational background (B) – and a proposition (p) corresponding to the complement of the modal. Necessity modals are taken to indicate that the argument proposition is necessarily entailed by – that is logically follows from – the conversational background (B) of the modal. (Def. 1) [[must/ necessarily (B, ϕ)]] [[ϕ]] is a logical consequence of B This means that the universal quantification over possible worlds of the necessity operator is restricted to the set of worlds where all the propositions in the conversational background are true. This set of worlds is called the modal base. In fact, within possible world semantics, saying that ϕ is a logical consequence of B means exactly that ϕ is true in all the worlds where all the propositions that make up B hold, that is in all the worlds of the modal base 1 . 1 A sentence ϕ is a logical consequence of a set of sentences Φ if and only if ϕ is true in all possible worlds in which all sentences in Φ are true. (Cf. Kaufmann, Condoravdi and Harizanov 2006).

Upload: usi-ch

Post on 22-Nov-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

© Cahiers Chronos

The Italian modal dovere in the conditional: future reference, evidentiality and argumentation.

Andrea ROCCI University of Lugano (Switzerland)

1. Broader context of the research

The present paper is at the same time a development of an ongoing research on epistemic modality in Italian and the very first step into a research on argumentation, modality and evidentiality in economic-financial news discourse that the author is currently developing.

1.1 Relative modality and the semantics of the modals

Here, we will be drawing on previous work on the epistemic readings of the Italian modals potere and dovere (Rocci 2005a and 2005b), of the Italian future tense (Rocci 2000 and 2005a), as well as of other Italian modal constructions (Rocci 2007a). This work explained how these linguistic structures can function as markers of inferential evidentiality and can establish argumentative discourse relations, using a version of the relative modality approach initiated by Kratzer (1981).

According to this approach modals are to be analysed as two-place relations, of the form R (B, p), taking as their arguments a set of propositions – their conversational background (B) – and a proposition (p) corresponding to the complement of the modal. Necessity modals are taken to indicate that the argument proposition is necessarily entailed by – that is logically follows from – the conversational background (B) of the modal.

(Def. 1) [[must/ necessarily (B, ϕ)]] ⇔ [[ϕ]] is a logical consequence of B

This means that the universal quantification over possible worlds of the necessity operator is restricted to the set of worlds where all the propositions in the conversational background are true. This set of worlds is called the modal base. In fact, within possible world semantics, saying that ϕ is a logical consequence of B means exactly that ϕ is true in all the worlds where all the propositions that make up B hold, that is in all the worlds of the modal base1. 1 A sentence ϕ is a logical consequence of a set of sentences Φ if and only if ϕ is

true in all possible worlds in which all sentences in Φ are true. (Cf. Kaufmann, Condoravdi and Harizanov 2006).

The relative semantics of the necessity modals can be also given equivalently “by translation” using an unrestricted logical necessity operator ( ) and the material entailment connective (→):

(Def. 2) Must/ Necessarily (B, ϕ) ⇔ ( B → ϕ)

This indirect formulation involves a strictly unnecessary step but will come in handy for descriptive purposes.

Within the relative modality approach the various interpretations of the modals can be understood in terms of the kind of propositions that compose the conversational background, or, equivalently, by the kind of worlds in the modal base. Three broad classes of conversational backgrounds are relevant to our discussion of the interpretations of the Italian modal dovere: alethic, deontic and doxastic backgrounds. We can render modalities with alethic (realistic backgrounds according to the terminology of Kratzer 1981) conversational backgrounds using the following informal paragraph schema:

(i) A possible state of affairs ϕ is entailed by or compatible with facts of kind B.

Deontic conversational backgrounds can be characterized by the formula:

(ii) An action α is entailed by or compatible with norms or ideals of kind B

And, finally, we have doxastic conversational backgrounds:

(iii) A (meta-represented)2 hypothesis ϕ is entailed by or compatible with a relevant set of beliefs B held by the speaker at the moment of utterance.

Research on the necessity modal dovere (Rocci 2005a and 2005b), in particular, suggested that in the epistemic readings of dovere in the indicative the conversational background was to be identified with a set of beliefs of the speaker at the moment of utterance, and thus with a doxastic modal base3.

2 Cognitively, epistemic/doxastic modalities relate to the higher faculty of

metarepresentation: that is the ability of an agent to represent his/her thoughts as representations distinct from the world, thus enabling the agent to cope with her partial and fallible access to the facts. In doxastic modalities the proposition ϕ is entertained as a metarepresentation – as a thought – and compared with the agent’s beliefs. For a detailed discussion of the role of metarepresentation in the analysis of epistemic modality see Papafragou (2000).

3 Note that this is not the “standard analysis” of the epistemic interpretation of the modals proposed for German and English by Kratzer (1981) and adopted by much of the subsequent literature, which does not involve doxastic backgrounds. The doxastic analysis of deve proposed in Rocci (2005a and

Only in epistemic-doxastic backgrounds the logical consequence relation is identified directly with an inferential, and, hence, possibly argumentative relation. With the other backgrounds this coincidence may or may not obtain (Rocci 2007c).

In the following pages we will recall some consequences of this analysis, as well as some of the arguments that support it in order to illuminate the contrast between the epistemic readings of the indicative (henceforth deve) and those of the conditional (henceforth dovrebbe), which react in a distinctively different manner with respect to a number of semantic and discursive variables, including notably the condition under which it is possible to establish argumentative discourse relations between the modalized utterance and adjacent utterances in the co-text.

1.2 Modality and the structure of argumentative discourse: the case of economic-financial news

The discourse genre of economic-financial news is an interesting test bed for investigating the contribution of the semantics of the modals to discourse organization and to argumentation in particular.

In contrast with other news genres, financial news are as much about predicting the future and evaluating possible outcomes than about reporting the past events (Cf. Del Lungo Camiciotti 1998, Walsh 2004). Future events – both in the form of forecasts and of alternative conditional scenarios – receive the same importance as the reporting of past events. Explicit argumentation supporting acts of prediction4, is also more prominent than in other news genres. This supporting argumentation is largely attributed to expert sources (e.g. financial analysts) and sometimes accompanied by further indirect argumentation on the source’s credibility.

At a semantic level this has a number of interesting consequences. Predictions appearing in financial news typically take the form of modalized utterances (cf. Walsh 2001, 2004 and 2006) referring to future events. These modalized statements typically appear as conclusions in arguments based on causal argumentation schemes (Walton 1996). Predictions and causal relations are often relativized to plausible or merely possible scenarios. Consider, as a partial illustration of these features, the following English

2005b), in this respect, is more closely related to the analysis of epistemic must in terms of metarepresented beliefs proposed by Papafragou (2000).

4 Merlini (1983) addresses predictions in economics papers from the viewpoints of Searlean Speech Act Theory. The analysis reveals an intimate connection between the nature of the illocutionary force of prediction and its role in argumentation. Merlini (1983) devotes particular attention to the conditional nature of predictions and the role of epistemic modalities in modifying the prediction along an epistemic gradient and an evidential-inferential one.

examples, taken from an article appearing in the Wall Street Journal Europe (WSJE) on September 14, 20065:

(1.a) Firm's fortunes may rise as commodity prices fall (Headline) (1.b) Major airlines around the globe continue to see strong passenger demand, so

profit could climb if they are able to raise prices while their own costs drop as fuel prices fall. (From the body text)

The headline in (1.a) consists of a modalized conclusion supported by an argument based on a form of “economic causality”, while the passage in (1.b) presents a more developed form of the same argument where the modalized conclusion introduced by (could) holds only within a conditional frame (if they are able to raise prices).

Explicit attribution to sources (financial analysts, rating agencies, etc.) and forms of quotative evidentiality6 create another kind of shifted discourse domain which interacts with conditional structures and epistemic modals, as illustrated by (2):

(2) A reduction of that percentage to 30% would likely lead Standard & Poor's to raise the company's corporate credit rating to "stable" from "negative," according to primary credit analyst Mary Ellen Olson. (WSJE, February 13, 2007)

The above semantic features make this discourse genre an ideal –and largely unexploited – environment to explore the interaction between modals, evidentials and conditional structures and their contribution to discourse organization. While the present paper basically remains a semantic study of the epistemic readings of Italian conditional modal dovrebbe, it can also seen as a first step in the exploration envisaged above.

For the present study of the discursive implications of the semantics of dovrebbe We will rely on a small corpus of examples collected from Italian ecomomic-financial daily Il Sole 24 Ore7.

2. Deve and dovrebbe: different degrees on a certainty scale?

5 The whole article contains 12 modal expressions (modal auxiliaries and modal

adverbs). 6 Interestingly Heard on the Street is the title of the recurring feature to which the

source article of examples (1.a,b) belongs. This title nicely points to the role of expert sources (sometimes anonymous) in this kind of discourse.

7 All the examples are taken from a small sample of 126 occurrences of the “epistemic” use of the conditional dovrebbe, extracted from the electronic archives of the Italian economic daily Il Sole 24 Ore. For all the occurrences extracted, the entire texts of the articles have been included in the corpus.

The epistemic interpretations of the modal dovere (‘must’) in the indicative mood (henceforth deveE) and in the conditional mood (dovrebbeE) – are exemplified in (3) below.

(3) A : Dov’è lo zucchero? (Where is the sugar ?) B1: Dev’essere da qualche parte nella credenza. (It must be somewhere in the cupboard) B2 : Dovrebbe essere da qualche parte nella credenza (It should be somewhere in the cupboard)

One way of looking at the difference between the two modals is to say that the conditional form dovrebbe is to be understood as a downgraded version of epistemic deve on a certainty scale. Certainty scales are a simple and much used way to represent differences between epistemic meanings.

Pietrandrea (2005: 70-76), for instance, presents the following one, which she considers as one of the key scalar dimensions structuring the epistemic domain in Italian:

DEVE

DOVREBBE

PUÒ

+ certain

- certain

However she also observes:

The conditional morpheme [= the one appearing in dovrebbe] is not an epistemic marker per se. It merely indicates that the truth of the modal that it modifies holds in a world where certain condition are met. That increases the degree of uncertainty only indirectly. (Pietrandrea 2005:76).

Here we follow this hint on indirectness, to argue that often scalar paradigms of this sort do not reflect meaning construction processes, but rather offer a map – a one-dimensional map actually – of post-pragmatic meaning effects. We will also argue that a description mirroring more closely meaning construction can lend deeper insights into discourse functions.

This issue is particularly relevant with respect to argumentative discourse. Modal scales such as the above are particularly prominent as an analytic tool in studies addressing modality in specific discourse genres. Consider, for instance, the classic functional treatment of modality in

Halliday (1994: 354-363)8 and its many applications in discourse analysis. Moreover, the tradition of argumentation studies, since Toulmin (1958), has essentially considered “modal qualifiers” as indicators of the degree of strength of a conclusion9, or, similarly, of the extent of commitment to the truth or acceptability of the propositional content of a standpoint (Snoek-Henkemans 1992).

Here, without discounting the indubitable relevance of scalar phenomena for the understanding of the quantificational aspect of modality (Cf. Horn 1972), we want to suggest that there are subtle differences between modals that have more to do with the nature and internal structure of their modal bases (or conversational backgrounds) rather than with their basic force of quantification, and that these differences have interesting consequences at the discourse level.

2. Deve vs Dovrebbe

A first observation suggesting that the difference between epistemic deve and dovrebbe cannot be reduced to different positions on a certainty scale is the following: in most of the contexts in which they occur, the two modals are not interchangeable. Consider (4.a,b):

(4.a) Ciò detto, ci dobbiamo preparare ad agire in un mercato negativamente influenzato dall' andamento fiacco del Pil, un mercato che nel 2002 dovrebbe calare dell' 8% in Italia a 2,2 milioni di vetture [...] (Corpus IL SOLE 24 ORE).

‘That said, we must be prepared to act in a market negatively influenced by the weak GDP figures, a market that in 2002 is expected to decrease by 8% in Italy to 2.2 million cars’

(4.b) Poco dopo la seconda guerra mondiale, ha ricordato, l' Italia ha vissuto il famoso miracolo economico, diventando in poco tempo uno dei Paesi più ricchi del mondo. Dev' essere scattata una molla particolare per innescare un processo di sviluppo così rapido, ha sottolineato l' economista. (Corpus IL SOLE 24 ORE)

8 Certainly the complex system of choices related to modality that Halliday

postulates goes beyond the scale encompassing high, median and low modalities. However this scalar dimension remains primary in the whole conception: “Modality is the area of meaning that lies between yes and no – the intermediate ground between negative and positive polarity” (Halliday 1994: 356).

9 According to the “Toulmin Model” of argumentation, the category of the modal qualifier is meant to provide an “explicit reference to the degree of force which our data confer to our claim in virtue of our warrant” (Toulmin 1958: 101).

‘Shortly after WWII – he reminded – Italy witnessed the famous Economic Miracle, becoming in a short time one of the richest countries in the world. Some very particular trigger must have gone off that sparkled such a rapid economic development process – the economist emphasized’

In (4.a) the substitution of dovrebbe with deve creates a bizarre effect of absolute certainty – which is not at all the usual “strong probability” meaning of epistemic deve: the “8% decrease” of the automotive market in Italy would be presented as an inevitable necessity. On the other hand, in (4.b) changing deve to dovrebbe destroys discourse coherence, making it impossible to consider the event of the ‘Economic Miracle’ as evidence that ‘some special trigger went off’. In other words, it becomes impossible to establish an argumentative discourse relation where the modalized proposition is interpreted as a conclusion, and the fact mentioned in the previous utterance as a premise. As it will be argued in the following sections, it is possible to account for the lack of mutual substitutability of deve and dovrebbe in contexts such as the above by looking at some basic differences between the two constructions. These differences emerge at three levels:

a. The possibility of epistemic interpretation of the modal when the embedded proposition refers to a future eventuality;

b. The presence of restrictions on the evidential source of the modalized proposition;

c. The presence of restrictions on the type of argumentative discourse relations inferable.

3. Epistemic dovere and future reference.

Elsewhere (Rocci 2005a) we have argued that epistemic interpretations of deve are completely incompatible with embedded propositions referring to future events. This means that, most of the times, epistemic interpretations of deve are available only when there is a perfect infinitive (5.a), or the simple infinitive of a stative predicate (5.b) as had been already observed by Bertinetto (1979 and 1986).

(5.a) Giovanni non è ancora arrivato. Deve essersi perso. ‘Giovanni hasn’t arrived yet. He must have lost his way.’ (5.b) Giovanni deve conoscere una scorciatoia. Perché è arrivato là prima di tutti. ‘Giovanni must know a shortcut. Because he arrived there before everybody

else’.

(5.c) Giovanni non conosce la strada. *Deve perdersi.10 ‘Giovanni does not know the way. *He must get lost.’

In (5.c), with a non-stative eventuality, the modalized process is interpreted as having future reference, and, at the same time, the epistemic reading of the modal deve is blocked. Similar constraints for English epistemic must have been observed by many authors11, and are, in fact, apparent from the translation of examples (5.a,b,c).

The above restriction on epistemic readings does not apply at all to the conditional form dovrebbe, as shown by (4.a) above. In fact, most of the occurrences of “epistemic” dovrebbe in our sample from the Sole 24 Ore Daily have future reference, many of them being predictions of how the economic situation will develop.

In contrast with the picture drawn above, Squartini (2004) argues that occurrences of deve like (6.a) and (6.b) have a future event as their argument and are indeed epistemic.

(6.a) Il cielo è pieno di nuvole. Deve piovere. ‘The sky is full of clouds. It’s going to rain’ (6.b) George Bush deve incontrarsi con il Primo Ministro israeliano Ehud Olmert

il 22 di Maggio. ‘Georges Bush is to meet Israeli prime minister Ehud Olmert on May 22’

One can show, however, that future reference examples like (6.a) and (6.b) differ from typical epistemic-inferential uses of deve in some important respects and there are quite strong grounds to reject Squartini’s hypothesis. It is worth paying special attention to these differences between epistemic and future reference dovere, because they will turn out to be one of the keys to understand the functioning of the conditional dovrebbe in predictions.

4. Inferential and quotative evidentiality

Much like the English epistemic must, the epistemic use of deve is deictically anchored to the moment of utterance and refers to the on-line inferences of the speaker (except in free indirect speech). It can be reconstructed as having a doxastic modal base, referring to beliefs meta-represented qua beliefs12. Epistemic deve functions as a marker of inferential evidentiality as it signals 10 The star refers only to the impossibility of an epistemic reading. A deontic

reading of (5.c) is perfectly possible both in Italian and in English. 11 Werner (2005) proposes an explanatory theory for this constraint in English. 12 On the role of metarepresentation in epistemic/doxastic readings see note 2

above.

that the speaker does not have direct access to the propositional content and infers it from contextual data. As such it is incompatible with knowledge from testimony and typically creates difficulties in the context of reportative expressions such as (7):

(7) Deve aver piovuto qui stamattina. ?Me l’ha detto Giovanni. ‘It must have rained here this morning. ?Giovanni told me that.’

Although it is still marginally possible, here, to interpret (7) in a ‘shifted context’ as a kind of free indirect speech reporting someone else’s inference, it is, in any case, not possible to take deve as referring to the degree of certainty of the speaker towards a proposition whose knowledge derives from testimony. Deve with future reference, on the contrary, does not create any difficulty when it is used in combination with reports:

(8.a) Deve piovere. Lo ha detto la meteo. ‘It’s going to rain. The weather forecast told that’ (8.b) Bush deve incontrare Olmert il 22 maggio. Lo ha comunicato un portavoce

della Casa Bianca ‘Bush is to meet Israeli prime minister Olmert on May 22. A White House

spokesman announced.’

According to Squartini (2004) – who considers, in particular, examples similar to (8.b), taken from a journalistic corpus– there are two distinct epistemic-evidential values of deve: inferential and reportative, which would explain the differences between (7) and (8). The same reportative evidential use would be found in the conditional form dovrebbe.

However, there are some strong arguments against this explanation. Future reference deve tolerates embedding in a number of epistemic environments where both inferential deve and the reportive evidential use of the conditional mood cannot 'survive', such as, for instance, the subjective epistemic adverb forse ('maybe'):

(9.a) Forse Bush deve incontrare Olmert il 22 maggio 'Maybe Bush is to meet Olmert on May 22' (9.b) ?? Forse Bush deve aver incontrato Olmert. ?? Maybe Bush must have met Olmert on May 22' (9.c) *Forse Bush incontrerebbe Olmert il 22 maggio (reportive reading of the

conditional) 'Maybe Bush is reportedly to meet Olmert on May 22'

Not only the epistemic adverb forse is incompatible with epistemic deve in (9.b) due to a clash between epistemic possibility and necessity; by situating

the source of belief in a subjective evaluation of the speaker13, forse also enters into conflict with the reportive evidential reading of the conditional mood in (9.c). Future deve, in contrast, can be embedded in the subjective evaluation expressed by forse, as shown by (9.a). Actually, the future uses of deve can be embedded indifferently just in any kind of epistemic structure, including constructions of the type ‘nobody thinks that p’ and ‘I don’t know whether p’.

At this point, then, It is quite natural to conclude that – contrary to what Squartini (2004) hypothesizes – the modality expressed by future-reference deve is neither epistemic nor evidential. The uses in (8.a) and (8.b) can be explained by resorting to non-epistemic modal bases.

In (8.a) we have a restricted-alethic modality (cf. Lycan 1994), or, to use Kratzer’s (1981) terminology a realistic modal base, where the conversational background consists of facts of a certain kind. Here the modal quantifies over possible future developments of a set of circumstances (weather conditions). Such a modality is causal in nature: it concerns circumstances that have physical causal power on the future event related by the modalized proposition.

In (8.b) we have a kind of normative modality, close to deontic modality, where necessity is defined by what is entailed by a schedule, a program, or other similar human arrangements. Kronning (1996 and 2001b) makes a similar hypothesis to explain future reference uses of French devoir. Kronning considers both cases as instances of restricted alethic modality. By doing so, however, he conflates the physical causality of (8.a) with the expectations brought about by the assumed respect of human norms in (8.b). Kronning (2001a) makes another interesting hypothesis as well. He considers the future reference uses of French devoir as the “source” of the apparently epistemic uses of the conditional form (devrait). Here we adopt a similar hypothesis to explain the “epistemic” uses of Italian dovrebbe. We will argue that they can be nicely explained as hypothetical versions of future reference deve (be it causal or schedule based).

We can find evidence for this hypothesis by looking at the restrictions in the type of argumentative-inferential discourse relations inferable with discourse sequences involving dovrebbe.

5. Epistemic deve and dovrebbe and the manifestation of argumentative discourse relations

Consider the following examples:

13 For a discussion of subjectivity as an evidential dimension manifested by a

number of epistemic modal expressions see Nuyts (2001).

(10.a) Giovanni ha lavorato molto. Dev’essere stanco. ‘John worked a lot. He must be tired’ (10.b) Giovanni è stanco. Deve aver lavorato molto. ‘John is tired. He must have worked a lot.’ (10.c) E’ tutto rosso in faccia. Deve essere fuori di sé. ‘He’s all red on his face. He must be out of his mind’

Epistemic deve can be used to manifest inferential relations both co-oriented with the direction of time-causation (inference from cause to effect), as in (10.a), and anti-oriented (inference from effect to cause) as in (10.b). It can also be employed to manifest inferences corresponding to temporal concomitance, like (10.c). Let us compare these sentences with their equivalents containing the conditional dovrebbe in (11):

(11.a) Giovanni ha lavorato molto. Dovrebbe essere stanco. ‘John worked a lot. He should be tired’ (11.b) Giovanni è stanco. *Dovrebbe aver lavorato molto. ‘John is tired. He should have worked a lot’ (11.c) E’ tutto rosso in faccia. *Dovrebbe essere fuori di sé. ‘He’s all red on his face. He should be out of his mind’.

We find that dovrebbe cannot occur in temporally anti-oriented inferences from the effect to the cause (11.a), and is also clearly excluded in certain cases of concomitance like, for instance the symptomatic argument in (11.c). If we go back to the authentic example in (4.b), we find that the impossibility of substitution of deve with dovrebbe stems from the same restriction on inferences from the effect (‘the Italian economic miracle’) to the cause (‘the hidden trigger of the event’).

The diverging behaviour of deve and dovrebbe with respect to argumentative discourse relations can find an explanation in the context of the hypothesis WE have been progressively developing in the previous sections.

Epistemic deve, selecting a meta-representational doxastic conversational background – corresponding to a set of beliefs held by the speaker at the moment of utterance – concerns the properly argumentative level, the form of argumentation, and can convey any kind of deduction (from cause to effect, from effect to cause, and many non-causal schemes): it is sensitive only to the form of the major premise that supports the deduction, and disregards its specific contents.

Dovrebbe, on the other hand, primarily conveys a causal relationship of a natural or deontic14 kind. As a further implicature, the assertion of this relationship may be taken as manifesting the major premise of an argument based on direct causality.

6. Dovrebbe and the Italian conditional mood

We have hypothesized that dovrebbe is a conditional version of the alethic and deontic future oriented readings of indicative deve but we have yet to flesh out in detail what it means exactly to be a hypothetical version of future deve and how this can give rise, in the end, to interpretations that appear “epistemic”.

It is now time to consider the semantics of the morpheme of the conditional mood and the way it contributes to the meaning of dovrebbe.

In this section we formulate the hypothesis that the basic invariant semantics of the Italian conditional mood morpheme is that of a relative necessity modal operator. The modal semantics we propose for the conditional is consistent with the diachronic development of the morpheme from Latin necessity modal constructions and parallels the modal semantics we have developed elsewhere for the Italian future tense (Rocci 2000 and 2005).

Such a semantics is also intended to fit with the idea introduced by Lewis (1973) and further developed by Kratzer (1981) and Frank (1996) that in natural language conditional construction always involve some explicit or implicit modal operator relating the protasis and the apodosis.

The basic semantics of the conditional mood can be captured by the following general formula:

(Def. 3) Conditional (B, ϕ) ⇔ ( B → ϕ) Procedural restriction on the saturation of B: ‘B is identified with some set of non-factual propositions’

Provided that it is not made up of factual propositions, the B of the conditional can be understood in context as comprising different sorts of propositions giving rise to the various readings of this mood. Let us consider three typical cases 15:

14 In this latter case the existence of a commitment/obligation/arrangement creates

the expectation of a subsequent action. 15 For a more extensive discussion on the derivation of the various readings of the

Italian conditional from its basic modal semantics see Rocci (2007b).

(12.a) Conditional sentence: ...un orientamento sul duplice controllo che, se accolto, comporterebbe l' obbligo di consolidare nel gruppo Pirelli anche i conti... (IL SOLE 24 ORE 04/01/2002)

‘A view of double control that, if accepted, would entail an obligation to consolidate in the Pirelli group also the finances...’

� ({the view is accepted, ... }→ there is an obligation to consolidate)

(12.b) Modal subordination in discourse: Siamo in pratica in tre: Edisontel, Atlanet e Albacom. Pur nelle differenti dimensioni penso che sarebbe una buona chance riuscire a fondere queste attività, in parte complementari. Nascerebbe un gruppo di telefonia fissa forte che potrebbe creare valore e dare soddisfazioni agli azionisti. (IL SOLE 24 ORE 12/01/2002)

‘There are in fact three of us: Edisontel, Atlanet and Albacom. Notwithstanding the different size, I think it would be a good opportunity to manage to merge these activities, which are, in part, complementary. A strong telephony group would be born that could create value and satisfy shareholders’

� ({the merger happens, ... }→ a strong telephony group is born)

(12.c) Reportive evidential: Barilla ha ufficialmente il 2,07 di Kamps (anche se secondo fonti finanziarie disporrebbe di una quota superiore) (IL SOLE 24 ORE 17/04/2002)

‘Barilla officially owns 2,07 % of Kamps (but according to financial sources it reportedly owns a larger share)’

� ({‘what financial sources say’}→ Barilla owns a share > 2,07%’)

In the prototypical case of the conditional sentence (12.a) the non-factual conversational background is reconstructed as including the protasis of the conditional construction16, while in the case of (12.b) it is a more or less explicit discourse antecedent which is inserted in the non-factual conversational background, according to a mechanism similar to modal subordination (Roberts 1989). Finally, in the case of the reportive evidential

16 In fact, the conversational background of a natural language conditional

includes the protasis but it is not limited to it. Many studies have pointed out that in order to get realistic truth conditions for the conditional we cannot maintain that the apodosis holds true in all the worlds in which the protasis is true, no matter what other conditions may obtain in that world. For instance, it runs against our intuitions that a sentence like If John gets the tenure, he will be happy entails that John is happy holds true in a world where he has got the tenure and simultaneously contracted a devastating disease. Due to space constraints, here we will completely ignore this important problem in the semantics of conditionals. For a discussion see Kratzer (1981), Mc Cawley (1996) and Moeschler & Reboul (2001).

use in (12.c) the non-factual conversational background is identified with what the quoted or anonymous sources say to be the case.

What kind of interaction should we envisage, then, between the modal semantics of the conditional and the semantics of dovere?

A relatively simple and descriptively interesting hypothesis is to consider that due to the conditional mood, the set of worlds over which the modal dovere quantifies is further restricted to the subset of worlds where an explicit or implicit antecedent holds. Where no antecedents are available or recoverable from the context, the further restriction is reconstructed, by default, as a “stereotypical background” (what is normal, usual, typical see Kratzer 1981)17.

(Def. 4) dovrebbe (B, ϕ) ⇔ • ({modal base ∪! conditional restriction}→ ϕ)

Where - the modal base is either alethic or deontic; - the conditional restriction is a set of non-factual propositions; - the sign ‘∪!’ indicates the set-theoretic operation of compatibility-

restricted union18.

5. Epistemic dovrebbe and the structure of argumentative discourse

The final part of this paper is devoted to applying the proposed semantics of the conditional modal dovrebbe to an extended analysis of an authentic example from our corpus of economic-financial news in order to show how such an analysis can help to understand the contribution of this modal to the structure of argumentative discourse. Let us examine example (13) below in detail:

(13) [...] secondo stime autorevoli, a fronte di consumi per 77,5 milioni di barili/giorno (mbg), l' offerta ora è di 79,9 mbg, con uno sbilancio che, in condizioni normali, dovrebbe far precipitare le quotazioni. (IL SOLE 24 ORE 05/12/2002)

17 An apparent drawback of this hypothesis, at least according to a certain view of

the relationship between linguistic signs and semantic structures, is that the semantics of dovrebbe is not compositionally derivable from the semantics of dovere and the semantics of the conditional morpheme. Rather, we can see the conditional modal dovrebbe as a motivated construction (cf. Goldberg 1995) where some semantic features are expressed redundantly by the mood morpheme and the modal verb, while others are clearly inherited from one or the other component.

18 On this set theoretic operation see Frank (1996).

‘According to authoritative estimates, with a consumption of 77.5 million barrel per day (MBD), the offer is now of 79.9 MBD with an unbalance that, under normal conditions. ought/would/should make the price fall’

Here the modal base of dovrebbe is contextually identified to the current situation of the economy, a set of factual propositions, which notably includes the following two propositions, presented in the immediately preceding co-text:

(a) Oil consumption is 77.5 MBD (b) Oil offer is now of 79.9 MBD

The conditional restriction of the modal is explicitly identified with the set of normal conditions by the phrase in condizioni normali (“under normal conditions”). Note that the mentioning of the evidential sources (secondo stime autorevoli “According to authoritative estimates”) does not enter into the conditional restriction, because it does not directly concern the causal relationship established by the modal, but only the evidential status of the two propositions (a) and (b). Thus, the predicate-argument structure of the modal can be reconstructed as follows:

• ({ situation of the economy: ‘Oil consumption is 77.5 MBD’, ‘Oil offer is now of 79.9 MBD’,...} ∪! {normal conditions}→ ‘Oil price falls’)

It can be observed that in this example economic causality is treated much like physical causality – the human agent involved is not “visible”. This is in accordance with the ontology assumed by classical models of market economy, which largely see market forces through the metaphor of physical forces.

The fall of oil price is a logical consequence of the compatibility restricted union of the propositions in the modal base and the set of propositions making up the normal conditions'. This means that the modal base alone might not be enough to license p as a consequence. So, the causal relation described by the modal can be taken as an argument for concluding that the embedded state of affairs will obtain in w0, to the extent that normal conditions do indeed apply in w0 at that time. By definition, normal conditions apply to most of the worlds most of the times, hence the frequent inference of a probability reading from the use of conditional dovrebbe. The outline of the resulting argument from the cause to the effect according to the laws of economics can be traced as follows:

The inference from the normal conditions to the probability meaning of dovrebbe can be blocked if the context provides more specific evidence that the normal conditions in the conditional restriction do not apply in the present situation. This is precisely the case in example (13) where, if we examine the broader co-text we find that in the specific situation normal conditions may well not apply in w0. In fact, in the preceding text, the article, written shortly before the US invasion of Irak, mentions the uncertainty and tension surrounding the work of UN weapons inspectors in Irak and concomitant strikes in Venezuela. Only at this point the author introduces the discussion of oil surplus with (14):

(14) Se però dimentichiamo i rischi “politici”... 'However, if we forget about “political” risks...'

So, the necessary consequence manifested by dovrebbe applies in a ‘normal scenario’ where we ignore political risks and we reason from a purely economical viewpoint.

But risks are not facts. Interestingly, the noun risk itself hides a modal component. So, the simplified scenario based on purely economical

reasoning might still apply. In the end it becomes quite difficult to associate the use of dovrebbe in (13) to a definite degree of certainty of the writer with respect to the modalized proposition.

***

If it is sometimes difficult to place dovrebbe in an epistemic scale of certainty it is because it does not convey directly an epistemic evaluation of probability, but a conditional necessity, and the epistemic evaluation of the conditions themselves may vary.

It is not unusual for economic-financial news texts to outline different scenarios without betting on any of them, leaving to the readers the task of weighing arguments and counterarguments. This, together with the attribution to sources, allows the authors to dilute their argumentative responsibility, and, at the same time, present their discourse as informative rather than persuasive.

References

Bertinetto, P.M. (1979). Alcune ipotesi sul nostro futuro (con osservazioni su potere e dovere), Rivista di Grammatica Generativa, 4, 1-2: 77-138

Bertinetto, P.M. (1986). Tempo, aspetto e azione nel verbo italiano, Florence, Accademia della Crusca

Del Lungo Camiciotti, G. 1998. Financial news articles and financial information letters: a comparison, in: Forms of argumentative discourse. Per un'analisi linguistica dell'argomentare, ed. Bondi, Marina, 195-205. Bologna:CLUEB.

Dendale, P. (1999). ‘Devoir’ au conditionnel : valeur évidentio-modale et origine du conditionnel, Cahiers Chronos, 4, p.7-28.

Goldberg, A. (1995). Constructions: A Construction Grammar Approach to Argument Structure, University of Chicago Press: Chicago

Halliday, M.A.K. (1994). Functional Grammar, Second Edition. London: Arnold

Horn, L.R. (1972). On the semantic properties of logical operators in English, UCLA dissertation.

Kaufmann, S. ; Condoravdi, C. ; Harizanov, V. (2006). Formal approaches to modality, in: W. Frawley (ed.) The Expression of Modality. Berlin / New York: Mouton de Gruyter

Kratzer, A. (1981), The Notional Category of Modality, in: H.J. Eikmeyer ; H. Rieser, (eds), Words, Worlds and Contexts, De Gruyter, Berlin

Kronning, H. (1996). Modalité, cognition et polysémie : sémantique du verbe modal devoir, Uppsala : Almqvist & Wiksell

Kronning, H. (2001a). Nécessité et hypothèse : ‘devoir’ non-déontique au conditionnel, in: P. Dendale ; L. Tasmowski (eds.), Le conditionnel en français. « Recherches Linguistiques », 25, Metz, Université de Metz

Kronning, H. (2001b). Pour une tripartition des emplois du modal devoir, in : P., Dendale; J. van der Auwera (eds.) Les Verbes modaux; Cahiers Chronos, Amsterdam: Rodopi.

Lewis, D. (1973). Counterfactuals, Oxford:Blackwell Lycan, W.G. (1994). Modality and meaning, Dordrecht ; Boston etc.:

Kluwer Academic Publ. Merlini, L. (1983). Gli atti del discorso economico:la previsione.

Status illocutorio e modelli linguistici nel testo inglese . Parma: Edizioni Zara - Università di Parma.

Mc Cawley, J.D. (1996). Conversational scorekeeping and the interpretation of conditional sentences, in: M. Shibatani ; S. Thompson (eds.), Grammatical Constructions: Their Form and Meaning, Oxford: Clarendon Press, 77-101.

Moeschler, J. ; Reboul, A. (2001). Conditionnel et assertion conditionnelle, in: Dendale, P. ; L. Tasmowski (eds.) Le conditionnel en français. Recherches linguistiques, 25

Nuyts, J. (2001). Subjectivity as an Evidential Dimension in Epistemic Modal Expressions, Journal of Pragmatics 33:383-400

Papafragou, A. (2000). Modality: Issues in the Semantics-Pragmatics interface, Amsterdam: Elsevier

Pietrandrea, P. (2005). Epistemic Modality. Functional Properties and the Italian System, Amsterdam/Philadelphia: Benjamins

Roberts, C. (1989). Modal Subordination and Pronominal Anaphora in Discourse, Linguistics and Philosophy, 12, 683-721

Rocci, A. (2000). L'interprétation épistémique du futur en italien et en français: une analyse procédurale, in: J. Moeschler (ed.) Inférences directionnelles, représentations mentales et subjectivité, Cahiers de Linguistique Française, 22: 241-274

Rocci, A. (2005a). La modalità epistemica tra semantica e argomentazione, Milano: I.S.U Università cattolica.

Rocci, A. (2005b). Epistemic Readings of Modal Verbs in Italian: the relationship between propositionality, theme-rheme articulation an inferential discourse relations, in: B. Hollebrandse, A. van Hout and C. Vet (eds.) Crosslinguistic Views on Tense, Aspect and Modality. Amsterdam/New York: Rodopi (Cahiers Chronos 13): 229-246

Rocci, A. (2007a). Epistemic modality and questions in dialogue. The case of the Italian interrogative constructions in the subjunctive mood and the ‘epistemic’ future tense. in: L. de Saussure, J. Moeschler ; G. Puskas (eds.). Tense, Mood and Aspect. Theoretical and Descriptive Issues. Cahiers Chronos 17. Amsterdam: Rodopi

Rocci, A. (2007b). Le modal italien dovere au conditionnel: évidentialité et contraintes sur l’inférence des relations de discours argumentatives. Travaux Neuchâtelois de Linguistique (TRANEL), 2006/45: 71-98

Rocci, A. (2007c). Modality and its conversational backgrounds in the reconstruction of argumentation, in: F.H. van Eemeren, J.A. Blair, Ch.A. Willard, B. Garssen (eds.) Proceedings of the Sixth Conference of the International Society for the Study of Argumentation, Amsterdam: SicSat:1185-1194

Snoek-Henkemans, F. (1992). Analysing complex argumentation. Amsterdam: SicSat

Squartini, M. (2004). Disentangling evidentiality and epistemic modality in Romance, Lingua 114: 873-895

Toulmin, S.E. (1958), The uses of argument. Cambridge: Cambridge UP

Walsh, P. (2001). Prediction and Conviction: Modality in Articles from The Economist; in: M. Gotti ; M. Dossena (eds), Modality in Specialized Texts, Bern, Switzerland: Peter Lang, 361-378.

Walsh, P. (2004). Investigating prediction in financial and business news articles, in: R. Facchinetti ; F.R. Palmer (eds), English modality in perspective : genre analysis and contrastive studies, Frankfurt am Main: P. Lang, 83-100

Walsh, P. (2006). Playing Safe? A Closer Look at Hedging, Conditions and Attribution in Economic Forecasting, in: eds. Bhatia& M. Gotti (eds), Explorations in specialized genres, Bern etc: P. Lang, 135-154.

Walton, D.N. (1996). Argumentation schemes for presumptive reasoning, Mahwah N.J.: L. Erlbaum.

Werner, T. (2005). The temporal interpretation of some modal sentences in English (involving a future /epistemic alternation), in: B. Hollebrandse, A. van Hout and C. Vet (eds), Crosslinguistic Views on Tense, Aspect and Modality. Amsterdam/New York: Rodopi (Cahiers Chronos 13), 247-259