A Context-Sensitive and Functional Approach to Evidentiality in Spanish or Why Evidentiality Needs a Superordinate Category 3631626363, 9783631626368

This study adopts a context-sensitive and functional approach to the qualitative analysis of Spanish linguistic means th

981 102 4MB

English Pages 462 [465] Year 2013

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

A Context-Sensitive and Functional Approach to Evidentiality in Spanish or Why Evidentiality Needs a Superordinate Category
 3631626363, 9783631626368

Citation preview

A Context-sensitive and Functional Approach to Evidentiality in Spanish or Why Evidentiality needs a Superordinate Category

Potsdam Linguistic Investigations Potsdamer Linguistische Untersuchungen Recherches Linguistiques à Potsdam Edited by/Herausgegeben von/Edité par Peter Kosta, Gerda Haßler, Teodora Radeva-Bork, Lilia Schürcks and/und/et Nadine Thielemann

Vol./Bd. 10

Anja Hennemann

A Context-sensitive and Functional Approach to Evidentiality in Spanish or Why Evidentiality needs a Superordinate Category

Bibliographic Information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data is available in the internet at http://dnb.d-nb.de. Zugl.: Potsdam, Univ., Diss., 2012 Cover Design: © Olaf Gloeckler, Atelier Platen, Friedberg Library of Congress Cataloging-in-Publication Data Hennemann, Anja, 1983A context-sensitive and functional approach to evidentiality in Spanish or why evidentiality needs a superordinate category / Anja Hennemann. pages cm. — (Potsdam linguistic investigations ; 10) Includes bibliographical references. ISBN 978-3-631-62636-8 1. Spanish language—Semantics. 2. Spanish language— Modality. I. Title. PC4385.H46 2013 4601.1'45—dc23 2013006369 517 ISSN 1862-524X ISBN 978-3-653-02066-3 (E-Book) DOI 10.3726/978-3-653-02066-3 ISBN 978-3-631-62636-8 (Print) © Peter Lang GmbH Internationaler Verlag der Wissenschaften Frankfurt am Main 2013 All rights reserved. Peter Lang Edition is an Imprint of Peter Lang GmbH All parts of this publication are protected by copyright. Any utilisation outside the strict limits of the copyright law, without the permission of the publisher, is forbidden and liable to prosecution. This applies in particular to reproductions, translations, microfilming, and storage and processing in electronic retrieval systems. www.peterlang.de

Editorial The series Potsdam Linguistic Investigations – Potsdamer linguistische Untersuchungen – Recherches linguistiques à Potsdam presents cutting-edge fundamental linguistics research carried out at the University of Potsdam. Its major goal is to publish collection of articles, conference proceedings and monographs on contemporary issues in the fields of Slavic languages and literature, Romance studies, English and American studies, German studies and general linguistics. A special focus of study is the formal, functional and cognitive description of language. The following areas of linguistics will seek to develop their own profile: phonology, morphology, syntax (with special attention to generative syntax), semantics, pragmatics (discourse analysis, speech act theory), sociolinguistics and language contact. We do not set any theoretical, methodological or geographical boundaries. The series will serve greatly as a forum for young scholars as well as other researchers working in various linguistic fields and frameworks in Potsdam or elsewhere. The indication of Potsdam stands for the crucial importance and outstanding quality of linguistics research at the University of Potsdam. On the other hand, researchers from other Universities with proven excellence of their work are most welcome to publish their doctoral dissertations, habilitation monographs or conference proceedings in this series. The languages of publication are German, English and French.

Editorial Die Reihe Potsdam Linguistic Investigations – Potsdamer linguistische Untersuchungen – Recherches linguistiques à Potsdam ist eine Plattform für linguistische Forschungen an der Universität Potsdam. Sie publiziert Sammelbände und Monographien zu aktuellen Fragen der zeitgenössischen internationalen Linguistik aus den Disziplinen Slavistik, Romanistik, Anglistik/Amerikanistik, Germanistik und Allgemeine Linguistik. Ein besonderer Schwerpunkt liegt in der formalen, funktionalen und kognitiven Sprachbeschreibung. Darin bilden vor allem die Bereiche Phonologie, Morphologie, Syntax (unter besonderer Berücksichtigung der generativen Syntax), Semantik, Pragmatik (Diskursanalyse, Sprechhandlungstheorie, Geschlechterforschung), Soziolinguistik und Sprachkontakt ihre eigenen Profile. Wir wollen keine theoretischen, methodischen oder lokalen Grenzen setzen. Deshalb richtet sich die Reihe sowohl an Nachwuchswissenschaftler als auch an Kollegen in Potsdam und außerhalb Potsdams, die in verschiedenen Richtungen, Modellen und theoretischen Ansätzen der modernen Linguistik arbeiten. Der Hinweis auf den Standort Potsdam soll zum einen die herausragende Bedeutung der linguistischen Forschung an dieser Universität signalisieren. Andererseits bedeutet die Nennung nicht, dass ausschließlich Forschungsergebnisse (einschließlich Dissertationen, Habilitationen und Konferenzsammelbände) veröffentlicht werden, die von Linguistinnen und Linguisten an der Universität Potsdam stammen. Die drei Publikationssprachen sind Deutsch, Englisch und Französisch.

Editorial La serie « Potsdam Linguistic Investigations – Potsdamer linguistische Untersuchungen – Recherches linguistiques à Potsdam » représente une plate-forme d’études linguistiques à l’université de Potsdam. Elle publie des recueils et des monographies sur les questions actuelles de la linguistique contemporaine internationale dans les domaines des études des langues slaves et romanes, anglaise et américaine, des langues germaniques et de la linguistique générale. Un point principal de recherche est posé sur la description formelle, fonctionnelle et cognitive des ces langues. Dans ces domaines, on met l’accent sur les profils de la phonologie, morphologie, syntaxe (en tenant compte de la syntaxe générative), sémantique, pragmatique (l’analyse du discours, la théorie des actes de la parole, la recherche sur le genre), la sociolinguistique où la linguistique de contact. Nous ne voulons pas poser des limites dans la théorie, la méthode et le lieu de recherche. C’est pourquoi la série invite les jeunes chercheurs ainsi que les collègues de Potsdam et des autres universités qui travaillent dans les secteurs de la linguistique moderne. Le titre de la série veut démontrer d’un coté l’excellente qualité de la recherche linguistique à Potsdam sans toutefois exclure les autres. Cela veut dire que nous acceptons et nous invitons les linguistes de Potsdam et de l’extérieur (inclus les thèses de doctorat et d’habilitation et les actes de colloques). Les trois langues de publication sont : l’allemand, l’anglais et le français.

7

Acknowledgements I would like to express my deep gratitude to my supervisor Prof. Dr. Gerda Haßler. I appreciate her expertise, her helpfulness and especially her sometimes severe but helpful comments she gave me during the whole process of the PhD writing. I also owe her my first academic teaching as well as my first conference papers. So many, many thanks! Furthermore, my research has been influenced by the work of Prof. Dr. Bert Cornillie. I have benefited greatly from his insightful comments and criticisms. However, I would not only like to thank him for his scientific support but also for his moral one. My gratitude also goes to Dr. Kasper Boye, who was always willing to answer my questions during my research and who was disposed to have discussions with me, which turned out to be fruitful. During the course of this work I received a great deal of support from many other people. I would like to thank my parents to whose home I sometimes invited myself for lunch or dinner in order to have a reason for leaving the desk. I must also acknowledge my friends, especially Katharina Stephan and Theresa Schulz, ‘my Spanish family’ and Marcus Neuß: some of them supported me predominantly morally, others ‘semi-scientifically’ and again others were nearly always disposed to simply distract and entertain me. They all have helped me on my way through this PhD. In summary, I am thankful to be surrounded by the people I am surrounded by.

9

Contents 1

Introduction

13

2

The notion of evidentiality in extenso

24

2.1

Where evidentiality and other linguistic categories meet

30

2.1.1

Where evidentiality and epistemic modality meet

31

2.1.2

Where evidentiality and deixis meet

38

2.1.2.1

On the deictic reading of modal verbs

53

2.1.2.2

Where to put inference?

63

2.1.3

Where evidentiality and (inter)subjectivity meet

77

2.1.4

Where evidentiality and polyphony meet

87

2.2

Concluding thoughts or proposal for solution:

105

speaker’s perspectivisation as superordinate category 3

What has been done so far concerning evidentiality in Spanish, other Romance languages and recently in general

128

3.1

Evidentiality and the grammaticalisation of German evidential expressions

129

3.2

Evidentiality as a kind of epistemic relativisation – exemplified by Spanish

131

3.3

Evidentiality and the processes of subjectification and modalisation

133

3.4

Evidential adverbs and their grammatical/lexical status

136

3.5

Evidentiality between lexicon and grammar in various languages

137

3.6

The relation between evidentiality and epistemic modality in Spanish

150

3.7

Evidentials and modal meanings in Guaraní

153

3.8

Evidentiality and its grammatical/lexical realisations in various languages

155

3.9

Evidentiality and modal certainty in English adverbs

158

10 3.10

Evidentiality and epistemicity in English verbs of cognitive attitude

163

3.11

Evidentiality and deixis in English and German verbs of perception

167

3.12

Evidential expressions and their scope

169

3.13

Evidentiality, epistemic modality and grammaticalisation in Italian

173

3.14

Evidentiality, modality, epistemicity and grammaticalisation in German

182

3.15

Evidentiality, modality and subjectivity

186

3.16

Summarising and concluding thoughts

195

4

Theoretical and methodological background of the present study

198

4.1

The research process

199

4.1.1

The pre-research phase – part I: detective novels

199

4.1.2

The pre-research phase – part II: online-news articles

202

4.1.3

The research phase

210

4.2

The syntactic position of evidentemente & Co.

213

4.3

The notion of procedural meaning

215

4.4

The notion of utterance

222

4.5

The pragmatic background of the present study

224

4.5.1

The Cooperative Principle – Grice’s maxims of conversation

226

4.5.2

Relevance Theory – how relevant is it?

228

4.5.3

The present study’s approach to evidentiality in Spanish

235

5

A context-sensitive and functional approach to the use of verbs of cognitive attitude and modal adverbs in Spanish

240

5.1

The evidential use of Spanish modal adverbs

244

5.1.1

aparentemente

248

5.1.2

evidentemente

253

11 5.1.3

obviamente

257

5.1.4

posiblemente

261

5.1.5

probablemente

266

5.1.6

seguramente

270

5.1.7

supuestamente

276

5.2

The evidential use of Spanish verbs of cognitive attitude

278

5.2.1

creer

279

5.2.2

pensar

282

5.2.3

Excursus: on other pragmatic functions of pensar

284

5.2.4

saber

303

5.2.5

suponer

307

5.3

Summarising and concluding thoughts concerning the verbs and adverbs

309

6

A context-sensitive and functional approach to the use of the Spanish modal verbs poder and deber

314

6.1

The evidential use of deber

318

6.2

The evidential use of poder

330

6.3

Summarising and concluding thoughts concerning the modals

341

7

A context-sensitive and functional approach to the use of the conditional form and the synthetic future

344

7.1

The evidential use of the conditional form

366

7.2

The evidential use of the synthetic future

381

7.3

Summarising and concluding thoughts concerning the conditional form and the synthetic future

399

8

Overall evaluation of the results and future prospects

404

8.1

Approach and data

404

8.2

Evidentiality and its related categories

409

8.3

The consequence: the notion of speaker’s perspectivisation

418

12

9

8.4

Spanish linguistic means that serve as evidential expressions

420

8.5

Future prospects

427

References

437

13

1 Introduction Evidentiality is not a linguistic category grammatically inherent in the language system of Contemporary Spanish as it does not have real evidentials like the languages Kashaya (Oswalt 1986), Fasu (Foley 1986: 165) or Quechua Wanka (Floyd 1997), for instance.1 Nevertheless, it can be expressed by lexical and grammatical2 forms. Famous studies like the ones by Boas (1947), Barnes (1984), Willett (1988), Aikhenvald (2003a, 2004) etc., in which evidentials are described from a semasiological perspective, were the basis for research in languages that do not possess real evidentials. That means, once having determined the meanings of evidentials, one can search for linguistic devices showing the same or similar function(s) (cf. Volkmann 2005: 84). The following examples show linguistic devices encoding evidentiality, while (1), (2) and (3) contain a lexical form (a modal adverb, a modal verb and a verb of cognitive attitude, respectively), (4) and (5) contain a grammatical one (conditional and synthetic future): (1) Una adolescente de 15 años originaria de Nigeria es la séptima menor liberada de una red de prostitución en Cataluña en el último mes. Tres proxenetas de Ghana y Nigeria explotaban, supuestamente, a la víctima, según informó ayer el Cuerpo Nacional de Policía (CNP). La joven ha pasado a disposición de la Fiscalía de Menores y, previsiblemente, acabará bajo la tutela de la Generalitat.3 (El País 17/07/2010) (2) “No era un puerto tan duro para soltarlo, se podía ir cómodo a rueda”, dijo el español “Contador debe de estar frustrado, no ha podido ponerse líder”, opinó el luxemburgués. Ganó el francés Riblon, otro hombre de velódromo disfrutando de las montañas ¡Sacrilegio!, ¡sacrilegio!, volverían a gritar al ver a dos pipiolos [...] (El País 21/07/2010) (3) [...] da pena ver cómo, por ejemplo, Inglaterra sale del Mundial con un gol legal no concedido o cómo después de expulsar a Zidane hace cuatro años a través del visionado de la jugada en vídeo no se hace lo propio con De Jong tras cargarse a Xabi Alonso. Pienso que ayudaría en jugadas determinantes y la polémica seguiría

1

2 3

This is thus another study from a scholar of a Romance language who regards “having a way of saying ‘apparently’ or ‘I do not believe’ […] a good enough pretext to put ‘evidentiality’ in the title of [her] paper” (Aikhenvald 2004: 5). ‘Grammatical’ is here used in the sense of belonging to grammar “as opposed to, e.g., what belongs to the lexicon” (Lehmann 2002: 8). In the present study, the linguistic device that is the subject under discussion will always be highlighted in bold. Contextually provided information which helps to determine the use of the linguistic means will be underlined.

14 dando vida a los forofos, porque fútbol y polémica siempre irán de la mano, con tecnología o sin ella. (El País 21/07/2010) (4) Sin embargo, la Policía ha asegurado que la situación “está bajo control” y que otros paquetes sospechosos recibidos en las embajadas de Dinamarca y Venezuela son “una falsa alarma”. El sobre recibido en la sede diplomática de Venezuela sería tan sólo una felicitación navideña. (El Mundo 29/12/2010) (5) “[...] llamaron a las tres de la mañana dos periodistas que, teóricamente, tenían hotel en Nantes: ‘Nin, estamos tirados en Mónaco; búscanos una cama y una ducha’, me dijeron”. Media hora después tenían cama, ducha y una botella de champán como bienvenida. Será por eso que asegura: “Lo más divertido de mi trabajo es superar retos”. No hay uno que se le resista. (El País 05/07/2010)

Example (6) represents a further subcategory of evidentiality, viz. (direct) quotation, which will only marginally be dealt with in the context of the present study: (6) “Los padres pueden mandar a los niños al colegio con una manzana y que luego ellos se compren en el cole una palmera de chocolate, una bolsa de patatas, un refresco... Lo que hay que hacer es educar al niño para que tenga conciencia de lo que come”, dice Susana Monereo, jefa de Endocrinología del Hospital de Getafe, quien pone un ejemplo del problema: “Si no, no comerán golosinas en el colegio, algo que está muy bien, pero estarán deseando salir para comerlas. [...]” (El País 24/07/2010)

As examples (1)-(6) indicate, this highly context-sensitive study will concentrate on data from daily newspapers because newspapers, such as El País and El Mundo, combine direct quotes that are oral in character as well as the journalist’s written consideration of a certain state of affairs ([p]4). The data are consequently retrieved from journalistic discourse which is said to be characterised by succinct transmission of information (cf. Cappon 2005: 11) which, in turn, is said to be enriched and made more ‘alive’ by quotes (cf. Cappon 2005: 121).5 So the newspaper writing style can be divided into two text types: the journalist’s part represents written discourse, while direct quoted speech can be oral in char4 5

In the present study ‘[p]’ stands for ‘state of affairs’ as well as for ‘proposition’ (cf. chapter 3.12 for Boye’s (2010b) contrary line of argumentation). Haßler (2003) describes quotes – or at least the reference to statements of other speakers – as one peculiarity of newspaper writing style: Gegenüber anderen Verwendungsweisen der Sprache weist die Nachrichtensprache eine kommunikative Eigenart auf, die man als durchgängige Mehrstufigkeit der Referenzebenen bezeichnen könnte. Das Referieren auf Aussagen anderer, die Metaaussage ist die Aussageform der Nachricht schlechthin. Der Produzent der Nachricht macht eine Aussage über die Aussage von Offiziellen und Gewährspersonen (Haßler 2003: 116).

15 acter, even though they clearly cannot be compared with ‘real’ spoken data. The third text type that is included in this study also represents written discourse but is characterised not only by orality but also colloquialism: a few examples were found in the chat forums of El País and El Mundo, where readers comment on newspaper articles and discuss certain issues. In these chat forums, most users write in the same way as they speak so that this text type is considered even ‘more oral’ than direct quoted speech. Nevertheless, intonation and paralinguistic devices cannot be considered here due to the text type(s) that is worked with. In summary, the text type(s) considered here are characterised by ‘medial writtenness’ and ‘conceptual orality’ (cf. Koch/Oesterreicher 1994: 587). Insofar as the utterances6 that are the subject under discussion are not uttered by a journalist, in which medial writtenness combines with conceptual writtenness, the utterances are marked by conceptual orality. If a formerly interviewed person is interviewed and then quoted, the direct quoted speech is oral in character and if in chat forums the users write in the same way as they speak, this text type is considered even ‘more oral’. Hence, in these two cases, medial writtenness combines with conceptual orality. As with regard to the conception of an utterance, the terms ‘oral’ and ‘written’ denote the endpoints of a continuum (cf. Koch/Oesterreicher 1994: 587), the speech in chat forums can be (conceptually) considered ‘more oral’ than direct quoted speech. Because of the context-sensitiveness, I will have to go beyond the sentence as “logical […] relations which are expressed within the sentence in one case may indeed be expressed between sentences in other cases” (Simon-Vandenbergen/Aijmer 2007: 82). Linguists like Haßler (1997, 1998, 2001, 2002, 2003, 2004, 2010a, 2010d), Volkmann (1997, 2005), Große (2011), Cornillie (2007a, 2007b, 2009, 2010a, 2010b), Cappelli (2007), Simon-Vandenbergen/Aijmer (2007), Wachtmeister Bermúdez (2004, 2005), Squartini (2001, 2004), Dendale/Tasmowski (1994) and Reyes (1996, 2002), who treat evidentiality in Spanish from an onomasiological perspective, understand this linguistic domain as a semantic-functional category. Semantically, the linguistic device under discussion must indicate the source for (a) certain proposition(s). In other words, the linguistic device must convey the meaning ‘source of information’. Evidentiality is also a functional category as it is expressed by linguistic means that fulfil the function of indicating the source of information for the transmitted content of (a) certain proposi6

For an explanation of the reasons why I treat the statements analysed in the present study as utterances, see chapter 4.4. However, I will deal with sentences, if I speak, for instance, about the fact that a sentence adverb has scope over a sentence or if I distinguish between modus and dictum at the sentence level (cf. chapter 4.2).

16 tion(s). And in order to capture that “evidential meanings range from lexical to grammatical functions” it should be spoken of a semantic-functional domain (Diewald/Smirnova 2010b: 1). This semantic-functional understanding of evidentiality is necessary if dealing with evidentiality in Spanish because the starting point for figuring out evidential meanings in a language that does not possess real evidentials is the function rather than the form. Nevertheless, in this study the semasiological and the onomasiological approach are combined as for certain linguistic expressions (in Spanish itself or other Romance languages) an evidential use was already shown, and it is one aim of this study to underpin those uses/functions for certain forms by means of language data7. This study is partly semasiological in character as the language data illustrate that the linguistic expressions under discussion may have different meanings, or rather may serve different functions, which depends on the context. Evidentiality is expressed by any (linguistic) form serving the function of indicating the source of information for the transmitted information. Thus any linguistic device – be it a grammatical form or a lexical expression – that encodes the source of information is treated as an evidential expression, evidentially used expression or rather an expression with evidential meaning. They will not be called evidentials as the term is applied to obligatorily used morphemes/affixes/particles, as for instance in Tariana (7) or in Tuyuca8 (8): (7) Juse iida di-manika-ka. ‘José has played football (we saw it)’ Juse iida di-manika-mahka. ‘José has played football (we heard it)’ Juse iida di-manika-nihka. ‘José has played football (we infer it from visual evidence)’ Juse iida di-manika-sika. ‘José has played football (we assume this on the basis of what we already know)’ Juse iida di-manika-pidaka.

7

8

I call them ‘language data’ because I do not work with (real) speech data from a (real) oral corpus. Even though they are not real speech data, may data can be said to be oral in character since interviewed persons are quoted directly or utterances found in forums are written as the persons would have spoken. That is why I call them language data. Language data comprise language use in what way soever. “Practically every paper on [evidentiality] makes at least some reference to the evidentiality system in the language of the Tuyuca […]” (Haßler 2010d: 95). And so do I.

17 ‘José has played football (we were told)’ (Aikhenvald 2004: 2-3).9 (8) díiga apé-wi ‘He played soccer (I saw him play)’ díiga apé-ti ‘He played soccer (I heard the game and him)’ díiga apé-yi ‘He played soccer (I have seen evidence that he played)’ díiga apé-yigi ‘He played soccer (I obtained the information from someone else)’ díiga apé-hyi ‘He played soccer (It is reasonable to assume that he played)’ (Barnes 1984: 257).

To summarise the part on the notional background, in this study only obligatorily used morphemes attached to the verb, as in the examples quoted above, are termed evidentials. The notion of evidential expression, in contrast, will be used here in the same sense as Diewald/Smirnova do in Evidentiality in German: The term evidential expression is a neutral label (a hypernym) used to denote any kind of linguistic string with evidential meaning in a particular context, regardless of its linguistic structure and degree of grammaticalization10 (Diewald/Smirnova 2010a: 41).

9

In A Grammar of Tariana, from Northwest Amazonia Aikhenvald gives another example to illustrate the use of evidentials. For describing the state of affairs ‘The jaguar killed a man’, each kind of information source requires the use of the corresponding evidential: For instance, in describing an event such as ‘The jaguar killed a man’, use of the visual evidential would imply that the speaker saw this event happening. The nonvisual evidential would be used if the speaker heard the noise of a man fighting the jaguar (or smelt the blood). The reported evidential would be used if someone told the speaker of the event, while the inferred evidential might be employed if the speaker had encountered a jaguar covered with human blood (Aikhenvald 2003b: 4).

10

The use of the term ‘grammaticalised’ or ‘grammaticalisation’ will be avoided with regard to the analysis of the linguistic means considered in the present study. If I use the term ‘grammaticalised’ or ‘grammaticalisation’ in certain text passages, it is used in the sense which is meant by the linguists whose studies are reviewed or referred to. For a detailed overview of the heterogeneously used – and consequently problematic – term ‘grammaticalisation’ and the meaning of grammaticalisation see Lehmann’s Thoughts on grammaticalization (2002: 8-14). Lehmann, in line with Kuryowicz (1965), defines

18 Describing the linguistic realisation of evidentiality in German as well as regularities in grammaticalisation of certain linguistic forms and constructions, Diewald/Smirnova (2010a) reserve the term ‘evidential’ to denote grammaticalised forms in German. In the present study, in contrast, it will be completely avoided using the notion of evidential to describe linguistic means – even though having evidential meaning – in Spanish. It would not be correct to term the synthetic future, for instance, an evidential. Hopper/Traugott define grammaticalization as the change whereby lexical items and constructions come in certain linguistic contexts to serve grammatical functions and, once grammaticalized, continue to develop new grammatical functions (Hopper/Traugott 2003: xv).

So the reason for avoiding using the notion ‘evidential’ concerning the Spanish synthetic future será(n) – to give an example – is that it is often used to express inference. However, it is to challenge whether it is (by now) its primary meaning, which would be the condition to enter the grammatical system of evidentiality (cf. Aikhenvald 2004: 1). While the morphemes in Tariana and Tuyuca are used to express evidential meanings only, the Spanish synthetic future form may be used to do so. Therefore, the term ‘evidentials’ is reserved for the ‘real ones’ in languages like Tariana, and evidential expression, evidentially used expression and expression with evidential meaning represent the notions which are worked with in the present study in order to refer to Spanish expressions or particular uses of Spanish expressions. The present study’s aim is not to show that some linguistic devices in Spanish can be used as an expression with evidential meaning, or rather, that evidentiality represents a meaning aspect which is semantically inherent in some linguistic means. This was already shown for Spanish (semi-)auxiliaries (Cornillie 2007a), adverbs (Haßler 2004, Cornillie 2010a, 2010b, Reyes 1996), future and grammaticalisation as a process “in which something becomes or is made more grammatical” (Lehmann 2002: 8). Kuryowicz explains: Grammaticalization consists in the increase of the range of a morpheme advancing from a lexical to a grammatical or from a less grammatical to a more grammatical status, e.g. from a derivative formant to an inflectional one (1965: 52; quoted in Lehmann 2002: 6). The bibliographical reference of Kuryowicz’s paper is: Kuryowicz, Jerzy (1965): “The evolution of grammatical categories”. In: Diogenes 51, 55-71. Bybee et al. (1994: 4) explain that “[s]ince the […] revival of interest in [grammaticalisation] theory in the early 1970s, two terms – grammaticalization and grammaticization – have been used, usually interchangeably” (cf. also Lehmann 2002: 8-9). Bybee et al. (1994: 4) decided to use the shorter term as it is the “more elegant” notion.

19 conditional forms (Squartini 2001, 2004), verbs of cognitive attitude (Volkmann 2005) and quotation (Reyes 1996, 2002). The present study, on the one hand, seeks to tie in with the research of Squartini (2001, 2004), Haßler (1997, 2002, 2004, 2010a, 2010b, 2010d), Volkmann (2005), Wachtmeister Bermúdez (2004, 2005) and Cornillie (2007a, 2010a, 2010b), who examined the epistemic/evidential use of certain linguistic means in Spanish. On the other hand, it aims to tie in with the studies by Simon-Vandenbergen/Aijmer (2007) and Cappelli (2007), who examined the epistemic/evidential use of English adverbs of certainty and English verbs of cognitive attitude, respectively. In detail, the present study has different goals: 1. It aims to fortify that Contemporary Spanish provides different means to encode epistemic and/or evidential meaning(s) with the help of lexical and grammatical means. In other words, what Dendale/Tasmowski (1994) have claimed for certain linguistic devices in French will be shown here for the Spanish language: Si le français ne dispose pas d’un système d’évidentiels aussi élaboré que le tuyuca, il n’en reste pas moins que les locuteurs français sont amenés dans certaines circonstances à marquer plus ou moins clairement la provenance de leur information. Le français dispose pour ce faire d’une série d’expressions (principalement lexicales, et en moindre mesure morphologiques, voire typographiques) permettant d’exprimer les principales catégories évidentielles. On peut accorder cette fonction à certains adverbes de phrase, tels que apparemment, visiblement (constatation), certainement, sûrement (inférence, supposition), à des constructions impersonnelles telles que il semble que, il paraît que (ouï-dire), aux verbes modaux devoir, pouvoir dans leur acception épistémique, à des verbes pleins tels que voir, entendre, sentir, aux verbes de déclaration, aux prépositions d’après, selon, pour, aux morphèmes du futur conjectural et du conditionnel d’ouï-dire, aux guillemets de citation (Dendale/Tasmowksi 1994: 5).11

2. The study aims to analyse the evidential use of verbs of cognitive attitude and modal adverbs in Spanish and will reveal that the analysis of these linguistic means must be a context-sensitive one as the meaning of a particular item in a particular context depends on contextual factors. So what Cappelli (2007) and Simon-Vandenbergen/Aijmer (2007) have shown for the use of verbs of cognitive attitude and adverbs of certainty still remains open to study for their Spanish equivalents, even though some little studies for the use of modal adverbs in Spanish (Haßler 2004, Cornillie 2010a, 2010b) already contribute to this field of research.

11

In the present study, the part of a quote where I want to put special emphasis on will be underlined.

20 3. The present study will be concerned with the context-sensitive analysis of the synthetic future and the conditional form and their reportive and inferential use. On the one hand, with the help of the work with language data it is to find out whether the Spanish conditional form alone cannot only be used to express inference but also be termed a linguistic device to indicate reported speech, as Wachtmeister Bermúdez illustrates by the following journalistic example: “El presidente renunciaría en las próximas horas” (2004: 7). On the other hand, this study aims to show that it is not always possible to differentiate between a quotative use and an inferential one (cf. Guentchéva 1994). Squartini (2001: 321) has found out that the Spanish synthetic future form is used to express inference, whereas the conditional form is used to express inference as well as it may be used to indicate reported speech. Squartini supposes for the quotative use that [...] the intrusion of the reportive value seems to be a new disrupting factor introducing a non-uniform feature, possibly producing a new development in the Spanish evidential system (Squartini 2001: 322)12,

while Wachtmeister Bermúdez (2004) takes the fact that the condicional is used to express both inference and quoted words for granted: En el uso periodístico el valor evidencial del condicional es doble. En el uso periodístico el condicional señala evidencia indirecta transmitida o mediada. [...] El otro valor del condicional expresa evidencia indirecta inferida [...] (Wachtmeister Bermúdez 2004: 7).

According to Gévaudan (2009: 118), the synthetic future may also be used to indicate reported speech. So both constructions según X + future and según X + conditional are to be found in journalistic discourse. That is why it is one subgoal to have a closer and comparative look at the quotative use of the future and the quotative use of the conditional. Because of the fact that the reportive use of the conditional is a prominent linguistic device in journalistic discourse, I adopt the translated notion of conditionnel journalistique (cf., for instance, Kronning 2002, 2004 or Sullet-Nylander 2006) and predominantly use the term ‘journalistic conditional’ to refer to this use. Nevertheless, it seems reasonable to assume that the journalistic conditional is no longer only a journalistic phenomenon: if a linguistic phenomenon is frequently found in news articles which are consumed on a daily basis by a large readership, it is highly likely that the reportive use of the conditional could also be found in oral speech data. 12

In a paper from 2004, Squartini explains: “In French, Italian and Spanish the Conditional is a consistent temporal (as future in the past), hypothetical and reportive marker” (Squartini 2004: 890).

21 4. Is the use of the Spanish modal verbs poder and deber (more or less) equivalent to the use of French pouvoir and devoir? Dendale (1994) and Tasmowski/Dendale (1994) have found out that […] devoir épistémique signale en premier lieu non pas la qualité épistémique d’une information, mais le recours même à une opération épistémique. L’opération dont il s’agit est une opération de création d’information que l’on considère en général comme une opération d’inférence (Dendale 1994: 37-38),

and that “[…] devoir signale une inférence valide et une conclusion justifiée à partir des éléments disponibles […]” (Tasmowski/Dendale 1994: 43), as well as that [les] énoncés comprenant pouvoir, tout comme ceux comprenant devoir, sont l’indice d’un processus mental qui consiste à créer des prémisses, à en inférer des conclusions et à évaluer ces conclusions pour n’en retenir que partie. De telles conclusions sont toujours subjectives puisque c’est le locuteur qui décide des hypothèses à retenir. Alors qu’avec devoir, le résultat est une sélection unique, pouvoir indique, le plus souvent sur le mode implicite, que le locuteur reste a priori ouvert à d’autres éventualités. Mais dans les deux cas, le locuteur est également source créatrice d’information, et dans ce sens, tant devoir que pouvoir sont des marqueurs d’évidentialité (Tasmowski/Dendale 1994: 55).

Hence, one aim of the present study is to figure out whether the same conclusions can be drawn for the Spanish modal verb equivalents. As French and Spanish belong to the same language family a similarity in the usage of poder/pouvoir and deber/devoir can be expected. A further subgoal with respect to the modals is to compare Pietrandrea’s findings concerning the conditional form of Italian dovere with the Spanish conditional form debería. In comparing the indicative form and the conditional form of dovere, deve and dovrebbe, Pietrandrea concludes that the former is used to express an inference based on objective knowledge and the latter is used to express an inference based on uncertain knowledge. The indicative form is thus said to convey unconditioned necessity and the conditional form conditioned necessity (cf. Pietrandrea 2005: 74-76; cf. also Giacalone Ramat/Topadze 2007: 22), that is, the conditional morpheme indicates “that the truth of the modal that it modifies holds in a world where certain conditions are met” (Pietrandrea 2005: 76). Pietrandrea (2005: 87) furthermore concedes dovrebbe – besides its ‘conditioned inferential’ meaning – a reportive reading, whereby the use of dovrebbe to mark reported information is less due to the semantic potential of the modal verbs but more to the conditional morpheme itself (cf. Squartini’s (2004) contrary line of argumentation concerning French devoir). Example (9) contains a reportive use of Spanish debería (+ según-phrase):

22 (9) “Figuras como el líder del Sinn Féin, Gerry Adams , no estarán en la comisión por su carácter partidista”, señalan fuentes de los mediadores internacionales. El ámbito de actuación de la comisión de verificación internacional debería afectar no solo al alto el fuego sino a la entrega de las armas de ETA, según señaló Brian Currin a EL PAÍS el pasado 11 de septiembre. (El País 22/09/2010)

5. As – at least – in Spanish, evidentiality is (sometimes strictly) related to other linguistic categories such as subjectivity, epistemic modality and deixis, it is one aim of this study to argue for a superordinate category, that is, an encompassing notion that comprises the categories that sometimes overlap. How, for example, could otherwise be stated that epistemic modality and evidentiality ‘sometimes overlap’ or that they have a relationship of ‘inclusion’? There must be a superordinate category that must be able to reflect the complex relationship between the linguistic categories involved. Thus, following Boye (2006, 2010a), a superordinate category will be extended insofar as it should not only encompass epistemic modality and evidentiality but also other linguistic categories. Understanding evidentiality as a semantic-functional category represents the first step – the condition – of dealing with this linguistic domain in a language that does not possess real evidentials. If the process of onomasiologically figuring out which linguistic devices can express evidential meanings has been completed, the semasiological aspect comes in: it is to analyse the different meanings associated with a certain form, e.g. how many meanings does evidentemente have, or rather, how many functions? The overall aim of this study is not only to contribute to the research field of evidentiality in languages that do not possess real evidentials but also to find – leaving aside languages that possess real evidentials – a superordinate category for evidentiality and the categories which are related to it. The study is structured as follows: before creating the theoretical and methodological background and pointing out where the study is theoretically embedded (chapter 4), an overview about the research field of evidentiality as well as an introduction of related concepts like epistemic modality will be given (chapter 2). Chapter 3 is a review chapter, which represents a survey of studies on evidentiality of the last decade. To take the defined goals of this study into account the following lexical expressions and grammatical forms are analysed and, where appropriate, grouped together: in chapter 5, the epistemic/evidential use of Spanish verbs of cognitive attitude and modal adverbs is analysed. Chapter 6 deals with the use of the modal verbs poder and deber. The synthetic future and the conditional form – being two grammatical forms – are grouped together in chapter 7 and their inferential or rather quotative use is analysed. When the condicional is used to indicate quoted words, it represents ‘a very indirect form’ of quotation, whereas speech-

23 accompanying verbs as decir and explicar may mark direct as well as indirect quotation. While the conditional is intensively studied in chapter 7.1, the function of speech-accompanying verbs is only briefly referred to in chapter 2.1.4. Sometimes a side glance at related linguistic phenomena or interesting approaches to topics related to this study seems to be useful. Where it is necessary the chapters include a summarising part, whereas chapter 8 represents an overall conclusion. All the results – despite the examples from the detective novels obtained in the pre-research phase (for further methodological information see chapter 4) – are taken from the Spanish daily newspapers El País and El Mundo. In order to obtain the results (analysed in this study) I used the search engine GlossaNet (http://glossa.fltr.ucl.ac.be/). Being a search engine GlossaNet allows searches in online-publications of newspapers. The user has the option of choosing between more than 80 newspapers in various languages. For the present study, it was decided to configure GlossaNet to search for the chosen items in El País and El Mundo. The search engine allows the user to set various parameters. These queries may consist of only one lexeme, a more complex expression and may even contain graphs. As the user is not able to use asterisks to replace a verbal ending, GlossaNet was configured to search for entire lexical items (see chapter 4.1.3). The results were received by email in text format. Hence, GlossaNet is a search engine that makes the constitution of an own corpus possible. Although in the time span 25/04/2010-27/04/2011 I received a remarkable quantity of results, this study does not aim to be a quantitative one. It is a qualitative study where the data are used to verify/refute the theoretical basis. It is, for example, not my intention to show in how many cases of a certain total amount of uses the Spanish synthetic future form is used to express inference. Working with other data, different text genres and perhaps working over a longer period of time the results would not be congruent. I ‘only’ want to show how certain linguistic devices are used to express evidential values, combined with the goals listed under the points one to five. Before creating the theoretical and methodological background for this study, it seems useful to firstly give an overview about evidentiality as a linguistic notion itself and about evidentiality as a research field in linguistics in general as well as in Spanish or rather Romance linguistics in particular (chapters 2 and 3).

24

2 The notion of evidentiality in extenso The notion of evidentiality to describe linguistic expressions conveying evidential meaning or rather fulfilling evidential functions is not accepted by every linguist who studies a Romance language (cf. also Hennemann 2010). While Dendale/Tasmowski (1994), for instance, who study French regarding its linguistic devices that indicate the information source, accept the notion of évidentialité without reservations, Guentchéva argues against the application of the notions of évidentialité and évidentiels with respect to the French language: Le terme “évidentiel”, un faux ami de l’anglais evidential, évoque l’évidence, c’està-dire la constatation directe. Or ni l’ouï-dire, ni le non-vu, ni l’inférentiel ne peuvent être considérés comme des évidences (Guentchéva 1996: 13).

Guentchéva therefore prefers to use the term “marqueur médiatif” to describe evidential expressions because, according to her, the primary function of evidential markers is to indicate that [p] represents mediated information. The concept of mediatedness consequently represents the basis for her notional choice, whereby it is applicable to grammatical as well as lexical markers: […] le terme “énonciation médiatisée” […] permet d’englober les procédés aussi bien grammaticaux que lexicaux capables de véhiculer la notion de médiation, constituant ainsi un moyen d’expression de la catégorie sémantique de la médiation (Guentchéva 2004: 27).

Hence, the term médiatif is widely used in French linguistics and not only by Guentchéva (cf. Kronning 2002, 2004 or Lazard 1956, 1996 among others). Kronning (2002) describes the médiatif as the grammatical category belonging to the semantic-functional category of médiation épistémique, whereby he explains the latter notion in ‘evidential terms’: Le médiatif est la catégorie grammaticale de la médiation épistémique (l’“évidentialité”) […], catégorie qui dénote le type de source épistémique (ou, plus généralement, la prise de conscience épistémique) de l’information que communique l’énoncé […] – sources épistémiques telles que la perception, l’inférence, l’emprunt à autrui (Kronning 2002: 564).

Kronning’s words, in contrast to Guentchéva, show that the use of the notion of evidentiality does not have to be totally avoided even if the notion of médiation is preferred. According to Guentchéva (1996: 13), the notion of évidentialité mistakenly conveys ‘direct evidence’ (“la constatation directe”). She justifies her viewpoint providing etymological information about French évident and évidence:

25 […] comme on peut le lire dans le Dictionnaire historique de la langue française (1998 : 1349), [“] évident, emprunt au latin evidens, -entis, est formé de e- ( ex-) et de videre ( voir)” et “se dit de ce qui s’impose clairement à l’esprit”; évidence, emprunt savant au latin evidentia et dérivé de evidens, -entis se dit de ce “qui se voit de loin”. Or ni le ouï-dire, ni le non-vu, ni l’inférentiel, termes souvent employés pour désigner certains de ces marqueurs, ne peuvent être considérés comme des “évidences” au sens français du terme. De plus, il correspond à l’intuition de nombreux linguistes non anglo-saxons qui ont décrit le même phénomène (parfois même avant Sapir et Boas) en employant des termes qui marquent clairement que celui-ci ne relève pas d’une quelconque évidence (Guentchéva 2004: 22-23).

The problem of this line of argumentation is that English evidence, German Evidenz and Spanish evidencia also etymologically derive from the same basis as French évidence and the corresponding linguistic category that comprises expressions which convey the information source or the type of evidence a speaker has for his statement is conveniently called evidentiality, Evidentialität and evidencialidad. The notion of evidentiality was derived from English ‘evidence’ which, in turn, derived from Latin evidentia (cf. Bußmann 2002: 206). As the notion that evidentiality derived is from ‘evidence’, as ‘evidence’ itself is not automatically associated with direct evidence and as evidence may be direct or indirect or rather mediated or personal (cf. the main difference that is made by Willett 1988 and Plungian 2001), there is no reason to reject the translations of evidentiality. In short, the notion of evidentiality does not make any reference to direct evidence but that is why Guentchéva rejects the notion of évidentiel and évidentialité. In my view, the Spanish notion of evidencialidad, based on the English original ‘evidentiality’ (as the French notion of évidentialité does) can be perfectly applied to certain Spanish expressions as they convey that there is some evidencia ‘evidence’ – regardless of whether direct or indirect but some – for the speaker’s statement. There is no reason to assume that evidence is only evidence if it is direct as ‘evidence’ developed further from Latin evidentia. In certain contexts, direct evidence represents the more reliable kind of evidence but the statement “[l]e terme ‘évidentiel’ […] évoque l’évidence, c’est-à-dire la constatation directe” (Guentchéva 1996: 13) is not considered correct and – as a consequence – this statement does not justify the rejection of the notion of évidentialité. The notion of médiation, in contrast, leaves the impression that expressions like I have seen that [p] were not subsumed below this category. “La catégorie du médiatif […] s’organise autour de trois valeurs fondamentales: 1) faits rapportés, y compris selon le ouï-dire; 2) faits inférés; 3) faits de surprise” (Guentchéva 1994: 9). This is different from evidentiality. Both Lazard (1996) and Guentchéva explain that the category of médiation and the médiatif markers are used to mark [p] as mediated information:

26 On peut dire que la catégorie du médiatif existe dans une langue quand elle possède une ou plusieurs formes verbales qui s’opposent à une ou des formes neutres par le fait qu’elle(s) présente(nt) les faits non pas purement et simplement, mais médiatement, c’est-à-dire à travers la connaissance que le locuteur en a prise, à laquelle il fait référence et par rapport à laquelle il prend par là-même un certain recul. Elles peuvent ainsi évoquer des faits connus par ouï-dire ou par inférence ou par une constatation inattendue […] ou de tout autre manière. (Lazard 1996: 28). […] l’information est obtenue par voie médiate, elle recouvre des faits recueillis auprès d’un tiers ou connus par ouï-dire, ou encore des faits inférés ou reconstruits à partir d’indices; il peut s’agir aussi de faits mythiques ou légendaires, ou encore de faits rêvés. L’énonciateur occupe donc une position qui le contraint à s’impliquer ou, au contraire, à dégager sa responsabilité par rapport à la situation décrite (Guentchéva 1996: 11).

Hence, it is not only a notional choice whether to prefer évidentialité or médiation but also obviously a conceptual one. The latter seems to comprise only those expressions which Plungian (2001: 353) subsumes below the heading of mediated evidence. But what about direct/personal evidence? As the category of médiation only comprises expressions that indicate indirect evidence, Guentchéva argues for a superordinate category that encompasses mediated enunciations and statements: En effet, on commence à l’utiliser pour désigner la notion de “médiation” dans des langues où le phénomène n’est pas grammaticalisé, voire pour désigner une (super)catégorie qui inclut la notion sémantique de “constatif”. On remarquera que dans tous le cas, il est indispensable de poser le problème de la relation entre énoncé assertif et énoncé médiatif (Guentchéva 1996: 12).

Discussing the relationship between médiation and other categories such as modality (cf. also Kronning 2004: 59), Guentchéva opposes the category of ‘mediated’ and assertion (cf. Guentchéva 1996: 30). This implies that statements which are introduced by I have seen that or I know that would count as assertions. An assertion is defined as an ‘event which is presented as certain’: Il me semble […] prudent de bien isoler et de décrire de façon autonome la catégorie de la médiation en l’opposant à l’assertion, à la simple “prise en charge” et aux modalités. La catégorie de l’assertion (un fait assumé est présenté comme certain) constitue un leu d’articulation entre la catégorie de la médiation et les catégories modales (Guentchéva 2004: 30).

This leads to the assumption that statements which are marked by I have seen that or I know that are interpreted as assertions since they convey information which is presented as certain. But besides presenting information as certain, such statements are evidentially marked: the speaker claims to have direct evidence for [p]. So such utterances can neither be described in ‘mediated terms’

27 nor are they adequately captured by the term ‘assertion’. This, in turn, is the reason why it makes sense to use the term évidentialité. It describes expressions that convey either direct/personal or indirect/mediated evidence. Dendale/Tasmowski (1994), who apply the notion of évidentialité to French evidential expressions and thus ignore the terminological discussion, understand the French equivalent of ‘evidentiality’ as a semantic-functional category and investigate which expressions may be used to indicate what kind of evidence. Hence, their definition is not exclusively based on semantics as Kronning (2004) erroneously states: Cette définition13 est de nature fondamentalement sémantique, car, par “marqueur évidentiel” ces linguistes désignent toute expression linguistique – qu’elle soit morphologique ou lexicale, voire typographique – qui indique (la nature de) la “source du savoir” (emprunt, inférence, perception) que communique l’énoncé (Kronning 2004: 37).

In common with Dendale/Tasmwoski, who study French expressions regarding their evidential use applying the notion of évidentialité to those expressions that are evidentially used, the notion of evidentiality or its translated versions is considered applicable to any linguistic expression that indicates the information source. Hence, certain Spanish linguistic devices can also be (and should be) analysed in evidential terms. In the present chapter and in the following subchapters the notion of evidentiality will thus be extensively dealt with. For some [for Aikhenvald (2003a14; 2004), for instance], evidentiality is a minor grammatical category manifested in a small number of the world’s languages. For

13

This is the definition by Dendale/Tasmowski (1994) Kronning refers to: Un marqueur évidentiel est une expression langagière qui apparaît dans l’énoncé et qui indique si l’information transmise dans cet énoncé a été empruntée par le locuteur à autrui ou si elle a été créée par le locuteur lui-même, moyennant une inférence ou une perception (Dendale/Tasmowski 1994: 5).

14

By way of example, she states the following: One of the current misconceptions concerning evidentiality is to do with a gratuitous extension of this term to cover every way of expressing uncertainty, probability and one’s attitude to the information, no matter whether it is expressed with grammatical or with lexical means; or whether it is the primary meaning of a category or not, or talking of evidentiality in a ‘broad sense’ – by Chafe’s […] definition: as marking speaker’s attitude towards his/her knowledge of reality as opposed to its ‘narrow sense’: marking the source of such knowledge. This is unhelpful and quite uninformative (Aikhenvald 2003a: 19).

28 others15, evidentiality is a semantic category which may be realised grammatically, lexically or paraphrastically (Mushin 2001: 17).

Mushin’s words illustrate that it is to differentiate between evidentiality in its narrower sense via evidentiality in its broader sense and evidentiality in its broadest sense. Evidentiality in its narrow(est) sense refers to morphological means of expressions, i.e. affixes, particles or clitics encoding the source of information. Thus, evidentiality in its narrow sense deals with languages possessing real evidentials, which “may be generally defined as markers that indicate something about the source of the information in the proposition” (Bybee 1985: 184).16 Evidentiality in its broader sense refers not only to the mentioning of the information source, but also to the speaker’s17 attitude towards the proposition (cf. chapter 2.1.1). Hence, evidentiality is understood as a semantic-functional category that may not only be realised grammatically but also lexically. That is why evidentiality in its broader sense can also be dealt with in languages without grammaticalised evidentials (cf. Volkmann 2005: 79, 453). “Even when languages lack explicit obligatory evidential categories, other devices may do similar things” (Haviland 1989: 36)18 because “the importance of evidentiality […] lies in its function rather than in its form” (Cornillie 2007a: 7). Linguists who Squartini’s words can be read as a brief summary of the ongoing controversy between typologists and linguists who consider evidentiality as a functional category: The increasingly abundant research on evidentiality has given rise to a sort of implicit confrontation between those, mainly typologists, advocating a strict definition of evidentiality within the domain of grammar, and those considering evidentiality as a more general functional category whose scope includes not only grammatical but also lexical phenomena and can therefore be extended to languages traditionally considered as unaffected by evidentiality (Squartini 2007: 1). 15 16

17

18

I definitely belong to ‘the others’. The collective monograph Evidentials. Turkic, Iranian and Neighbouring Languages, for instance, is concerned with such ‘markers’. It contains studies about “evidential categories found in the verbal systems of Turkic and Iranian languages as well as in some of their contact languages, e.g. Slavic, Finno-Ugric, Tungusic, Caucasian and Armenian” (Johanson/Utas 2000: v). In the present study, the one and only personal pronoun to refer to ‘a speaker’ will be ‘he’. Even though I consider myself a politically correct person I reject the s/he-orhis/her-writing style for the sake of an ‘easy-on-the-eyes reading’. Croft states nearly the same: In general, cross-linguistic identification cannot be accomplished on purely formal (structural) grounds for two reasons. First, variation across languages is too great […] Second, formal definitions are internal to the structural system of a single language […] (Croft 1995: 88).

29 perceive evidentiality as a semantic-functional category and consequently analyse linguistic means serving evidential functions term this linguistic domain “a heterogeneous category, of which the only general criterion is the fact that certain elements mark the source of evidence” (Haßler 2002: 159). De Haan (2006) also says justifiably: Evidentiality has long been considered an exotic category, associated with Native American languages primarily. However, it is a category that manifests itself in languages on every continent, including in some well-studied languages of Europe […] (de Haan 2006: 57).

Chafe is a linguist, who coined the notion of evidentiality in its broadest sense (1986: 262). He himself summarises his understanding of the linguistic category at the end of his paper “Evidentiality in English Conversation and Academic Writing” as follows: “‘Evidentiality’ has been used here broadly to cover any linguistic expression of attitudes toward knowledge” (Chafe 1986: 271). In this respect, Willett observes: Chafe [...] opts to regard evidentiality [...] as roughly equivalent to epistemology in Lyons’ sense or epistemic modality in Palmer’s sense; that is, as an indication of the source and reliability of a speaker’s knowledge (Willett 1988: 55).

Generally, it should be strictly differentiated between a pragmatic understanding of evidentiality and a typological understanding of this notion. This is what is also pointed out by Nuyts, whereby he addresses the interactional function of evidentials relating it to the concept of (inter)subjectivity (cf. chapter 2.1.3). But this relationship is usually not recognised in the studies of typologists, which thus represents the essential difference between pragmatic and typological research in the field of evidentiality: The […] formulation of the dimension of (inter)subjectivity [...] assumes a crucial role for the speaker’s assumptions about the hearer’s knowledge, i.e., it (often) crucially involves the relation between speaker and hearer in the actual conversational setting [...]. So maybe this interactive dimension is not too exceptional either, and may even be more widespread than has been acknowledged so far. Typological research usually works with informants and hardly ever considers real conversational or discursive data […] (Nuyts 2001a: 396-397).

Obviously, there is (still) a lot more to say about evidentiality than only ‘evidentiality is entailed in any linguistic form expressing the source of information for a certain proposition’. This is the functional-semantic definition worked with here (on an interim basis up to chapter 2.2), but one should have a detailed view on the notion itself and all its surroundings, namely the categories it overlaps with.

30

2.1 Where evidentiality and other linguistic categories meet Evidentiality overlaps and interfaces with other linguistic domains such as epistemic modality or deixis. This overlapping with other categories is what makes the notion of evidentiality so complex. Hence, in the following subchapters it will be dealt with different linguistic categories and where they meet evidentiality, whereby evidentiality is – for now – understood as a semantic-functional domain having to do with the kinds of “proof” speakers are able to adduce in order to underpin their statements. Thus the study of evidentiality in language is concerned with the linguistic means languages provide for referring to “reasons”, “indications”, “evidence”, or – in short – sources of information speakers have for expressing statements (Diewald/Smirnova 2010a: 2).

I agree with this understanding of evidentiality but I would like to draw attention to the condition for a linguistic means to be analysed in terms of evidentiality: the semantic-functional domain of evidentiality has to be linguistically encoded by concrete evidential expressions. To my mind, it is not sufficient to say that evidentiality has to do “with the kinds of ‘proof’ speakers are able to adduce in order to underpin their statements”. The category needs to be represented by (a) concrete form(s). Otherwise whole stories told by speakers could be termed ‘evidential’ if, for instance, the story was told in order to answer the question ‘Do you have proof for that you were not at home last night?’. Then, a whole story would serve as a ‘proof’ for a speaker to underpin his statement that he was not at home last night. This is the only detail that should be accentuated regarding the definition proposed by Diewald/Smirnova (2010a). Therefore, their definition of evidentiality presented in their volume on Linguistic Realization of Evidentiality in European Languages is slightly preferred, where they speak of an “explicit encoding” of an information source: The basic characteristic of linguistic evidentiality is the explicit encoding of a source of information or knowledge (i.e. evidence) which the speaker claims to have made use of for producing the primary proposition of the utterance. The type of evidence the speaker adduces may be of various kinds (i.e., different modes of perception or cognition), which may be encoded by the evidential expressions but also may be left unspecified (Diewald/Smirnova 2010b: 1).

Keeping this in mind for the following subchapters, let us now have a look at evidentiality, other linguistic categories and where they meet.

31

2.1.1 Where evidentiality and epistemic modality meet Before treating the relation between evidentiality and epistemic modality, it is useful to devote a few words to modality in general and then to the linguistic category of epistemic modality itself. First of all, ‘modality’ remains one of the most problematic and controversial notions: there is no consensus on how to define and characterise it, let alone on how to apply definitions in the empirical analysis of data. And there are no signs that the debates are heading in the direction of a final solution (Nuyts 2005: 5).

As there is no consensus on how to define modality19, in the following a few definitions and statements – of diverse nature – concerning modality will briefly be compiled. The following words by Magni (2010) may roughly explain why there is no consensus on how to apply a ‘correct’ definition of modality in the analysis of data. Above all, it seems to depend on the kind of data because modality ‘has scope over’ various areas of grammar and can thus be investigated in various areas: The term modality designates a wide conceptual domain […]. From a linguistic point of view, modality affects all areas of grammar, interacting with other categories such as negation, tense, and aspect. […] modal functions do not relate to the verb alone or even primarily, but to the whole sentence. […] Indeed, various ideas and notions have been considered for the identification and delimitation of modality: attitudes and opinions of the speaker […] (Magni 2010: 193).

The last mentioned attitudes and opinions of the speaker are also referred to by Dubois (1969), who describes modality as the “attitude du sujet en face de l’énoncé qu’il formule” (Dubois 1969: 118) and by Roulet (1979) who speaks about “traces de la prise en charge de la prédication par l’énonciateur” (Roulet 1979: 43). While those linguists refer to the attitude of the speaker concerning his enunciation, Bally (1965 [1932]) – referring to the distinction between modus and dictum (cf. chapter 4.2) – even applies the notion of modality to sentences. For him ‘modality is the soul of the sentence’, which implies that modality is said to be the soul of every sentence, as, in his view, ‘there is no utterance without modality’ (cf. Bally 1965: 35). But if modality was not bound to certain linguistic units, how then discover modal expressions? Givón (2001), in contrast, seems to have a pragmatic understanding of modality, explaining that modality’s propositional scope “is not the propositional semantics of the atomic event or state, but rather the pragmatics, or connectivity of the clause vis-à-vis its discourse context” (Givón 2001: 285). Halliday (1970) also relates his under19

Dendale/van der Auwera laconically summarise: “La modalité, on le sait, est une notion aux définitions multiples et aux frontières floues” (Dendale/van der Auwera 2001: i).

32 standing of modality to the discourse context, or rather, to the speech event: “Modality is a form of participation by the speaker in the speech event. Through modality, the speaker associates with the thesis an indication of its status and validity in his own judgements” (Halliday 1970: 335). As – being obviously such a wide conceptual domain (cf. Magni 2010) – modality does not only affect all areas of grammar but may also be represented by lexical means, modality should be termed a semantic-functional category (cf. Becker/Remberger 2010: 1; cf. the review of Nuyts’ paper (2005) in chapter 3.15). In the following, that type of modality which is of high importance for the present study will be dealt with, viz. epistemic modality. Although appropriate, the following remark on epistemic modality may seem a little bit too general: “[e]pistemic modality has to do with knowledge” (Portner 2009: 2). Nuyts (2001b) explains more precisely that epistemic modality expresses “the evaluation of the chances that a certain hypothetical state of affairs under consideration (or some aspect of it) will occur, is occurring or has occurred in a possible world” (Nuyts 2001b: 21). In defining epistemic modality, Coates (1995) also refers to the expression of the evaluation of chances that the state of affairs under consideration is happening/has happened/will happen. In contrast to Nuyts, Coates speaks about the assumptions of a speaker or his assessment of possibilities: Epistemic modality is concerned with the speaker’s assumptions or assessment of possibilities, and in most cases it indicates the speaker’s confidence or lack of confidence in the truth of the proposition expressed (Coates 1995: 55).

The speaker’s confidence or lack of confidence in the truth of the proposition represents, strictly speaking, only one kind of epistemic modality. It is well summarised by Volkmann (2005: 53, 54, 453, 454, 456) that there are two distinct kinds of epistemic modality a speaker can express: evaluation and validation. So the speaker’s (modal) attitude may be evaluative or ‘validative’ in nature.20 The former refers to the evaluation of a proposition in negative-positive terms and the latter refers to the evaluation of a proposition in terms of its truth value. This implies that, dealing with epistemic modality in connection with evidentiality, it is always the kind of ‘validative’ epistemic modality that linguists refer to.21 20

21

The terms ‘validative attitude’ and ‘evaluative attitude’ are not only adopted but also translated from Volkmann (2005). She speaks of “validative Einstellung” and “evaluative Einstellung” (2005: 53) or “validative Einstellungsbekundung” and “evaluative Einstellungsbekundung“ (2005: 54). For the (non-existent) distinction between subjective and objective modality, see chapter 2.1.3.

33 Dendale/Tasmowski (2001: 341-342) do not state without reason that the relationship between evidentiality and epistemic modality can be described (very) differently in the literature.22 They observed three kinds of interpretation of the relationship between the two categories: 1. disjunction: evidentiality and modality are conceptually distinguished from each other; 2. inclusion: one domain is included in the other (evidentiality is mostly regarded as being included in modality); 3. overlap: evidentiality and modality partly overlap (they share an intersection area). Boye (2006) also observes these three kinds of interpretation of the relationship between evidentiality and epistemic modality. Like Dendale/Tasmowski (2001), he notes: […] modality and evidentiality are sometimes considered completely distinct categories, sometimes distinct but overlapping categories, but at other times the former category is taken to comprise the latter, and at still other times the latter category is taken to comprise the former […] (Boye 2006: 1).

Palmer (2001), for instance, seems to conceive the relationship of evidentiality and epistemic modality as an inclusion-like one, as he speaks about “evidential modality” (2001: 8, 24; 2003: 7). He says verbatim: […] epistemic modality and evidential modality are concerned with the speaker’s attitude to the truth-value or factual status of the proposition and may thus be described as “propositional modality”. The basic difference between epistemic modality and evidential modality is that with epistemic modality speakers make judgments about the factual status of the proposition, whereas with evidential modality they indicate what is the evidence that they have for it (Palmer 2001: 24).23

Although clearly differentiating between epistemic modality and evidential modality, both categories are termed ‘modality’, which leads to the conclusion that he perceives their relationship as an inclusion, whereby he regards evidentiality as falling within the category of modality. In distinguishing epistemic modality 22 23

Hereunto: Wachtmeister Bermúdez (2004: 8), de Haan (1999, 2001a) and Volkmann (2005: 90-95). Also in his earlier work, Palmer (1986: 51) ‘imposes’ the notion of modality on evidentiality: the term ‘epistemic’ should apply not simply to modal systems that basically involve the notions of possibility and necessity, but to any modal system that indicates the degree of commitment by the speaker to what he says. In particular, it should include evidentials such as ‘hearsay’ or ‘report’ (the quotative) or the evidence of the senses.

34 from other kinds of modality, de Haan (1997) also explains, partly quoting Palmer (1986): I reserve the term epistemic for those categories that indicate “the degree of commitment by the speaker to what he says” (Palmer 1986: 51) plus the information on which the speaker bases his utterance (de Haan 1997: 5).

Hence, he ‘adds’ evidentiality to the category of epistemic modality. Bybee/Fleischman (1995) argue in the same direction: they speak about modality as the domain including the other. They may overlap, or modality may even encompass evidentiality: […] as applied to natural language, there is no reason to restrict the epistemic notion just to necessity and possibility, as is traditional in philosophy of language. For one thing, commitment to the truth of a proposition is often a matter of degree. For another, epistemic modality can be seen as overlapping with, or even encompassing, another grammatical category, namely evidentiality (Bybee/Fleischman 1995: 4).

Hoye, on the one hand, simply speaks only about an ‘association’ of one category with the other: […] since evidentiality is concerned with the nature and source of evidence on which a speaker’s assumption or judgement is so often based, it may clearly be associated with the concept of modality, specifically epistemic modality – that is the category of modality which concerns the speaker’s ‘knowledge’ or lack of ‘knowledge with regard to a particular state of affairs (Hoye 2008: 153).

On the other hand, he seems to go for a relation of the third kind, namely overlapping, because he assumes an ‘interaction’ of both categories: The broad, functional take on evidentiality adopted here, recognizes that, whilst there are differences between the evidential and (epistemic) modal systems, it must also be acknowledged that the two are closely associated and clearly interact […] (Hoye 2008: 165).

For Aikhenvald, both linguistic domains are neither associated, let alone closely, nor do they interact: Evidentiality is a linguistic category whose primary meaning is source of information. […] this covers the way in which the information was acquired, without necessarily relating to the degree of speaker’s certainty concerning the statement or whether it is true or not (Aikhenvald 2004: 3).

In making clear that in her opinion evidentiality and epistemic modality have a relationship of disjunction, she simultaneously admits an extension of the meaning of evidentials to epistemic modality: “That evidentials may have semantic extensions related to probability and speaker’s evaluation of the trustworthiness of information does not make evidentiality a kind of modality” (Aikhenvald

35 2004: 7-8).24 Although evidentials may have extensions related to epistemic modality, the fact that evidential morphemes as well as modal ones can – at the same time – be combined with a verb seems to be even more relevant for typologists. This allows them to claim a relationship of disjunction, even though admitting an extension of the meaning of evidentials related to the speaker’s evaluation of the trustworthiness of information expressed. However, Große (2011), for instance, also shows that the Guaraní hearsay marker ndaje, which is mostly translated as Spanish dicen que ‘they say that’ or se dice que ‘it is said that’ (Große 2011: 242), is often used as a modal marker if used in the Spanish language variety spoken in Paraguay. The modal use of the evidential is confirmed by examples in which hearsay adverbs such as aparentemente or supuestamente are used together with ndaje (Große 2011: 249; cf. chapter 3.7). Thus, evidentials may indeed convey a modal reading. In Romance languages, adverbs that etymologically denote the source of knowledge via visual sense or sense of hearing are nowadays also used as modal adverbs (cf. Haßler 2010a on Italian ovviamente, but also Haßler 2004). Indeed, the relation between evidentiality and modality is described differently. In my view, we clearly deal with different (semantic-functional) categories. They seem to be even ‘more different’ when dealing with the typological notion of grammatical evidentiality, which “is only found in about 25 per cent of the world’s languages” (Aikhenvald 2004: 17). But having a closer look at linguistic devices serving evidential functions in languages that lack such a grammatical system of evidentiality unequivocally reveals that the two domains may overlap as the analysis of empirical data from Spanish will show (cf. chapters 57.3). The fact that epistemic modality and evidentiality may sometimes overlap can even be emphasised by a language belonging to the above mentioned 25 per cent: Eastern Pomo. Eastern Pomo is one of seven distinct but related Pomoan languages spoken in northern California in the coastal range of mountains between the Pacific Ocean and the Sacramento Valley. All formally differentiate the sorts of evidence on which an assertion is based, but they differ among themselves in both the forms used and the number of categories distinguished. All are agglutinative, verb-final languages with rich phonological, morphological, and semantic systems (McLendon 2003: 101).

24

For Turkic, Iranian and neighbouring languages, Comrie also explains that in a particular context an evidential form may seem to be related to epistemic modality: […] the evidential systems as such do not necessarily involve any casting of doubt on the reliability of the information conveyed, although a form that indicates an indirect source for information may, in a particular context, receive such an interpretation, but crucially not as its invariant meaning (Comrie 2000: 2).

36 The evidential system of Eastern Pomo provides a suffix to mark ‘direct knowledge’: The direct knowledge suffix -ya (-a after vowels) indicates that the event referred to by the verb is actually happening or has just happened and that the speaker has direct knowledge of it, primarily because the speaker performed or experienced the action, process, or state (McLendon 2003: 106).

Generally, the “direct knowledge suffix -(y)a […] reflects the speaker’s point of view as well” (McLendon 2003: 113). So if the direct knowledge suffix represents one sort of evidence a speaker can provide for his assertion, epistemic modality and evidentiality clearly intersect. Or is there something ‘more epistemic’ than direct knowledge? However, we do deal with two distinct semantic-functional categories: As a rule of thumb, one can say that while evidentiality indicates the source of evidence a speaker has for making a statement, without necessarily accompanying that with a factuality judgment, epistemic modality, [sic] is concerned exactly and exclusively with the latter, i.e. with the degree of factuality a speaker attributes to a proposition (Diewald/Smirnova 2010a: 6).

Therefore, various attempts to draw a clear-cut distinction – i.e. a relationship of disjunction – between both domains can be observed. Squartini (2004), for instance, tries to disentangle evidentiality and epistemic modality in Romance and de Haan (1999) wants to set boundaries. Cornillie’s analysis of the relationship between the two domains “also contrasts with accounts based on the ‘inclusion’ or the ‘overlap’ of the two categories” (2009: 44). He presents three case studies – of epistemic auxiliaries, adverbs and verbs of cognitive attitude (Cornillie 2009: 49) – which illustrate “that an expression is either evidential or epistemic and that within these qualifications different readings are possible” so that he concludes “that there is neither inclusion nor overlap between the two qualifications”. (Cornillie 2009: 59). Another approach to disentangle the categories is with the help of deixis (cf. chapter 2.1.2). Although Fitneva (2001) does not mention the concept of deixis when speaking about “Epistemic marking and reliability judgments”, offering evidence from Bulgarian, she describes the speaker’s reference to the addressee as a property of evidentiality and thus uses this distinctive feature to distinguish the source-of-information marking from epistemic modality. Speaker attitude [epistemic modality] is used when the speaker is in a position to competently decide on the reliability of information; source of information [evidentiality] is used when the speaker-hearer dyad is better off to negotiate the reliability of the information (Fitneva 2001: 405).

37 If the speaker’s purpose of communicating the information source “is to solicit the opinion of the hearer” (Fitneva 2001: 405) and to discuss its reliability, the speaker then involves the hearer “in determining what is reliable” (Fitneva 2001: 418). This implies that a closer look at the speaker-hearer interaction is necessary, which in turn means analysing evidential markers in their (pragmatic) use. Mushin (2001) even terms evidentiality ‘a deictic category’ and consequently relates the study of its linguistic manifestation to the field of pragmatics: “Like other deictic categories, the semantics of evidentiality can only be understood alongside a detailed account of its pragmatic properties” (Mushin 2001: 35). But mentioning ‘pragmatic properties’ is not enough to call evidentiality a deictic category, as other linguistic categories also have pragmatic properties. Nevertheless, it is correct to call evidentiality a category with deictic properties. The relationship between evidentiality and deixis will be dealt with in more detail in chapter 2.1.2. In her paper “Epistemic modality and evidentiality and their determination on a deictic basis: the case of Romance languages” Haßler (2010a) addresses problems in distinguishing between evidentiality and epistemic modality, underpinning them with examples from Romance languages containing adverbs. Because of the fact that “there are demarcation problems in the Romance languages between epistemic modality and evidentiality and also overlaps between the different kinds of evidentiality” (Haßler 2010a: 237; cf. also Guentchéva 1994, Squartini 2004), she uses the concept of deixis to identify evidentially used means: With the determination of the category of evidentiality as marker of the origin of the speaker’s knowledge there is a category centred on the speaker. It therefore seems justified to include deixis in the determination of evidentiality (Haßler 2010a: 237).

As deixis is defined as a specific mechanism of referentialisation – because it refers on the one hand to the context of the utterance and establishes on the other hand a relationship to the speaker – evidentiality, in contrast to epistemic modality, is described as a deictic phenomenon (Haßler 2010a: 223, 239, 243). While epistemic modality adds the stance of the text producer to the predication, evidentiality presumes the production of a reference to the source of knowledge by the recipient and thus a judgement of trustworthiness (Haßler 2010a: 243; cf. also Haßler 2010d: 106).

What makes evidentiality – according to Haßler (2010a, 2010d) – in contrast to epistemic modality, a deictic phenomenon is the fact that epistemically used expressions contribute to the stance of the speaker without also referring to ‘something extra-linguistic’.

38 Similar ‘deictic approaches’ to evidentiality are used by Wachtmeister Bermúdez (2005), de Haan (2005) and Frawley (1992), which will be dealt with in the following chapter, even though – generally speaking – “descriptions of the deictic nature of evidentiality have been given relatively little focus in the literature” (Mushin 2001: 33).

2.1.2 Where evidentiality and deixis meet25 First of all, a few words should be dedicated to the notion of deixis itself: usually, deixis is defined as the process of pointing via language, more precisely, via verbal expression or non-verbal expression, i.e. gestures (Bußmann 2002: 149). ‘Classical’ deictic elements, which are commonly acknowledged as being deictic, are the personal pronouns I and you, local adverbs such as here and there, temporal adverbs such as now and then, other time references such as yesterday and tomorrow, and demonstratives like this and that etc.26 All these deictic elements depend on the situation of utterance (cf. Bußmann 2002: 149; Allott 2010: 55), that is, without considering the context in which they appear, they cannot be interpreted correctly.27 Dictionaries of pragmatics or linguistics in general describe deixis as “[a] technical term used in semantics and pragmatics for linguistic items that encode sensitivity to context and for uses of linguistic items that involve this kind of sensitivity” (Allott 2010: 54; my emphasis) and as “[t]he way in which the reference of certain elements in a sentence is determined in relation to a specific speaker and addressee and a specific time and place of utterance” (Matthews 2007: 96; my emphasis), respectively. These definitions illustrate that deixis may also be understood in a broader sense: not only ‘classical’ deictic elements encode sensitivity to context or have to be interpreted in relation to a specific speaker and addressee. Non-deictic elements may also encode sensitivity to context or should be analysed in relation – at least – to the speaker. That is why it is usually distinguished between “a set of linguistic items which are primarily deictic” and other elements “which are not but may be used deictically” (Allott 25

26

27

A big part of the remarks on the relationship between deixis and evidentiality is taken from my master thesis Das Zusammenspiel der linguistischen Kategorien Evidentialität und Deixis im Beispiel spanischsprachiger Kriminalromane (2008; unpublished document). Overall, it is to distinguish between person deixis, spatial dexis, temporal dexis, discourse deixis and social deixis (Bußmann 2002: 149; Allott 2010: 57), whereby Bußmann (2002: 149) refers to demonstratives as this and that in terms of object deixis. For further information on the complex and problematic notion of deixis – especially in dissociation of ‘anaphora’ – see Consten’s study (2004).

39 2010: 55; cf. also Eguren 1999 and Lyons 1995). As the crucial feature of deixis is its context-boundedness, as deixis describes the process of pointing, and as “[i]t is usual to understand deixis in terms of deictic centres” (Allott 2010: 57), it is, for instance, possible and reasonable to describe a particular use of modal verbs as deictic (Diewald 1999, 2000; cf. chapter 2.1.2.1). So modals represent one example of those elements which may be used deictically. Due to this broader understanding of deixis, it is also possible and reasonable to interpret even whole linguistic categories such as evidentiality, epistemic modality and subjectivity as deictic or as – at least – being related to deixis (cf. chapter 2.2). Nevertheless, the views of what exactly should be termed ‘deictic’ or rather which expressions should be described as ‘used deictically’ differ (cf. chapter 2.2 for the contrary lines of argumentation by Haßler 2010a, 2010d and Diewald/Smirnova 2010a). In summary, evidentiality may be considered as being related to deixis and the use of evidential expressions may be described as deictic. In the following the study will thus focus on the deictic character of evidentiality and deal with papers that treat evidentiality as a deictic phenomenon. The deictic character of grammatical evidentiality has already been pointed out in earlier works (e.g. Jakobson 1971). Indeed, in a few works on grammatical evidentiality the ‘deictic nature’ of evidentiality has – albeit mostly implicitly – been already mentioned. Let us have a look at the studies by Hensarling (1982), Lowe (1972), Gómez Rendón (2006), Haviland (1989), Schlichter (1986) and de Haan (2005, 2001b). Below the chapter title “Participant systems” Palmer explains: “In some languages there are systems indicating more direct reference to the participants in the discourse” (2001: 62). In other words, there are evidential systems in which the discursive position of the hearer is part of the system. Such a language system is studied by Hensarling (1982; quoted in Palmer 2001: 62-63), who analyses the Kogi evidentials (spoken in North Colombia). “In Kogi [...] there is, Hensarling (1982) tentatively suggests, an ‘evidential’ system indicating ‘who knows what about the situation being discussed’” (Palmer 2001: 62). These are the particles by which the speaker refers to the hearer’s state of knowledge: Table 1: Kogi evidentials (cf. Palmer 2001: 62) PARTICLE

SPEAKER

HEARER

MEANING

ni na shi skaN ne

+ + – – –

+ – + – ?

‘to remind’ ‘to inform’ ‘to ask’ ‘to doubt’ ‘to speculate’

40 By using the particle ni ‘to remind’ the speaker not only indicates that he knows about a certain state of affairs but also that even the hearer should know about it28 – provided that the hearer is able to remember. If the speaker uses the particle na ‘to inform’ he indicates that he presupposes that the addressee does not know (yet) about the state of affairs. The particle shi ‘to ask’ implies that the hearer is thought to know about a certain state of affairs but not the speaker.29 Frawley, who also studies the ‘deictic nature’ of the Kogi evidentials, remarks as well: “[t]he three [...] forms, ni, na, and shi, explicitly make reference to the addressee” (Frawley 1992: 415). The particle with the semantic meaning ‘to doubt’, skaN, indicates that neither speaker nor hearer exactly know about a certain state of affairs. In this connection Lyons (1995: 191) explains: […] if speakers express doubt as to the truth of a particular proposition, in conversation rather than in soliloquy, they may well be understood in context (and expect to be understood) to be inviting the addressee to resolve their doubt for them […].

By using the particle ne the Kogi speaker expresses that he (only) assumes a certain state of affairs, whereby he does not consider the hearer’s state of knowledge (cf. Palmer 2001: 63). “A more complex system is proposed by Lowe (1972) for Nambiquara (Brazil)” (Palmer 2001: 63). It is the ‘event verification system’ in which the speaker as well as the hearer is involved (cf. Lowe 1972: 367-372): Individual verification forms treat of an event verified by the speaker only, who then relates his account to the addressee; collective verification forms treat of an event verified simultaneously by both speaker and addressee (Lowe 1972: 369; emphasis in the original).

The linguistic forms of the ‘collective verification system’ have the general meaning ‘I report what both I and the addressee saw the actor doing’ (Lowe 1972: 371), more precisely, the speaker tells what both he and the hearer have seen. In this situation – in which the speaker talks about a state of affairs that 28 29

Cf. Wachtmeister Bermúdez’ (2005: 15-17) arguments about the “acceso compartido a la información”. Bendix shows for the Newari evidentials that they are used to take the hearer’s position into account because the employment of the Newari evidentials is even in questions obligatory. So here the speaker has to presume the addressee’s mode of knowing; he has to presuppose how the addressee might have gotten to know about the state of affairs: In a question, I request you to make an assertion and give you the basic propositional content that I want you to put into your assertion […]. Since, in Newari, the evidential endings remain grammatically obligatory in the interrogative, I must also select the kind and specificity of evidence that I indicate to you I hope you will base your answer on (Bendix 1993: 229).

41 both he himself and the addressee have witnessed – he clearly refers to the addressee’s state of knowledge. Let us have a look at a few examples: wa3kon3tait1ti2tu3wa2 3

3

1

1 2

2

1 2

3

wa ko nain tait ti tu wa 3

3

3

1

‘Both you and I saw that he worked.’

3

wa kon ya sain tait ti tu wa

‘Both you and I saw that they worked.’ 2

‘Both you and I saw that I and they worked.’

(cf. Lowe 1972: 371).

If a Kogi speaker verbalises similar utterances, he would surely use the ni particle with the meaning ‘to remind’. However, in Kogi as well as in Nambiquara the hearer is considered by the speaker and the two evidential systems involve both speaker and addressee. Gómez Rendón (2006) investigates “Interpersonal Aspects of Evidentiality in Ecuadorian Quechua”, which is also the title of his paper. Gómez Rendón goes beyond the analysis of the evidentials’ semantics by analysing them from a pragmatic-interactive understanding of evidentiality: [E]videntiality is a pragmatic aspect more than a purely semantic one and any interpretation of evidentials must incorporate the identity of participants and their assumption of each other’s knowledge about the source of information (Gómez Rendón 2006: 38). [E]videntiality is anchored in the pragmatics of the speech act and […] all analyses need to start at the interpersonal level (Gómez Rendón 2006: 41).

He therefore proposes an “interpersonal reading” (2006: 37) of evidentiality as in his view the use of evidentials has to be interpreted as the result “of an interlinking of the speaker’s assumption about his/her interlocutor’s knowledge of the information source and the identity of participants in the speech event” (2006: 37). Although a speaker may report what both he and his interlocutor saw, a speaker usually mentions the information source when he firstly thinks the hearer does not know it and when he secondly assumes that the mentioning of the information source is relevant for the communicative situation. Ifantidou also describes evidentials within the framework of Relevance Theory (2001) as representing the speaker’s intention to make his utterance relevant enough to grasp the attention of the hearer (cf. chapter 4.5.2 of the present study). Saying this, Ifantidou also mentions the interactive function of evidentials: An assumption with no strength (i.e. a totally unevidenced assumption, for example a groundless speculation) can achieve no relevance. Relevance is achieved by modifying a set of existing assumptions, by strengthening them, contradicting and eliminating them, or combining with them to yield contextual implications. An assumption with no strength can achieve relevance in none of these ways. Yet a speaker aiming at optimal relevance must intend her [or his] utterance to be relevant enough to be worth the hearer’s attention. It follows that she [or he] must expect at least

42 some of the assumptions expressed and implied by her [or his] utterance to be strong enough (i.e. evidenced enough) to achieve the intended effects. (Ifantidou 2001: 196).

Back to the evidentials in Ecuadorian Quechua: the following utterances are verbalised by one and the same speaker. In the first case (a) he transmits the information source but he does not in the second case (b): (a) Shina-ka parla-shun fiesta-ta, Sanjuan fiesta-ka shina-mi ka-rka. ‘So let’s talk about the festival, Saint John’s festival was like this’ (cf. Gómez Rendón 2006: 41).

When the utterance was expressed, the speaker was standing in front of a young audience that was already informed about the festival by their grandparents. The speaker himself participated at the festival in the 1930s. This is expressed by the direct evidential mi. “The speaker makes the data source explicit because he thinks the addressee lacks such information and he considers it to be relevant for a felicitous exchange” (Gómez Rendón 2006: 41). Of course, reports about events are seen as more reliable if the state of affairs was witnessed by the speaker and can thus be attested. So it obviously depends on the speaker-hearer constellation if a state of affairs is evidentially marked or not. If a speaker uses an evidential to be seen as more reliable and to be believed by the addressee, he regards the hearer’s discursive position. As mentioned above, example (b) does not contain an evidential. The same speaker addresses hearers of the same age. He now knows that they also attended Saint John’s Festival. In view of the speaker-hearer constellation the use of an evidential does not seem necessary; in this communicative situation it is not necessary to evidentially support his reliability. Aside from that, the following example is interesting because, by uttering the postpositioned question, the speaker refers to ‘buddy’s’ state of knowledge: (b) Chai fiesta-ka primerotoro-huan-pash karka, nachu taitiku? ‘The first festivals included bullfighting, isn’t it true buddy?’ (cf. Gómez Rendón 2006: 41).

The speaker only requires verification from ‘buddy’ because he knows that the hearer also acquired his state of knowledge through personal experience. In this connection, Gómez Rendón speaks about an “interplay between participation and non-participation in the narrated event” (2006: 41). Hence, he has shown that by the (non-)use of Ecuadorian Quechua evidentials the hearer’s state of knowledge is referred to, albeit the term ‘deixis’ is not mentioned once in his study.

43 In his article “‘Sure, sure’: Evidence and Affect” (1989), where the use of the Tzotzil evidentials is analysed, Haviland clarifies his ‘very pragmatic’ understanding of evidentiality. He introductorily explains that evidentials are used to express emotions: When people fight, they typically display their feelings; and since people often fight over matters of fact, declarations of truth and accusations of deceit are frequent vehicles for affective language. […] evidentials can be about feelings and commitments as well as about truth (1989: 27).

Consequently, Haviland – just like Gómez Rendón – analyses the use of evidentials employed in interactional situations. He then introduces different Tzotzil evidential particles (Zinacantán, Mexico) and illustrates via examples from conversation that the speaker by using Tzotzil particles considers the discursive position of the addressee. In the following, what Haviland’s investigations concerning the use of the particles nan and yu’van have revealed, will briefly be summarised: [One] evidential particle is nan, glossed by Laughlin (1975) as “perhaps”. An alternative gloss might be: “I am not sure (although I am tentatively suggesting).” […] expressing doubt need not simply be a matter of propositional uncertainty, but can instead be a device for conveying interactional (perhaps even moral) effect […]. In Tzotzil, nan conveys hesitant suggestion. […] Just as nan can provide a storyteller with a device for introducing unattested speculation into a narrative, the particle can also furnish an interactive device for extracting an audience’s reaction to some bit of speculation (Haviland 1989: 40-41).

So the nan particle is often used by Tzotzil speakers if a hearer’s reaction is required. For yu’van Haviland has shown that the particle is usually employed if the speaker explicitly refers to the addressee’s state of knowledge: The particle yu’van, in utterance final position, suggests “of course, indeed, what I am now saying is true, despite what you might think (and probably you should have known it, despite the fact that you appear to have forgotten it or to be ignoring it, perhaps deliberately)” (Haviland 1989: 47).

If a speaker uses the particle yu’van at the beginning of his utterance and expresses at the end of his enunciation the phrase a-na’-oj ‘you know’, he signals the hearer that his utterance must be interpreted ironically: An initial yu’van often combines with an ironic final a-na’-oj “[…], you know, you are in possession of the knowledge that”, to form a sentence that will be heard as a sarcastic rhetorical question (Haviland 1989: 49).

If a speaker expresses utterances that are meant ironically or even sarcastically he presupposes that the addressee is nevertheless able to interpret his enunciation correctly. It could be argued that especially the use of irony or sarcasm is

44 the most complicated/complex form to refer to the addressee’s state of knowledge. In these cases, the sentence’s grammatical structure and prosody do not go hand in hand: In speech, the grammatical structure and the prosodic structure of utterances are generally complementary and mutually supportive, but […] they may also be in apparent conflict. For example, a declarative sentence may be uttered ironically to express a proposition that contradicts the proposition which, taken at face-value, it purports to express (e.g., That’s a clever thing to do!) […] (Lyons 1995: 181; emphasis in the original).

Hence, Haviland has shown that especially by using the evidential particle yu’van the speaker refers to the hearer. However, Haviland mentions the term ‘deixis’ as often as Gómez Rendón does. Schlichter (1986), who investigates “The Origins and Deictic Nature of Wintu Evidentials”, also mentions the importance of taking the speaker-hearer interaction into account when analysing the Wintu evidentials: The interpretation of the relationship between events is subjective and is marked with evidentials. It is up to the addressee to interpret the evidence he is confronted with by looking at the speaker to see if he is joking or serious, looks reliable or guilty of a lie, is in pain or happy, smart or old enough to know, or just bragging. Using evidential suffixes, the speaker appeals to the hearer as a rational person who can correctly estimate the strength of evidence (Schlichter 1986: 57).

Highly similar words are to be found in the study by Wachtmeister Bermúdez (2005): Es el oyente el que debe elaborar la referencia a la fuente de información hecha por el hablante y realizar todos los ajustes necesarios en función de su identidad, su individualidad y su posición epistémica frente a las fuentes, que puede ser diferente respecto del hablante, y a partir de allí otorgar un grado de fiabilidad (2005: 28).

Finally we should have a closer look at de Haan’s paper “Encoding Speaker Perspective: Evidentials” (2005), in which he describes evidentiality as a deictic category. He explains that evidentiality cannot be termed a modal category because evidentiality fulfils “the same function for marking relationships between speakers and actions/events” as demonstratives “for marking relationships between speakers and objects” (de Haan 2005: 1). He illustrates his ‘deictic understanding’ of evidentiality with the help of examples from the Sanuma language (Brazil; Venezuela); its evidentials were analysed by Borgmann (1990). In present tense of Sanuma, morphemes of spatial deixis relate the state of affairs to the local standpoint of the speaker with respect to the moment of the event’s happening. The following table illustrates that in the present “actions are

45 located with respect to the position of the speaker, i.e. spatially” (de Haan 2005: 13): Table 2: Sanuma morphemes (cf. Borgmann 1990: 166; quoted in de Haan 2005: 13) MORPHEME

MEANING

kule kulai / kulaai

‘Near speaker.’ ‘Fairly near, having been seen by the speaker, but at the moment hidden by some obstruction.’ ‘Farther away from the speaker but on the same level.’ ‘Upriver or across the river or even on land when there are one or more low spots or valleys between the speaker and the other person or object.’ ‘Far away inland from the river.’ ‘Far away downriver.’ ‘Up above in air, tree etc.’ ‘Down below in hole, earth, etc.’ ‘Going away from speaker on same level.’ ‘Going away from speaker upriver or across river.’ ‘Going away from speaker upriver or across river.’ ‘Going away from speaker downriver.’ ‘Toward speaker.’

kulati / kulahati kulali / kulahali

kulatili kulakili kupoli / kupoholi kupokili kimati / kuimati kimani / kimahani kuimani / kuimahani kimakili kimi

So in the present tense of Sanuma the use of evidentials implies the spatial relation between the state of affairs and the speaker at the moment of the event happening. That is why, in de Haan’s view, evidentiality is ‘more deictic’ than modal “although it must be stressed that (epistemic) modality may enter the picture at some point” (de Haan 2005: 1). But “[i]t should not be thought that epistemic modality is part of the basic meaning of evidentiality but it can be added as a pragmatic feature” (de Haan 2005: 30). So according to him, evidentiality primarily marks the distance between speaker and action: [Evidentials] are used to denote the relative distance between the speaker and the action. A speaker will use an indirect evidential to state that the action takes/took place outside the speaker’s deictic sphere, whereas the use of a direct evidential shows that the action takes or took place within that deictic sphere (de Haan 2005: 2).

Finally, evidentiality is described as ‘propositionally deictic’: […] I propose to add evidentiality to the category deixis as an example of propositional deixis. An evidential grounds an action or event with respect to the speaker; just as a demonstrative grounds an object with respect to the speaker. In other words, the relation between a proposition and an evidential is analogous to the relation between a noun (phrase) and a demonstrative (de Haan 2005: 29; emphasis in the original).

46 A similar relational structure between speaker as point of origin and proposition as point of destination – or simply, goal – is to be found in the study of Diewald (1999), who illustrates that modal verbs may display a deictic reading (see chapter 2.1.2.1). The interaction of evidentiality and deixis in Sanuma does not represent an isolated case. The same interaction is to be found in Komi (Finno-Ugric): There are two past tenses in Komi, usually referred to as the First and Second Past tense. The Second Past tense is used to denote indirect evidence. […] The use of the First Past […] marks direct evidence while the use of the Second Past on the verb […] shows that the speaker was not present at the act of doing. Thus, the different uses mark different deictic distances between speaker and action. […] consequently the choice of evidentiality, is motivated by deictic forces […] (de Haan 2005: 7-9).

De Haan has thus shown that evidentiality may be described as a notional category “which directly reflects the degree of the speaker’s involvement (or lack thereof) in the action he or she describes” (de Haan 2001b: 36). According to him, evidentiality is characterised by a deictic component and consequently by a deictic opposition: The various evidential categories […] (Sensory, Inferential, Quotative) have different degrees of speaker involvement. […] there is a deictic component (presence – absence of speaker at the place of the action), as well as an opposition witnessed – not witnessed (by speaker) to evidentiality (de Haan 2001b: 36). When the speaker uses a sensory evidential, such as a visual or auditory evidential, he/she is showing that the action somehow took place in the vicinity of the speaker, within sight or earshot, so that the action takes place in the deictic sphere of the speaker (de Haan 2001b: 36).

Because of the deictic nature of certain grammatical evidentials a few scholars (cf. Diewald/Smirnova 2010a, Haßler 2010a, 2010d, Wachtmeister Bermúdez 2005 and Hennemann 2008, 2012a) assume a similar interaction between deixis and evidentiality in languages that do not possess a system of grammatical evidentiality. Before having a closer look at the interaction between evidentiality and deixis in Romance languages and to understand the line of argumentation by Frawley (1992), Wachtmeister Bermúdez (2005) and Haßler (2010d) the introduction of Bühler’s concept of origo seems to be necessary as the notion of origo is crucial for the deictic understanding of evidentiality described in the studies referred to below. In his Sprachtheorie. Die Darstellungsfunktion der Sprache (1934) Karl Bühler presents his Zweifelderlehre ‘two-field theory’, which consists of the deictic field of language and the symbolic field. The former is represented by the

47 here-now-I system of subjective orientation (cf. Bühler 1990: 117): “Das Zeigfeld der Sprache im direkten Sprechverkehr ist das hier-jetz-ich-System der subjektiven Orientierung [...]” (Bühler 1999: 149). This system, also called origo, is crucial for the understanding of de Haan’s (2005) or Frawley’s (1992) line of argumentation in describing evidentiality in terms of deixis. But let us start at the beginning and quote Bühler’s introductory words on the deictic field of language and deictic words: [A]n Wegverzweigungen oder irgendwo im weglosen Gelände ist weithin sichtbar ein “Arm”, ein “Pfeil” errichtet; ein Arm oder Pfeil, der gewöhnlich einen Ortsnamen trägt. Er tut dem Wanderer gute Dienste, wenn alles klappt, wozu vorweg nötig ist, daß er in seinem Zeigfeld richtig steht. Kaum mehr als diese triviale Einsicht soll mitgenommen und die Frage erhoben werden, ob es unter den lautsprachlichen Zeichen solche gibt, welche wie Wegweiser fungieren. Die Antwort lautet: ja, ähnlich fungieren Zeigwörter wie hier und dort (Bühler 1999: 79).

So Bühler explains that in a countryside lacking tracks or in places where the pathway branches an ‘arm’ or ‘arrow’ that usually bears a place name helps the traveller to find his way. For this purpose, “the first requirement is that it must be correctly positioned in its deictic field” (Bühler 1990: 93; emphasis in the original). Then he poses the question “whether spoken language contains signs that function as signposts” (Bühler 1990: 93), that is, signs that fulfil the same function as an arm or an arrow bearing a place name. “The answer is yes, deictic words such as here and there have a similar function” (Bühler 1990: 93; emphasis in the original). Bühler further explains that the speaker – in a concrete speech event – plays a similar role to the one played by the arm or arrow that guides the traveller in countryside lacking branches. This role is to be distinguished from the role of the receiver. So Bühler distinguishes the roles of sender and receiver in a concrete speech event (cf. 1999: 79), whereby the speaker refers to himself by ‘I’ and to the receiver by ‘you’30: Wenn ein Sprecher auf den Sender des aktuellen Wortes “verweisen will”, dann sagt er ich, und wenn er auf den Empfänger verweisen will, dann sagt er du. Auch “ich” und “du” sind Zeigwörter und primär nichts anderes (Bühler 1999: 79).

In this way he concludes, […] that everything that involves linguistic deixis belongs together because its meaning is fulfilled and made definite in the deictic field of language and not in the symbolic field; fulfilment and definiteness can only accrue to deixis in the deictic 30

In the translated version of Bühler’s Theory of Language (1990) it is explained that the receiver is referred to by ‘thou’ as it only refers to the addressee, while ‘you’ can also include others who are not directly referred to (cf. Bühler 1990: 93).

48 field. What ‘here’ and ‘there’ is changes with the position of the speaker just as the ‘I’ and ‘thou’ jumps from one interlocutor to the other with the exchange of the roles of sender and receiver (Bühler 1990: 94).31

And this deictic field of language is represented by the here-now-I system of subjective orientation in a concrete speech event (cf. Bühler 1999: 149). It is the coordinate system of subjective orientation, whereof all participants of a speech event are parts (cf. Bühler 1999: 102-103). These findings represent the background for his concept of the origo (cf. Bühler 1999: 102): “Let two perpendicularly intersecting lines on the paper suggest a coordinate system to us, O for the origin, the coordinate source[.] My claim is that if this arrangement is to represent the deictic field of human language, three deictic words must be placed where the O is, namely the Zero point: here-now-I

deictic words here, now and I” (Bühler 1990: 117).

Fig. 1: Bühler’s concept of the origo

Consequently, every speaker is to be represented by the concept of the origo (cf. Innis 1982: 22) because it represents the deictic centre, from where deictic words have to be interpreted in order to be correctly understood. The origo therefore is a very special point of origin because it refers to the zero point of the coordinate system of every speaking ‘ego’ and it is thus ineliminable (cf. Diewald 1999: 168). As later studies on deixis show, on the one hand the origo is abstract, on the other, it is egocentric, that is to say, it is always situated in the current speaker (cf. Diewald 1999: 168). Deictic words always need an origo, that is, a reference point because they are ‘open terms’ as Eguren also explains in the Gramática descriptiva de la lengua española:

31

Compare to the original: [...] daß alles sprachlich Deiktische deshalb zusammengehört, weil es nicht im Symbolfeld, sondern im Zeigfeld der Sprache die Bedeutungserfüllung und Bedeutungspräzision von Fall zu Fall erfährt; und nur in ihm erfahren kann. Was “hier” und “dort” ist, wechselt mit der Position des Sprechers genau so, wie das “ich” und “du” mit dem Umschlag der Sender- und Empfängerrolle von einem auf den anderen Sprechpartner überspringt (Bühler 1999: 80).

49 [...] se trata de “términos abiertos”, cuya referencia no está fijada de antemano ni se mantiene constante, sino que se establece, crucialmente, cada vez que cambian el hablante, el oyente o las coordenadas espacio-temporales de los actos de enunciación (Eguren 1999: 931).

Naming words, in contrast, function as symbols and get their specific meaning or semantic specification in the synsemantic context; these language symbols belong – unlike deictic words – to the symbolic field of language (cf. Bühler 1999: 81, 151). Nevertheless, deictic words also represent symbols to a certain (semantic) extent: here and there symbolise a place surrounding the current speaker where ‘something’ that the speaker refers to is to be found (cf. Bühler 1999: 90). Just as here and there symbolise the place that surround the speaker, today, I and you represent symbols to a certain extent: today is the epitome of all days when the word is uttered and I and you represent all possible information senders and receivers of information (cf. Bühler 1999: 90).32 The most important criterion to distinguish deictic words from naming words is though: “While index words demand embedment in a situation in order to be used meaningfully, symbols need a context wherein they can make sense” (Innis 1982: 25). Every linguistic sign – whether deictic or naming word – consequently gets certain field values in a concrete speech event (cf. Bühler 1999: 84), and Bühler (1999: 149) names ‘situation’ and ‘context’ as those two sources where to get the precise interpretation, or rather the right ‘field value’ from. The concept of origo is crucial for the understanding of those studies where evidentiality is related to deixis, or rather, where evidentiality is described as a deictic phenomenon. Now, as the concept of origo is introduced, it can be dealt with the studies by Haßler (2010d) and Wachtmeister Bermúdez (2005). In these studies evidentiality is perceived as a deictic phenomenon. Below, in chapter 2.1.2.2, the paper by de Haan (2005) will be dealt with in more detail along with the study by Frawley (1992) that Haßler (2010d) and Wachtmeister Bermúdez (2005) rely on. Usually it is distinguished between direct and indirect evidentiality (cf. also Willett 1988), whereby a speaker uses direct evidential expressions if the state of affairs he talks about was directly witnessed (visually, auditive or via other 32

Compare to the original: [E]in da und dort symbolisiert, es nennt einen Bereich, nennt den geometrischen Ort sozusagen[,] d.h. einen Bereich um den jeweils Sprechenden herum, in welchem das Gedeutete gefunden werden kann; genau so wie das Wort heute den Inbegriff aller Tage, an denen es gesprochen werden kann, faktisch nennt und das Wort ich alle möglichen Sender menschlicher Botschaften und das Wort du die Klasse aller Empfänger als solcher (Bühler 1999: 90).

50 kinds of sensory perception) (cf. de Haan 2001a: 203). In this case the state of affairs is said to have occurred/is occurring within the origo of the current speaker (cf. de Haan 2005: 2). The speaker uses indirect evidential expressions if he did not witness the state of affairs directly, but if he inferred it or was informed by others. De Haan (2005: 2) thus says that by using indirect evidential expressions the speaker indicates that the state of affairs has happened beyond his deictic centre. So according to de Haan (2005) evidentiality is a deictic phenomenon because with the use of direct evidential expressions the speaker signals that the state of affairs happened/is happening within his origo, while with the use of indirect evidential expressions the speaker indicates that it happened outside his origo. In this connection, as mentioned above, de Haan (2005: 29) speaks about propositional deixis. Wachtmeister Bermúdez (2005: 29), similar to de Haan, speaks about deixis evidencial. While de Haan, in describing evidentiality as a deictic phenomenon, only refers to the opposition ‘state of affairs happened/is happening within the origo of the speaker’ vs. ‘state of affairs happened beyond the origo of the speaker’, Wachtmeister Bermúdez (2005: 39) describes the ‘deictic characteristics’ of evidentiality with respect to the relation between speaker and hearer. Downing also describes evidential expressions “as interactive devices or resources for redefining common ground between interlocutors” (Downing 2001: 251) because “evidentials encode not only what a speaker knows and how he knows it; but also what an addressee can be taken to know or should know or apparently (perhaps culpably) fails to know” (Haviland 1987: 343; quoted in Downing 2001: 278-279). So the interactive component that Wachtmeister Bermúdez and Downing take into account is not considered by de Haan. Wachtmeister Bermúdez further introduces the notion of ‘perspectivisation’ regarding the interaction between the categories of deixis and evidentiality (cf. Wachtmeister Bermúdez 2005: 28, 30-31, 43): “[...] la deixis implica una perspectivización desde el punto de vista del hablante que el oyente debe reinterpretar desde su propia posición discursiva” (Wachtmeister Bermúdez 2005: 28; emphasis in the original). Haßler (2010d) argues in the same direction. In her article “Epistemic modality and evidentiality and their determination on a deictic basis” Haßler (2010d) includes deixis in the definition of evidentiality to resolve the demarcation problems of evidentiality and epistemic modality in Romance languages, whereby deixis is understood as “a specific mechanism of referentialisation which refers to the context of the utterance and establishes a relationship to ego-hic-nunc” (Haßler 2010d: 102). Thus, deixis is not dealt with in its canonical way. Evidentiality is seen as a deictic phenomenon because evidential expressions refer to the source of information, which represents an element outside the linguistic context, and to the speaker who has access to the in-

51 formation source (cf. Haßler 2010d: 102). So evidentiality is described as a “marker of the origin of the speaker’s knowledge”, that is, evidentiality is “a category centred on the speaker” (Haßler 2010d: 101). To illustrate her theory, Haßler applies the categories suggested by Frawley (1992) regarding the deictic classification of the epistemic area, so that four subcategories emerge (cf. also Wachtmeister Bermúdez 2005: 25-26): 1. coming from ego (inference), 2. going toward ego (visual perception scale, auditory, other senses), 3. coming from alter (quote, report, rumour…), and 4. going to alter (one to all possible conversation partners) (Haßler 2010d: 102). Her consideration of the conversation partners shows that she works with a dialogically enriched notion of deixis. Evidentiality is thus not merely a reference to the source of information reported in an utterance, but also a deictic phenomenon which refers to the speaker and his/her complex relationships to information and sources (Haßler 2010d: 104).

Just as Haßler (2010d) relies on Frawley’s study (1992), Wachtmeister Bermúdez (2005) highlights the importance of Frawley’s ‘deictic access’ to evidentiality. Furthermore, recognising the deictic character is considered helpful in distinguishing evidentiality and epistemic modality (cf. also Haßler 2010d: 102; Haßler 2010a): Sin embargo, a pesar de que en su trabajo incluye la evidencialidad dentro de la modalidad epistémica, la puntualización que Frawley (1992) hace es importante cuando se trata de resolver justamente la tensión entre modalidad epistémica y evidencialidad. Reconocer el carácter deíctico de la evidencialidad y tratar de describirla en términos deícticos ayuda a entender el lugar de la evidencialidad en la configuración del lenguaje y a sentar las bases para distinguir evidencialidad de modalidad epistémica (Wachtmeister Bermúdez 2005: 27).

If evidentiality is categorised in terms of deixis it should be recognised that evidentiality is not a ‘prototypical deictic category’. If evidentiality is described as being a deictic phenomenon it should be highlighted that deixis must be understood in a non-canonical way. And so explains Mushin (2001) as well: […] evidentiality is not a prototypical deictic category. The semantics of prototypical deictic markers like personal pronouns I and you, can only be understood in terms of indexicality. That is, they have no meaning outside of their function to point out the current speaker and addressee. Evidential markers clearly have some semantic content independent of their indexical function – the categories of source of information and their associated meanings of epistemological assessment […] (Mushin 2001: 33-34).

52 To sum up, if evidentiality should nevertheless be called a deictic phenomenon it would belong to the kind of deixis that is (very) ‘impure’ because the semantics of evidential expressions are always partly non-deictic: [...] puede también hablarse de “deixis pura” y “deixis impura”. Los “deícticos puros” son aquellos cuyo significado es de naturaleza exclusivamente deíctica: los pronombres personales yo y tú, por ejemplo, se refieren al hablante y al interlocutor sin añadir ningún otro significado. El significado de los “deícticos impuros”, en cambio, es en parte no deíctico, como es el caso de los pronombres personales él, ella y ello, que a su significado deíctico añaden el significado no deíctico de género (Eguren 1999: 935).33

Lyons (1995) also differentiates between pure and impure deixis. Expressions “whose meaning can be accounted for fully in terms of the notion of deixis” belong to the former kind of deixis, while “expressions whose meaning is partly deictic and partly non-deictic” belong to the latter kind of deixis (Lyons 1995: 307). He also denotes the third-person singular pronouns he, she and it as deictically impure because they semantically encode the feature ‘masculine’, ‘feminine’ and ‘neuter’, respectively (Lyons 1995: 307). Eguren (1999), furthermore, ‘warns’ that a linguistic item which expresses subjectivity in some way should not automatically be called deictic because if a speaker’s subjectivity is expressed no kind of reference has occurred so far (cf. Cappelli’s (2007) line of argumentation in chapter 2.1.3). But for a linguistic element to be deictic, both criteria have to combine: a deictic element has to originate in an origo to be egocentric and it should refer to something: [...] tampoco basta con que exista una orientación egocéntrica para que pueda hablarse de deixis: fenómenos lingüísticos que expresan de una u otra manera la subjetividad del hablante, como los modos verbales [...], no pueden considerarse deícticos en sentido estricto, dado que no son formas de referencia. Ambas propiedades – referencialidad y egocentricidad – han de combinarse, en suma, para poder considerar una unidad o una expresión como “deíctica” (Eguren 1999: 933).

33

Bühler also calls pronouns a ‘semantically mixed class’ as they are half-deictic and halfnon-deictic: [...] es war mir eine Ermutigung nachträglich zu finden, daß [die Unterscheidung von Zeig- und Nennwörtern] von den ersten griechischen Grammatikern genau so und an derselben Stelle wie es mir notwendig erschien, bereits gezogen worden war. Später kam eine gewisse Verdunkelung und Verwischung auf durch die Dominanz des Interesses an der Mischklasse der Pronomina; niemand wird ihre Existenz bestreiten, aber daß sie semantische Mischlinge sind, den Nachweis müssen sie sich gefallen lassen (Bühler 1999: 81-82).

53 However, later it will be shown that the assumption of evidentiality being a deictic category is supported by various linguists. The deictic nature of evidentiality is, for instance, also addressed by Diewald/Smirnova (2010a) and Leiss’ studies (2008, 2009) are also concerned with the deictic description of the use of modal verbs and adverbs (cf. chapters 2.1.2.1 and 2.2). It will furthermore be shown that it is not only evidentiality itself that can be considered a deictic phenomenon but also the categories that are related to it, which again underlines that a superordinate category should not only be found for evidentiality and epistemic modality alone.

2.1.2.1 On the deictic reading of modal verbs A particular use of modal verbs leads Diewald (1999; cf. also 2000) to term modal verbs as deictic signs, whereby she describes the modal verbs’ newly acquired ‘deicticity’ as a result of grammaticalisation (cf. Diewald 1999: 170). Diewald differentiates between two central uses of modal verbs, which are described as two distinct subsystems: Bei den beiden zentralen Gebrauchsweisen handelt es sich [...] um zwei distinkte Subsysteme mit unterschiedlichem Grammatikalisierungsgrad und unterschiedlicher Funktion, deren lexematische Einheit (also die Tatsache, daß je ein Modalverblexem beide Funktionen realisiert) durch die Existenz einer gemeinsamen abstrakten Bedeutungsschablone begründet ist (Diewald 1999: 5).

These two uses of modal verbs differ in respect of function and degree of grammaticalisation and thus represent two subsystems. Nevertheless, the two uses are related through the semantic structure they share. The feature that distinguishes the one use from the other is ‘deicticity’ (cf. Diewald 1999: 14): [D]er nichtdeiktische Gebrauch stellt (typischerweise) einen Zustand des Satzsubjekts dar, wobei das Modalverb Bestandteil der dargestellten Szene ist; es fungiert als charakterisierenden Zeichen, als “Nennwort” i[m] Sinne Bühlers [...]. Im deiktischen Gebrauch dagegen bringt das Modalverb eine sprecherbasierte Faktizitätsbewertung zum Ausdruck; es ist nicht Bestandteil des dargestellten Sachverhalts, sondern repräsentiert, wie es für die Zeichenklasse der Deiktika (“Zeigwörter”) typisch ist, die Beziehung zwischen dem sprachlich Dargestellten und dem Sprecher, der deiktischen Origo [...] (Diewald 1999: 14).

If the modal verb is used deictically, the denoted object is ‘localised’ and related to the origo, from where the object has been denoted. This relational structure between the speaker’s origo and the event – or in Diewald’s terms ‘denoted object’ – is part of the semantics of a deictic sign (cf. Diewald 1999: 15-16). The relational structure holds “between a modal source, path, and goal” (Diewald 2000: 23; cf. also Diewald 1999: 27-28). Hence, the relational structure of the

54 grammaticalised modal verbs represents features of a deictic process: the speaker’s origo is – through the use of the modal verb – the origin of the evaluated event and the proposition is the goal (Diewald 1999: 206).

Origo

MODAL-DEICTIC VALUE

Proposition

Fig. 2: The relational structure of grammaticalised modal verbs (cf. Diewald 1999: 170)

That means that an utterance which contains a deictically used modal verb can only be understood correctly if reference to the speaker’s origo is given. If the modal verb is used deictically, the modal source is obligatorily represented by the speaker, the deictic centre, the origo. Thus, the deictic use always indicates a speaker-based factuality judgement of the proposition as a whole (cf. Diewald 1999: 46). And the degree of factuality is represented by the deictic value: […] the deictic value in the modal dimension is the degree of factuality of the proposition as it is judged by the speaker. The point of reference for the degree of factuality of the proposition is the degree of factuality that the speaker attributes to himself/herself (Diewald/Smirnova 2010a: 83).

The non-deictic use, in contrast, is characterised by having the origin in an authority that is part of the proposition, while the deictic use of the modal verbs is characterised by having the source of origin beyond the proposition. In this case, it is the speaker’s origo where the factuality judgement of the proposition comes from (cf. Diewald 1999: 19). To illustrate the non-deictic as well as the deictic use of modal verbs, Diewald (1999: 13, 16) offers the following examples from German: (a) Durch ihr mutiges Verhalten konnte eine 69jährige Frau einen jugendlichen Räuber in die Flucht schlagen. Æ The woman was able to do this. (b) Ich kann mich getäuscht haben. Æ I have possibly deluded myself. (c) So meinte sie, man müsse seinen Kindern vermitteln, dass sie auch “nein” sagen dürfen. Æ to be allowed to say ‘no’. (d) In den Wintermonaten Januar/Februar allerdings dürfte die Kurve – wie immer in der kalten Jahreszeit – einmal noch weiter nach oben gehen. Æ The graph goes probably further up.

Usually, the use of the modal verb können in (a) is described as deontic, and in (b) as epistemic. Diewald calls these two uses “lexical, less grammaticalised (non-deictic)” and “grammatical, grammaticalised (deictic)”, respectively (cf.

55 Diewald 1999: 13). Consequently, the modal verb’s scope is different: the less grammaticalised use represents a narrow-scope use, while the more grammaticalised use represents a wide-scope one (Diewald 2000: 23). In sentence (a) the state of being able to do something refers to the subject of the proposition, to the 69-year old woman. In sentence (b), in contrast, the modal verb’s reference point is the speaker’s origo which is not be found within the proposition since ‘I’ is the speaker who utters ‘I have possibly deluded myself’. The sentence pair (c) and (d) are to be analysed in an analogous manner: in sentence (c) the use of dürfen is non-deictic, lexical and has the meaning ‘to have permission to do something’, while in sentence (d) the same modal verb is used to utter a factuality judgement, that is, the use is deictic (cf. Diewald 1999: 16).34 Such an evaluation always originates in the speaker’s origo, whereby the proposition that is to be evaluated represents the goal. As Diewald describes the factuality judgement as a speaker’s subjective judgement of an event, the presence of an origo as reference point is indispensable. Roughly speaking, the speaker has the choice whether to term a certain state of affairs as real or nonreal: Mit der Bewertung eines Sachverhalts als faktisch sagt der Sprecher “wahr von mir aus gesehen” und nicht “absolut – nach den Maßstäben der Logik oder vorliegenden allgemeinen Evidenzen – wahr”. Mit der Beurteilung eines Sachverhalts als in irgendeiner Weise nichtfaktisch dagegen sagt der Sprecher “von mir aus gesehen nicht als wahr einschätzbar” und nicht “absolut – nach Maßstäben der Logik oder vorliegenden allgemeinen Evidenzen – nicht wahr” (Diewald 1999: 174-175).

No matter how the speaker evaluates a certain state of affairs under question, the speaker is the source of this process of evaluation and represents the centre of perspectivisation (cf. Diewald 1999: 211). The last point of Diewald’s line of argumentation that should be addressed is a little problematic. Her justification of the semantic template she applies to the deictic use of modal verbs is, in my opinion, too one-dimensional: Betrachtet man dagegen den deiktischen Gebrauch als Ausdruck eines von den Umständen ausgehenden Zwanges auf den Sprecher, der daraufhin zu einer Faktizitätsbewertung veranlaßt wird [...], dann ist der Sprecher ja als das Ziel (nicht als der Ausgangspunkt) der ihn zwingenden Umstände konzipiert. […] Dies würde eine völlig andere relationale Struktur voraussetzen, und zwar eine, in der die Faktizitätsbewertung des Sprechers eingebettet ist in eine übergeordnete Relation, die den Einfluß der Umstände auf den Sprecher darstellt. Eine solche Struktur ließe sich 34

One can also differentiate between the deictic and non-deictic use of modal verbs at the syntactic level: the non-deictic, less grammaticalised use is, according to Diewald (1999: 46), very similar to the use of main verbs, while the deictic, grammaticalised use behave like modal auxiliaries.

56 nicht aus der Semantik der Modalverben ableiten […] Die deiktische Verwendung ist von der objektiv epistemischen durch die Spezifizierung der modalen Quelle als Origo geschieden, d.h. daß der deiktische Gebrauch, anders als häufig angenommen, keinen Bezug zu Umständen, Beweisen, Indizien usw. herstellt (obwohl solche natürlich vorhanden sein können) (Diewald 1999: 214).

In summary, Diewald denies any relation between the speaker and circumstances that may cause the speaker to utter a factuality judgement. In this case, the speaker would represent the goal and not the proposition that is subjectively evaluated. This would obviously require a totally different relational structure: in addition to the semantic template [origo Æ factuality judgement Æ proposition], a superordinate relation that represents the influence of circumstances that cause the speaker to subjectively evaluate an event should then be applied as well. But such a structure cannot be derivated from the modal verbs’ semantics and consequently, according to Diewald, the deictic use is not related to circumstances, evidences, etc. This raises the following questions: is not every uttered factuality judgement of a state of affairs the result of circumstances? Are not the (extra-)linguistic circumstances always the reason to cause exactly this evaluation – and not another one? It surely depends on the fact whether the circumstances are linguistically encoded or not and in analysing language data, one can only rely on what is contextually provided. But arguing from a logical point of view, I doubt that a speaker expresses a factuality judgement without feeling compelled by any circumstances to do so. To take up example (d) again: the speaker who utters ‘The temperature graph – as always in the cold season – goes probably further up’ surely has a reason to suppose this. There must be circumstances that lead him to utter this assumption. This use of dürfen in sentence (d) in particular should be described as more evidential/inferential than modal, which is underlined by the parenthesis. The modal verb is definitely used inferentially; it is a case where knowledge “derived from tradition or common knowledge” (Lazard 2001: 365) is expressed: in the cold season, it is usually the case that the temperature graph goes further up. The consistent relational semantic template could nevertheless be perfectly applied. In Evidentiality in German, Diewald/Smirnova (2010a: 86) say that modal verbs “may be contextually associated with any type of information source”. This is what happens, in my view, to example (d), where the parenthesis “wie immer in der kalten Jahreszeit” shows that the use of the modal verb dürfen is inferential. Example sentence (e) in Evidentiality in German – opposed to (f) to illustrate the difference between epistemically used and evidentially used modals or auxiliaries – is much more appropriate to illustrate the epistemic use of dürfen than the sample sentence (d): (e) Die Lieferung dürfte/könnte/mag größer sein als erwartet.

57 ‘The delivery might/could/may be larger than expected.’ (f) Die Lieferung scheint/droht/verspricht größer zu sein als erwartet. ‘The delivery seems/threatens/promises to be larger than expected.’ (Diewald/Smirnova 2010a: 86).

As there is no further contextual information provided that might lead to an inferential reading of dürfen, example (e) is considered more appropriate than sample sentence (d) to illustrate the epistemic use of this modal verb. Actually, sentence (e) and (f) are used to distinguish epistemic modality from evidentiality: the auxiliaries in (f) show that the “origo has an information source for P”, while the origo in (e) simply “attributes an uncertain degree of factuality to P” (Diewald/Smirnova 2010a: 86). Examples (d) and (e) illustrate very well that even modals like dürfen, which are through and through epistemic, may be inferentially used, although “they do not contain this feature as part of their semantic structure” (Diewald/Smirnova 2010a: 86). But their evidential reading may be contributed by contextually provided information. Hence, the classification of their use depends on contextually (non-)provided information. And in example (d) the contextually provided information or rather the proof for the modal verb being used to express inference, is even to be found at the sentence level itself. As Diewald/Smirnova opt for a clear distinction between evidentials (evidential expressions) and modals, they explain that – if overlaps arise – they are “the results of conversational implicatures and not inherent semantic components of these verbs” (Diewald Smirnova 2010a: 80; emphasis in the original). I agree with the fact that there is ‘nothing evidential’ inherent in the semantics of dürfen. I deny that if a certain linguistic item shows a particular meaning aspect which is not inherent in its semantics it must automatically be the result of conversational implicatures. But Diewald/Smirnova explain: […] the evidential meaning of German epistemic modal expressions is assumed to arise via pragmatic strengthening of particular conversational implicatures. The speaker marks the epistemic modal status of the proposition characterizing it as more or less sure, i.e. indicating his/her degree of commitment to the truth of the proposition. In this case, on the hearer’s side, the implicature may easily arise that the speaker’s attitude towards the proposition (and the proposition itself) is based upon some evidence available to him/her (Diewald/Smirnova 2010a: 93).

This may be the case. But it often simply is contextually provided information which shows that a certain modal expression is evidentially used. Then, it is not even conversationally implied but obvious. So there is a lot of room between a meaning aspect being encoded by a lexical element and a meaning aspect being due to conversational implicatures. The meaning aspect may simply be provided by contextually available information.

58 It should be kept in mind that a speaker never utters a factuality judgement of an event ‘without being caused by any circumstances’. The communicative situation, the (extra-)linguistic context, always provides some information that may have caused the speaker to express the factuality judgement concerning the described event. The semantic template for the deictic use of modal verbs can nevertheless be perfectly applied. For the use of evidentials Diewald/Smirnova (2010a) admit a kind of second relational structure. With regard to the development of evidentials they explain that it “is highly deviant from the development of the modals” (Diewald/Smirnova 2010: 90) because it implies the reinterpretation of a “cause” of movement or effect into an evidential source causing the speaker to make a statement, i.e. the evidential source is not the deictic origo (as opposed to the modal source in the deictic use of the modals). This process is called here the relocation of the source entity (Diewald/Smirnova 2010a: 90; emphasis in the original).

And then they outline: [This] feature points to the most important fact that the evidential source is conceived of as distinct from the speaker (but co-present with the speaker). By using the evidential the speaker points to this source, i.e. s/he locates the source and only from there makes a statement concerning the (primary) proposition. In other words: in evidentials the speaker points to the existence of the evidential source and makes the statement on the basis of the factuality of this evidential source (Diewald/Smirnova 2010a: 90; emphasis in the original; cf. also the ‘deictic line of argumentation’ by Haßler 2010a, 2010d and Wachtmeister Bermúdez 2005).

Consequently, there are four components that have to be taken into account when developing a model that captures the deictic character of evidentials. While the deictic relational structure for modals contains three components (1. origo, 2. path and 3. goal – or 1. origo, 2. relation between origo and entity, and 3. described entity) (cf. Diewald/Smirnova 2010a: 11), the evidential relation between origo and described entity is slightly more complex (cf. Diewald/Smirnova 2010a: 53) […] because the speaker sees the primary event through the secondary event, i.e. the evidence can be equalled to a pair of glasses through which the primary event is perceived. […] the defining crucial feature of evidentials is that […] the existence of a pair of glasses, and even the specification of what kind of glasses [that is, the kind of evidence] this is, is the central semantic feature of evidentials (Diewald/Smirnova 2010a: 53).

The above mentioned secondary event, the evidence, is situated between origo and the primary event which represents the described entity and “which is logically situated between origo and primary event, as a third point in the vectored relation” (Diewald/Smirnova 2010a: 54). The additional arrow that points back

59 from the secondary event to the origo illustrates a reflexive relation between these two components, and it depends on the ‘glasses’ how the origo perceives/perceived the primary event (cf. Diewald/Smirnova 2010a: 54):

Origo

Secondary event functioning as evidence

evidential dimension

Primary event

Fig. 3: The relation between origo, secondary event and primary event (cf. Diewald/Smirnova 2010a: 53)

Leiss (2008, 2009) regards epistemic/evidential adverbs and epistemic/evidential modals as deictic but differentiates between two kinds of deixis: she describes the former as simple deictic and the latter as double deictic35 (Leiss 2009: 13) or as a kind of complex deixis (Leiss 2008: 34). So according to Leiss epistemic/evidential modal adverbs “lack the feature of double displacement” (Leiss 2008: 15): [t]here is […] an essential difference between modal adverbs and epistemic modals. Modal adverbs, such as probably or certainly are less complex in terms of deixis. They are simple shifters […]. They are in sharp contrast to epistemic modals where an additional deictic function is involved […]. Epistemic modals are clearly more complex. Complex deixis means that, additionally to shifting, there is some sort of additional displacement of the viewpoint of the speaker involved. In contrast to Diewald’s (1999) terminology we refer to epistemic modals as complex shifters which are clearly invested with more deictic force than epistemic adverbs (Leiss 2008: 34).

That is why adverbs and modals are not considered functionally equivalent (cf. Leiss 2008: 15, 34), although both adverbs and modals “signal the speaker’s assessment of the probability of the reported event” (Leiss 2008: 35). Leiss criticises that this shared feature seems to be sufficient for many translators to “rely on epistemic adverbs when they have to translate epistemic modals into a language which has a restricted system of modal verbs” (Leiss 2008: 35). Consequently, Leiss does not accept the common practice to translate German wollen by perhaps, sollen by probably and müssen by certainly (cf. Leiss 2008: 35). 35

Leiss’ terms of complex deixis (2008: 34) and double deixis (2009: 13) should not be confused with Volkmann’s term of doppelte Deixis (2005: 232-234; cf. also Volkmann 2009) and Reyes’ term of doble deixis (1990: 17), as will be shown in chapter 3.2.

60 And she is right when she explains that modals, if translated as adverbs lose their feature of ‘information source’: While it is true that epistemic modals and epistemic adverbs are functionally equivalent insofar as an assessment of the reliability of information is commented on by the speaker, only epistemic modals such as German wollen, sollen, and müssen can additionally encode the source of information on which the speaker’s judgement relies on (Leiss 2008: 35).

The following examples illustrate that German modals indeed encode the information source besides the speaker’s attitude towards [p]: (a) Anja will den ganzen Tag am Computer verbracht haben. (b) Anja muss den ganzen Tag am Computer verbracht haben. (c) Anja soll den ganzen Tag am Computer verbracht haben (cf. Leiss 2008: 35).

When using wollen, the speaker indicates, on the one hand, that the information source is Anja herself and, on the other hand, that he does not totally rely on [p]. The evidential subcategory thus involved here is quotation. In example (b) the speaker indicates with müssen that he expresses an inference and “the source on which the speaker relies upon is the speaker himself” (Leiss 2008: 36). In example (c) the speaker indicates that he got to know from another person that Anja had spent the whole day in front of the computer (cf. Leiss 2008: 35-36). While Leiss (2008: 36) explains that the use of sollen “refers to a person close to the speaker, being close enough to witness the event of which truth conditions are to be evaluated,” it should be mentioned that sollen is also used to indicate hearsay, i.e. when the speaker obtained the information without being able or without wanting to specify at the moment of speaking who exactly told him.36 However, it is obvious that epistemic modals (may) encode both, epistemic judgement and source of information. In the examples above the information source is quotation/hearsay and inference but the degree of probability that [p] is the case varies from example to example. With wollen the speaker indicates that the source (Anja) is not really reliable. Sollen stands for an intermediate degree of probability, due to the fact that the source of information “refers to a person who is neither the subject of the proposition nor the speaker himself” (Leiss 2008: 36). And with müssen [p] is assigned the highest probability without being absolutely sure. Because of the fact that the speaker’s attitude towards the proposition and

36

With reference to sollen and wollen, Maché (2009: 41) explicitly speaks about a quotative reading of the German modals. According to him ‘an epistemic reading in its narrow sense’ is not present (cf. Maché 2009: 41). As indicated above, I would disassociate myself from the claim that there is nothing epistemic conveyed by sollen and wollen.

61 the information source are present in the German modals, Leiss (2008: 37) speaks about a “double displacement”: First, the speaker distances herself from the proposition in judging it (this being true for both epistemic modals and epistemic adverbs), and second, the speaker is presented as distant from the source of evidence, this being true only for epistemic modal verbs (Leiss 2008: 37).

The process of double displacement is said to be even applicable to cases in which the speaker quotes himself. The speaker is then seen as “splitting himself up into the source of judgement and into the speaker on whom the judgement relies” (Leiss 2008: 37). The three German modals in their use mentioned above have to be seen as a special case where epistemic modality and evidentiality go hand in hand (in equal shares). Therefore, Leiss (2008: 37) concludes “that epistemic modals are to be classified as pertaining to the category of evidentials” as, according to her, “[e]videntials involve both the source of information and the speaker’s judgement [!]” (Leiss 2008: 38).37 Consequently, both German evidentials and epistemic modals are said to share the feature of double displacement. While epistemic modals are described as involving both domains, epistemic modality as well as evidentiality, modal adverbs are said to only convey the speaker’s attitude and evidential adverbs to indicate the information source (cf. Leiss 2009: 12). Analysing German, she classes vielleicht ‘maybe, perhaps’, wahrscheinlich ‘probably’ and möglicherweise ‘possibly’ among the modal adverbs, and offensichtlich ‘obviously, evidently’, scheinbar ‘apparently’ and angeblich ‘allgegedly, reportedly, supposedly’ among the evidential adverbs. No adverb encodes, according to Leiss, the speaker’s attitude as well as information source (Leiss 2009: 12). At this point I would like to take up example (b) again to prove her statement. As sollen is also used to convey a hearsay reading, one might argue that it therefore could be translated as allegedly, supposedly or reportedly. Indeed, then the modal component was gone. Translating sollen as probably – as criticised by Leiss – the evidential component was gone. So concerning the translation of the German modals Leiss is right in stating that no adverb encodes both the speaker’s attitude and the information source. At this, ‘encodes’ has to be understood in the sense of primary encoding. If the use of the German adverbs is considered, epistemic modality and evidentiality may also overlap. In Spanish, the fact that a use of an epistemic/evidential adverb 37

Even though Kosta (2011: 258) makes a clear-cut distinction between modalité épistémique and évidentialité, the aim of his study is to explain why evidentiality and epistemic modality often coincide in one and the same lexeme, more precisely, in one and the same adverb (cf. Kosta 2011: 271).

62 may convey the other reading should also not be excluded. Contextual information may provide (at least additional) epistemic/evidential ‘nuances’. The Spanish modal adverb probablemente, for instance, is commonly known as an epistemic adverb and its conceptual meaning is undoubtedly modal. But the two following examples show that it may be evidentially used, namely to express an inference, which is underpinned by contextually provided information: (10) “La cultura había tomado una dirección tan equivocada que creíamos que todo tenía que ser repensado. No ha pasado lo mismo con las guerras de Afganistán e Irak, probablemente porque el reclutamiento no es obligatorio”. (El País 02/08/2010) (11) Yo misma nunca pude sustraerme al impulso de agacharme sobre una doble página de diario puesta en el suelo recién fregado para leer unos titulares tal vez ya del día anterior. Si ustedes sienten esa misma pasión, probablemente dedicarán muchas horas de las vacaciones a leer, por lo que no me resisto a recomendarles unos cuantos títulos. (El País 02/08/2010)

In example (10) the speaker indicates why ‘probably [p]’. He explains that ‘the same has not happened in the wars of Afghanistan and Iraq’ and then adds ‘probably because the recruitment is not obligatory’, thereby indicating the result of an inferential process, namely the (probable) reason for [p]. In example (11) the evidential use of probablemente is confirmed by the inferential use of the future form (dedicarán): if the readership has the same passion, it ‘will probably dedicate (lit.)/certainly dedicates many hours to reading’. Thus, the speaker expresses a ‘conditioned near-necessity’ such as ‘if X then certainly Y’. So one should at least speak about ‘evidential nuances’ which contextually provided information may apply to adverbs. Furthermore it is suggested that such nuances may also be conveyed by adverbs of other languages. Nevertheless, the meaning of modals and adverbs represents a particular problem in translation science. Regarding the double displacement, however, Leiss strictly distinguishes between adverbs – being modal or evidential – and modals as well as evidentials by applying the features Sprechereinschätzung ‘+/– speaker’s attitude’ and Quelle ‘+/– information source’ so that the following matrix emerges: Table 3: Kind of deixis, speaker’s attitude and information source in adverbs, modals and evidentials (cf. Leiss 2009: 13; slightly adapted) LINGUISTIC ITEM epistemic adverbs evidential adverbs epistemic modals evidentials

KIND OF DEIXIS simple deixis simple deixis double deixis double deixis

SPEAKER’S ATTITUDE + – + +

INFORMATION SOURCE – + + +

63 I wonder then where the difference between epistemic modal verbs and evidentials is, or – asking in another way and adopting Leiss’ point of view – why at all differentiate between evidentials and epistemic modals? As both are said to have the same features, both encoding speaker’s attitude as well as source of information, are they not the same according to Leiss’ theory of double displacement? While epistemic adverbs can be distinguished from evidential adverbs according to their distinct features ‘speaker’s attitude’ and ‘information source’, the matrix for epistemic modals and evidentials is equivalent. The line of argumentation concerning the three German modals wollen, sollen and müssen can be followed very well but the fact that evidentials have the same features as modals applied to them is the result of defining evidentials as involving “both the source of information and the speaker’s judgement” (Leiss 2008: 38). If at all, the feature of ‘involving both the source of information and the speaker’s judgement’ may be applied to certain expressions of languages that lack a grammatical system of evidentiality, to expressions that encode both domains – sometimes even conveying them simultaneously. The German modals wollen, sollen and müssen, for instance, belong to exactly those expressions. Anyway, Leiss’ studies (2008, 2009) implicate one important message: epistemic/evidential adverbs and modals are deictic – be it simple or double (Leiss 2009) or complex (Leiss 2008). To my way of thinking, even simple deixis is complex enough.

2.1.2.2 Where to put inference? By describing evidentiality in terms of deixis, the evidential subcategory of inference represents a major point for discussion. It even represents a disputatious topic for scholars like van der Auwera/Plungian (1998) and Cornillie (2009), who discuss whether inference is ‘more modal’ or ‘more evidential’. In “Modality’s semantic map” van der Auwera/Plungian represent “crosslinguistically relevant synchronic and diachronic connections between modal, premodal, and post-modal meanings or uses” (1998: 79). In doing so, they claim to pay special attention “to meanings that are vague between possibility and necessity” (1998: 79). Nevertheless, they explain that the inferential reading amounts to epistemic modality and more particularly necessity: for both categories we are dealing with the certainty of a judgment relative to other judgments. From this point of view it also causes no surprise that inferential evidentials often receive an English translation with epistemic must. Inferential evidentiality is thus regarded as an overlap category between modality and evidentiality (van der Auwera/Plungian 1998: 85-86).

64 Even though it indeed makes sense to regard inference as the evidential subcategory where (epistemic) modality and evidentiality overlap, it is not very prudent to analyse every inferential reading as amounting to necessity. It depends on the evidence that is the basis for the conclusion, and hence [p] expressed as an inference may be possible, probable or necessary (or even something in between as highly probable or almost necessary). The following two examples, for instance, contain the synthetic future form será, which is used to express a strong inference because it is accompanied by the noun phrase sin duda ‘no doubt’ (cf. also Volkmann 2005: 299). The inference expressed is thus more necessary than only possible: (12) [...] afirma José María Cosculluela, presidente de Vitalia Home, una empresa con geriátricos en varias regiones. “El sector debe enfrentar una gran reconversión en un plazo de unos cinco años”, vaticina. Así será, sin duda, cuando las comunidades vayan completando sus decretos sobre los requisitos que han de cumplir estos centros para poder recibir a ancianos […] (El País 30/09/2010) (13) En 2001 recibió la máxima distinción del jazz francés, el Django de Oro en la categoría de Nuevo Artista. Su pianismo lee con personalidad las mejores páginas del jazz americano, a las que suma palabras clásicas y europeas. Sin duda, será una de las revelaciones del festival. (El Mundo 31/10/2010)

The following instance of será, in contrast, is used with the verbal phrase creo que ‘I believe/think that’. In comparison to the former examples, this inference seems more possible than (logically) necessary. According to the journalist, the football player is nevertheless ‘convinced’ (está convencido) that the current season ‘may’ (puede) be better, that is, the speaker does actually not express an inference that [p] necessarily is the case. (14) Seydou Keita: “Creo que será un buen año”. El jugador del FC Barcelona Seydou Keita está convencido que este año el Barça puede hacer una muy buena temporada. (El Mundo 02/09/2010)

Cornillie also takes a stance against van der Auwera/Plungian’s proposal because it assumes that all inferential statements are concerned with certainty of a judgment. That is, inference is seen as a monolithic concept that lacks any variation. […] Moreover, the proposed correlation between inference and necessity implies that other modes of knowing, such as quotative (or hearsay) evidentiality, are automatically outside the realm of necessity. This implicitly correlates them with weaker forms of modality such as possibility and probability (Cornillie 2009: 49).

Cornillie (2009) is obviously right when rejecting the assumption that inferential statements should generally be concerned with certainty. The modal verbs poder and deber, for instance, can be used to express inference, whereby the former is

65 normally used to express a tentative conclusion, the latter to indicate a stronger one. Diewald/Smirnova (2010a: 92) also “assume that the sub-domains of epistemic necessity and inferential evidentiality can and must be distinguished”. They provide two reasons: (a) not every inferential statement needs to have an epistemic modalization; (b) not every statement expressing epistemic necessity needs to represent the result of an inferencing process (Diewald/Smirnova 2010a: 92).

However, this is not always as simple as it sounds here. As indicated above, the assumption that inference automatically amounts to epistemic necessity should be rejected. Spanish poder and deber can both be used to express inference but the modal values going hand in hand with an inference expressed by poder and deber are, roughly speaking, concerned with lower certainty and higher certainty, respectively. That means there is ‘something modal about their use’. Therefore, the correlation between inference and necessity in particular should only be rejected, but not – as Diewald/Smirnova formulate with their two reasons – between inference and epistemic modality at all. The discussion about ‘where to put inference?’ is also addressed by other scholars such as Squartini (2001): [...] inferential markers, which, even if often consistently treated as evidentials[,] are described as a case of interaction between epistemic modality and evidentiality by Van der Auwera & Plungian (1998: 86). Their intermediate status is also recognized by Palmer [...], who treats inferentials under the heading judgements when languages such as English or German are concerned [...], while mentioning them in the section devoted to evidentials in a language such as Tuyuca (Squartini 2001: 297298).

Gómez Rendón (2006: 49) says in this connection: “[…] an interpersonal approach makes clear that inference in E[cuadorian] Q[uechua] may be read not only as a cognitive exercise but also as a commitment of the speaker to the truth of a propositional content.” This is easy to understand since speakers always like to believe in their own conclusions. Nevertheless, Palmer seems to go the right way as he argues that the place of inference is language specific. Statements about inference which lay claim to universal validity should generally be avoided. Volkmann (2005) explains the special status of inference as follows: Schlussfolgerungen fallen zum einen in den Bereich der Evidentialität, weil sie eine Form des Zugangs zur Information darstellen, zum anderen fallen sie in den Bereich der epistemischen Modalität, weil sie mehr als die anderen Arten von Informationsquellen Wissen bereits voraussetzen (2005: 120).

As Volkmann (2005) explains, on the one hand evidentiality offers a place for inference because it represents one kind of access to information, and conse-

66 quently represents one information source. On the other hand, inference is the evidential subcategory that relies more than any other subcategory on knowledge, which obviously lets it appear ‘more modal’ or ‘more epistemic’. Boye (2006) also describes expressions that indicate an assumption as (possibly) being partly epistemic, partly inferential: […] the ‘assumption’ may be taken to indicate a certain range of degree of epistemic support – assumption in contrast to e.g. certainty – but it may also be taken to indicate a type of evidence – assumption in contrast to e.g. reportive evidence […]. Thus, in lack of further information, items and constructions characterized with the term ‘assumed’ or ‘assumptive’ may be taken to encode the meaning of indirectinferential evidence, the meaning of e.g. partial epistemic support, or both these meanings […] (Boye 2006: 66).

But if evidentiality is assumed to be a deictic category, the discussion becomes even more interesting – especially if dealing with evidential expressions in languages that do not possess a grammatical system of evidentiality (cf. chapter 2.1.2). De Haan (2001b, 2005) has argued that evidentials are used to denote the relative distance between the state of affairs and the speaker talking about that event. Inferentials, he says, are used “for those instances in which the speaker has not witnessed the action personally, but has witnessed evidential traces of that action” (de Haan 2005: 16). The fact that he here forgets that inferences may also be drawn on the basis of knowledge will later be taken up again. De Haan proposes to see the subcategory of inferentiality as a ‘hybrid’ category: Although the inferential is usually grouped with the quotative to form the category of indirect evidentiality (see e.g. Willett 1988 and Palmer 2001), it is in fact a hybrid direct/indirect evidential category, because the speaker is aware of the evidence for the action (de Haan 2005: 16)

He argues for accepting inferentiality as a hybrid category because, according to him, inferentials are ambiguous with respect to the speaker’s role: On the one hand, they have a certain level of personal involvement. When the speaker makes a statement based on visual evidence after the action occurred, he/she is involved at some level (de Haan 2001b: 38). On the other hand, the speaker has not witnessed personally the action he/she is describing, and as such inferentials are like quotatives (de Haan 2001b: 38-39).

In his opinion, by using a quotative, whereby the speaker introduces another person who becomes responsible for the information, “the speaker disavows any responsibility for the truthfulness of the information because the information in the statement does not come from personal witnessing […]” (de Haan 2001b: 38). In other words, the event did not happen/is not happening within the

67 speaker’s deictic sphere. This is truly the case for quotatives. The problem is, however, that the inferential process clearly takes place within the deictic sphere of the speaker. If he infers a certain state of affairs to be/have been the case he obviously relies on some indices to be found beyond his deictic sphere, and that is why it could make sense to speak about a hybrid category – but only if considering conclusions drawn on the basis of evidence. What about reasoning? Is there to be made a difference? If a speaker infers a certain state of affairs on the basis of his (world) knowledge, where no ‘evidential traces’ are involved, is this reasoning process then happening ‘totally’ within the deictic sphere of the speaker and consequently a kind of ‘direct evidentiality’ or is it ‘indirect evidentiality’ because there is no (real) event involved when concluding the occurrence of a certain event? Languages that have real evidentials have a solution for this problem. Nearly every matrix of evidentials shows which component is at work in a particular language: […] languages decide whether the deictic component or the witnessing component is the underlying factor that drives the evidential system. […] language[s] like Kashaya Pomo and Hualapai highlight the deictic component, represented by means of the feature direct […]. They do this by making the inferential responsible for part of the sensory information, usually the nonvisual, nonauditory ones. Other languages, such as Takelma and Sherpa, prefer to view the inferential as denoting an [sic] nonwitnessed event. In these languages, the feature firsthand [inferentials being non-firsthand] underlies the evidential system […]. The inferential evidential in such languages typically has a quotative […] interpretation as well (de Haan 2001b: 39).

So languages in which evidentiality is part of the grammatical system solve the problem of ‘hybridity’ themselves. It is not so easy in languages that do not possess real evidentials, such as Romance languages. This is a problem which Squartini (2001) comes across. He tests the different models proposed by Willett (1988) and Frawley (1992) that differently represent the internal structure of evidentiality. Squartini comes to the conclusion that Willett’s model is more appropriate to represent the internal structure of evidentiality in Romance languages, that is, the direct/indirect opposition: Willett’s model, based on the primary distinction direct vs. indirect type of evidence, is better suited to account for evidentiality in Romance, while the notion of source [in Frawley’s model] should be considered as independently interacting with the type of evidence/mode of knowing (Squartini 2001: 297).

In his paper “The internal structure of evidentiality in Romance” Squartini (2001) tests the appropriateness of Willett’s and Frawley’s model by analysing the evidential use of the synthetic future, the conditional and the imperfect of

68 different Romance languages. What Squartini nevertheless rejects is the hierarchical order of evidential values that is the basis for both models (cf. Squartini 2001: 298, 301). He pleas for a representation of source of evidence and type of evidence38 in a semantic map – such as in Anderson’s paper (1986: 284)39 – to illustrate the non-hierarchical interaction between source and type of evidence (cf. Squartini 2001: 301). Before continuing, and before commenting on Squartini’s analysis, we should first have a look at Willet’s and then at Frawley’s model of the internal classification of evidentiality.

DIRECT

TYPES OF EVIDENCE

ATTESTED

VISUAL AUDITORY OTHER SENSORY

REPORTED

SECOND-HAND THIRD-HAND FOLKLORE

HEARSAY

INDIRECT INFERRING

RESULTS REASONING

Fig. 4: Willett’s types of evidence (1988: 57)

The main differentiation that Willett makes is the one between direct and indirect evidentiality. A speaker uses a direct evidential if – according to de Haan (2005) – the described event has taken/is taking place within the origo of the speaker and if the speaker himself is able to attest (visually, auditorily or by other senses) the event he is talking about (‘Attested’). If the speaker was told about the state of affairs (‘Reported’) or if he had inferred it (‘Inferring’), he provides indirect evidence for the event. While within the category of ‘Reported’ it is differentiated between ‘Second-hand’, ‘Third-hand’ and ‘Folklore’, the category of ‘Inferring’ is split up into the subcategories ‘Results’ and ‘Reasoning’. Inferences that are drawn on the basis of perceivable evidence belong to the former, inferences that represent a “mental construct” (Willett 1988: 57) can 38

39

What is called ‘type of evidence’ here is a notion originally coming from Anderson (1986): his ‘type of evidence’ is equivalent to Chafe’s (1986: 263) ‘mode of knowing’. Both designate the way, or how, the speakers obtained the information. The term ‘source of evidence’ is interpreted differently. As we will see, Frawley (1992: 413) distinguishes between an internal source (“self”) and an external one (“other”) (cf. Squartini 2001: 302). Because of the complexity of Anderson’s semantic map (1986: 284) I consciously waive to reproduce the map here.

69 also be based on intuition, logic, dream or experience and consequently belong to the latter. This leads to the question of whether de Haan (2005) would name ‘Reasoning’ – together with ‘Results’ – as a hybrid category or whether he would argue that ‘Reasoning’ clearly happens only within the deictic sphere of the speaker but ‘Results’ within and beyond. Even though Willett’s model is based on a typological study of 38 languages with real evidentials, the main differentiation between direct and indirect evidence is rejected by Plungian (2001: 353). The reason is that he – like de Haan (2005) obviously – came across languages with real evidentials “which seem to prefer the parameter of speaker’s involvement” (2001: 353). That is why in his model the main differentiation is made between “personal evidence” and “mediated evidence”, whereby the former is split up into “direct evidence” (personal perception) and “reflected evidence” (inference). Direct evidence and reflective evidence are taken together as personal evidence because the speaker’s distance to the state of affairs is not as far as if the state of affairs was mediated (cf. also Volkmann 2005: 83).40 Furthermore, Plungian’s model shows well that inference represents a kind of personal evidence as well as indirect evidence, and that is what is also addressed by de Haan (2001b, 2005): Table 4: Plungian’s types of evidence (cf. Plungian 2001: 353) INDIRECT EVIDENCE REFLECTED EVIDENCE MEDIATED EVIDENCE (Inferentials and Presumptives) (Quotatives) PERSONAL EVIDENCE

DIRECT EVIDENCE

40

Volkmann (2005), as well as Plungian (2001), prefers to distinguish between the speaker’s reference to his own centre of consciousness and the speaker’s reference to another centre of consciousness. Thereby she subsumes sensory perception and inference below the former. Furthermore, she calls attention to the fact that even the (simple) perception of a certain state of affairs leads the speaker to cognitively handle the information which in turn leads to an interpretation of the perceived information: Sinneswahrnehmung bleibt nicht nur auf die bloße Aufnahme von Reizen beschränkt, sondern führt auch zur Interpretation des Reizes, Umwandlung in Wissen und dessen Weiterverarbeitung. [...] Der Übergang von Interpretation des Reizes, über Verarbeitung bis zur Informationsschaffung ist fließend. [...] Die semantische Nähe von Informationsaufnahme und Informationsschaffung und die Opposition zur Kategorie Informationsübernahme spiegelt sich auch in der Einteilung dieser Arbeit wider. Die perzeptive Evidentialität, d.h. die Markierung von Propositionen als vom [aktuellen Sprecher] selbst aufgenommen und weiterverarbeitet, spielt hier unter dem Aspekt Relativierung auf das eigene Bewusstsein eine Rolle (Volkmann 2005: 105).

70 So Plungian, in contrast to Willett, chose another starting point for his model of the internal structure of evidentiality, namely the criterion of speaker involvement. The weak point of Plungian’s model is that the kind of inference that Willett labels ‘Reasoning’ is neither regarded nor offered a place in the model. So Lazard’s remark on common knowledge is perfectly comprehensible: I would only suggest that it perhaps would be advisable to make a special place in the table for knowledge derived from tradition or common knowledge, which is not the same as knowledge derived from reported speech (hearsay). The former is normally part of what the speaker has practically always known, and, moreover, it is usually shared by the speaker and hearer. This is rather different from reported news. More importantly, there are languages where common knowledge is expressed by special evidential and clearly distinguished from hearsay (Lazard 2001: 365).41

For his model of the internal classification of evidentiality, Frawley (1992) uses several categories to deictically categorise the epistemic domain42: If we understand the requirements of epistemic modality by the categories of source/direction of knowledge and the scale of strength of knowledge, we can simplify and unify accounts of epistemic modality. To do so, we must first reexamine the basic deictic structure of modality, via deixis itself, which […] has three components: deictic points, direction, and remoteness (Frawley 1992: 412; emphasis in the original).

41

42

Große (2000), who investigates “Evidencialidade no português brasileiro”, also says in this connection that one may even speak about evidentiality “no âmbito dos provérbios, que representam de certo modo uma parte do conhecimento lingüístico de uma comunidade de fala” (Große 2000: 412) because “conhecimentos geralmente admitidos e transmitidos pela tradição” (Große 2000: 412) may represent an information source for a speaker. Frawley (1992) speaks about an epistemic standpoint of the speaker because he describes the speaker’s state of knowledge and especially his attitude towards his knowledge as a relationship of dependency to the “strength of knowledge”, which may be better described as ‘kind and strength of information’ (cf. Frawley 1992: 412).

71 SOURCE OF KNOWLEDGE

STRENGTH OF KNOWLEDGE

Self FROM

Scaled Categories of Inference necessary > possible

TO

Scaled Categories of Sensation visual > auditory > other senses > feel43 _____________________________________________________________ Other FROM

Scaled Categories of External Information quote > report > hearsay > other

TO

Scaled Categories of Participants other > all else

Fig. 5: Frawley’s deictic categorisation and scaling of epistemic modality (1992: 413)

Frawley understands the two kinds of information source (“source of knowledge”) as deictic points: “Sometimes the self is the centre of the epistemic stance (as in judgments) and sometimes the other (nonself) (as in hearsay)” (Frawley 1992: 412). If then the criterion of directionality is applied, which Frawley (1992: 412) calls “from the self, to the self, from the other, to the other”, four subcategories arise as a result: 1. inference (from the speaker); 2. sensory perception as scaled categories with hierarchical order (to the speaker, more precisely acting on the speaker); 3. mediated evidence, that is, quotation, hearsay etc. (from the other) and 4. one or more interlocutors (to the other(s)). Frawley (1992: 413) includes the third deictic component (“remoteness”) to capture the criterion of speaker involvement. He thereby describes epistemic modality and evidentiality as having a relationship of dependency: When we say that a speaker is less committed to a proposition, the evidence is less strong or direct, or the speaker is less confident, we mean that, just as in any deictic system, the distance between the reference point and located point is relatively large. Confidence, commitment, strength, and so on are ways of talking about the relative distance between the source of knowledge (the reference world) and the object of knowledge (the converging expressed world) (Frawley 1992: 413; emphasis in the original). 43

For Frawley’s model I would suggest to indicate a possible interaction between the categories of sensation and inference because sensory evidence may represent the basis for an inference.

72 Frawley then illustrates the appropriateness of his model by offering examples from Kogi that Hensarling (1982) has already described as being deictic in nature. The evidential system of Kogi represents a good example because, as shown above, it also considers the “properties of the addressee” (Frawley 1992: 415). The evidentials ne to express an assumption and skaN to express doubt would belong to Frawley’s category “from self“. Shi meaning ‘ask’ would belong to the category “from other” because by using this particle the speaker signals that he requires information from his interlocutor. Ni ‘remind’ and na ‘inform’ consequently belong to the category “to other” (cf. Frawley 1992: 415; also Palmer 2001: 62). So Frawley’s deictic classification of evidentials is not unfounded. There are languages which the deictic model may be perfectly applied to. Squartini summarises the main difference between Willett’s model and Frawley’s one in the following way: Thus, in Willett (1988) the primacy of the opposition direct vs. indirect mode of knowing produces a unification between reported and inferring evidence, which are lumped together within the general notion of indirect evidence. On the contrary, in the organizing system proposed by Frawley (1992: 413), since the deictic source of information is taken as the primary organizing principle, inferences and reports fall into different categories, inferences being more naturally embedded in the Self and reports in the Other (Squartini 2001: 302-303).

Indeed, Squartini’s analysis of the evidential use of the future and the conditional show that Willett’s model is more appropriate to represent the internal structure of evidentiality in Romance languages. His summarising table representing the use of the future as well as the conditional form to mark inference (I) and/or reported (R) illustrates this (cf. Squartini 2001: 321): Table 5: The inferential and reportive use of the future and conditional form in Romance languages FRENCH Future I44 Conditional I / R 44

45

SPANISH I I / (R)45

PORTUGUESE I/R I/R

ITALIAN I R

In his paper “Where mood, modality and illocution meet: the morphosyntax of Romance conjectures”, where the domain of epistemicity is divided into three subdomains – into the dubitative, the inferential and the conjectural mood – Squartini describes the French future as a linguistic device to express conjectural mood, whereas the French conditional is described in terms of dubitative mood (cf. Squartini 2010: 119). So interestingly, the same linguistic elements can also be dealt with below the heading of modality. ‘Reported’ is here put in parentheses because Squartini only supposes that the conditional may be systematically used to mark reported information: “[...] the intrusion of

73 Hence, in these Romance languages the future is used to mark inference “but they vary as far as the distribution of report and the functions associated with the Conditional are concerned” (Squartini 2001: 321). It is eye-catching that only the use of the Italian future and conditional would perfectly fit into Frawley’s model because the future is only used to indicate inference while the conditional is used to mark reported information. Thus, Frawley’s main differentiation between “self” and “other” is completely unproblematically applicable to Italian since inference – expressed by the future form – would belong to the “self”category and reported – expressed by the conditional – to the “other”-category (cf. Squartini 2001: 326). Looking at the distribution of ‘I’ and ‘R’, one can see that especially Portuguese represents an ‘extreme case’ because the future as well as the conditional can be used to indicate inference and reported information. Consequently, Squartini ascertains: [O]nly in Willett’s categorial analysis can the case of a neutralization between inference and report be admitted, since these two depend on one general notion (indirect evidence), so that it can be concluded that the Portuguese Future and Conditional and the French Conditional express the general notion of indirect evidence, without distinguishing between inference and report. In Frawley’s proposal there is no such [a general] notion accounting for neutralizing forms, given that in his model inference and report belong to two different major categories (Self and Other) (Squartini 2001: 323).

The categories of inference and reported – more precisely, the linguistic representation of these categories via the conditional and the future – in Portuguese, French and Spanish cannot be represented by Frawley’s model because “[...] the fact that a form is used for two different meanings is prima facie evidence for a semantic correlation between the two meanings which has to be reflected in the classification” (Squartini 2001: 301). The point in which inference and reported ‘semantically correlate’ is that the evidentially marked event was not personally witnessed at the moment of its occurrence. Arguing from a ‘less linguistic’ and ‘more logical’ point of view would result in a vicious circle because inference nevertheless seems to be less indirect than quotation as inference – simply speaking – directly occurs within the origo of the speaker while in the case of quotation at least one other origo enters the scene, namely the one which functions as information source, which transmits information. So during an inferential process the speaker does not need more than himself (and perhaps perceiv-

the reportive value seems to be a new disrupting factor introducing a non-uniform feature, possibly producing a new development in the Spanish evidential system” (Squartini 2001: 322). In a later study, in contrast, Squartini (2004: 890) shows that the conditional is used to indicate reported information in French, Italian and Spanish.

74 able evidence) to induce or deduce46 a certain state of affairs. If a speaker quotes, in contrast, his origo is shifted to another centre of consciousness (cf. von Roncador 1988: 298). Mushin (2001), who also explains that evidentiality can be characterised “as a deictic category, one that functions to index information to some point of origin (Bühler’s ‘origo’)” (Mushin 2001: 33), mentions the origo shift as well: [E]videntiality [is useful] to enable the speaker to maintain and shift the point of origin from which the propositional information modified by the evidential marker. For example, the Quechua reportive evidential clitic –shi […] enables the hearer […] to index the propositional information […] to an origo that is someone other than the current speaker […] (Mushin 2001: 34).

While in languages with a grammatical system of evidentiality reported information has to be marked as reportive information, speakers of languages that lack such a system are pragmatically motivated if they choose to mark the information they transmit as being reportive. A speaker quotes because he does not want to bear responsibility for the transmitted information or because he uses the words of the quoted person to support his own words: “[…] although a hearsay marker often signals reduced reliability, if it is used when citing the words of an expert it has the opposite effect […]” (Nuyts 2006: 11).47 By all means, inference involves (only) one origo and quotation involves more than (only) one (cf. also Volkmann 1997: 257). Applying Frawley’s criterion of directionality inference always comes ‘from the self’ and reported information always comes ‘from the other’ – no matter how it is linguistically expressed. So by understanding evidentiality as a deictic phenomenon, inference and reported clearly represent two non-related categories. This is perfectly mir-

46

Volkmann (2005: 112) explains the difference between inductive and deductive inferences as follows: Während deduktive Schlüsse so wahr wie ihre Prämissen sind, also von wahren Prämissen auf notwendigerweise wahre Sachverhalte deduktiv geschlossen werden kann, kommt dem Ergebnis von Induktion und Abduktion nur Wahrscheinlichkeit zu, denn es handelt sich um eine Möglichkeit unter anderen.

47

The highest truth value is consequently found in deductive inferences. Cornillie (2009: 50) even differentiates between three kinds of inference: circumstantial inference (direct (visible) evidence for the inference expressed), generic inference (for example, inferences from reasoning), and conjectured inference. Therefore, he disagrees with the argumentation line of van der Auwera/Plungian (1998). In the present study, for instance, I quote to achieve the opposite effect in the majority of cases. For example, in the chapter where it is argued for a superordinate category all linguists I quote ‘are used’ to support my argumentation.

75 rored by Frawley’s model (even though it is less perfect in other respects48). But the fact that both subcategories of evidentiality may be expressed by one and the same linguistic element in Spanish is also dealt with by Wachtmeister Bermúdez (2004). Consequently, he is also concerned with the ‘problem of inference’. Wachtmeister Bermúdez’ paper “La categoría evidenical del castellano: metonimia y elevación del sujeto” (2004) illustrates that the Spanish conditional is used both to indicate inference and reported information (Wachtmeister Bermúdez 2004: 7). He consequently concludes (like Squartini 2001) that “la clasificación de Willett se adapta mejor a los datos del castellano” (Wachtmeister Bermúdez 2004: 1). It is thus significant that in Evidencialidad. La codificación lingüística del punto de vista Wachtmeister Bermúdez (2005) sticks to Frawley’s model in developing his concept of “deixis evidencial” because, in his view, evidentiality is a deictic phenomenon: La evidencialidad puede [...] pensarse como un fenómeno deíctico, dado que los marcadores evidenciales son índices que apuntan a elementos del contexto extralingüístico, a saber: la fuente de información y el participante (típicamente el hablante) que tiene acceso a tal fuente (Wachtmeister Bermúdez 2005: 25).

His concept obviously represents the Spanish counterpart to Frawley’s model as well as to Frawley’s description of his model of the “deictic categorization and scaling of epistemic modality” (cf. Frawley 1992: 412-413), even though Wachtmeister Bermúdez’ concept is ‘more evidential than epistemic’: La deixis evidencial puede [...] describirse en términos de esos tres conceptos: puntos de referencia, distancia y dirección. [...] Los puntos de referencia serían los participantes con (o sin) acceso a la fuente (esto es, la magnitud acceso a la información), la distancia a la fuente de información estaría representada por el continuo entre información personal  información ajena (fuente de información) y la dirección (de acceso) a la información sería la forma (sensorial  cognitiva) en la que se llega a la información (modo de acceso a la información) (Wachtmeister Bermúdez 2005: 29-30; emphasis in the original). 48

Frawley describes – just as, for instance, Chafe (1986: 263) or Oswalt (1986: 43) do – evidentiality and epistemic modality as having a relationship of dependency. Visual perception, for example, is seen as more reliable than the perception of an event through other senses. Such a hierarchy is questioned by Willett (1988: 85) but more critically commented by Wachtmeister Bermúdez (2005: 26-27) or Fitneva (2001), who explains that “there is no intrinsic reason why perceptual information should be trusted more than inferred information: when travelers see an oasis in the desert, they rely on logic to correct what they know to be a perceptual error” (Fitneva 2001: 404). I had a funny conversation with Gesina Volkmann, who asked me: “if I were sitting in front of two cans – one filled with sugar, the other one with salt: how do I benefit from visual perception? To distinguish sugar from salt the most reliable sense in that situation is my sense of taste” (cf. also Volkmann 2005: 104, 119).

76 However, considering the linguistic forms that are used to express reported information as well as inference Wachtmeister Bermúdez (2005) – such as Squartini (2001) – also goes for Willett’s model: [...] la clasificación de Frawley supone que la categoría de la inferencia y la fuente directa sensorial conforman un conjunto (información cuya fuente es el yo) enfrentado como un todo a la información transmitida (cuya fuente es el otro). Sin embargo, en [el artículo “La categoría evidencial del castellano: metonimia y elevación del sujeto”] mostramos que el castellano utiliza un mismo tipo de estructuras para referirse a la información transmitida y a la inferencia, mientras que usa otros mecanismos para indicar que la información es sensorial. Esto sugiere [...] que son más bien las categorías de información transmitida e inferencia (es decir, fuente indirecta) las que conforman un conjunto enfrentado a la información sensorial y endofórica (fuente directa) (Wachtmeister Bermúdez 2005: 27).

He nevertheless seems to reject the idea of applying inference to indirect evidentiality as inferentially marked information represents a direct result of “mental operations” of a “thinking subject” (cf. Volkmann 2005: 110): “el conocimiento que proviene [...] de la inferencia tiene su fuente en el yo: es el yo el que infiere o deduce [...] Por el contrario, la cita y el rumor tienen su fuente en los otros” (Wachtmeister Bermúdez 2004: 10). So where to put inference then? Firstly, inference is the evidential subcategory where (epistemic) modality and evidentiality overlap, whereby it depends on the evidence that is the basis for the conclusion as to whether the inference amounts to necessity or not. An inferentially marked proposition may be possible, probable or necessary (or something in between such as highly probable, almost necessary etc.). Even though epistemic modality and evidentiality seem to overlap in inferences, in an inferential use of a certain expression it will be decided whether the evidential component or the epistemic one is more prominent. If the evidential component is regarded as more prominent the expression is labelled ‘inference’. By contrast, if the epistemic component is considered more prominent in a particular use of an expression, the expression is labelled ‘assumption’ (cf. Boye 2006: 66). In summary, inference is situated in between epistemic modality and evidentiality, whereby in an inferential use of a certain expression, the evidential domain is considered clearly more prominent than the epistemic one. Secondly, with regard to direct and indirect evidentiality or rather with regard to the ‘within or outside the speaker’s deictic centre’ problem, it is argued that inferentiality should not be said to belong to indirect evidentiality. Inference always comes ‘from the self’, while reported information, for instance, always comes ‘from the other’ – no matter how it is linguistically expressed. So by understanding evidentiality as a deictic phenomenon, inference and reported

77 clearly represent two non-related categories. However, the point which inference and reported have in common is the following: the state of affairs was not personally witnessed at the moment of its occurrence. This explains why one and the same form may be used to express one subcategory as well as the other and why Squartini (2001) favours Willett’s model rather than Frawley’s model. Willett’s model reflects better that in Portuguese, French and Spanish both subcategories, inference and reported, may be reflected by one and the same form. Thirdly (and obviously), in answering the question ‘where to put inference?’ the line of argumentation from a ‘more logical’ point of view and the line of argumentation from a ‘more linguistic’ point of view clash (cf., for instance, the lines of argumentation in Wachtmeister Bermúdez 2004 and 2005). Following the former line of argument, inference is more direct than quotation because inference directly occurs within the speaker’s origo. Fourthly (and conclusively), an appropriate model for the representation of the internal classification of evidentiality in Romance language should be developed. It must be a model that aims for both linguistic correctness and logical comprehensibility.

2.1.3 Where evidentiality and (inter)subjectivity meet Benveniste (1966) explains that a language sans expression de la personne ne se conçoit pas. […] Le langage est marqué si profondément par l’expression de la subjectivité qu’on se demande si, autrement construit, il pourrait encore fonctionner et s’appeler langage. Nous parlons bien du langage, et non pas seulement de langues particulières. Mais les faits des langues particulières, qui s’accordent, témoignent pour le langage (Benveniste 1966: 261).

That means that the expression of subjectivity is said to be inherent in language and that all utterances are somehow subjective, as they originate in the respective speaker/writer. Evidentially marked propositions do so as well: an evidentially marked utterance is always subjective in a certain way. Even if the evidence or rather the information source for the proposition is objective in nature, i.e. that something was visually perceived and this source was thus accessible to other speakers, the visually perceived information source for a certain state of affairs has to be verbally expressed; it has to be encoded. And if it is verbalised by the speaker, it has to be cognitively processed, and consequently interpreted. Imagine speaker A watching speaker B and C talking to each other with raised voices. Speaker A then tells speaker D about the event he had observed. He may now say: (a) I have seen that speaker B and C were verbally fighting with each other; (b) I have seen that speaker B and C were discussing with each other;

78 (c) I have seen that speaker B and C were heavily discussing with each other; (d) I have seen that speaker B and C were having a small dispute or (e) I have seen that speaker B and C were having a big dispute. So even visually perceivable information, which is commonly acknowledged as objective evidence, runs, if the information is cognitively processed, through the subjective ‘filter’ of the observing individual. Therefore, even though a speaker may deal with an obviously objective evidence or source of information for a state of affairs, because it was visually perceptible for instance, an utterance containing an evidential expression can never be totally objective, but at least subjective, at most intersubjective. According to Cappelli (2007: 133), the ‘most subjective’ utterances are those that contain an evidential expression indicating affective evidence, whereby – in my view – metaphorically used lexemes such as ‘fighting’ in sample sentence (a) cannot be said to be less subjective. A similar high degree of subjectivity is expressed via inferential expressions because inferences drawn by the speaker’s ego represent his cognitive results, although they may rely on objectively perceivable evidence. Defining the dimension of subjectivity, Nuyts (2001a) differentiates on the one hand between subjectivity and objectivity, but on the other hand between subjectivity and intersubjectivity, thus establishing a second opposition pair and taking up as well as refining Lyons’ subjectivity-objectivity opposition: (i) […] in (my reassessment of) Lyons’ view, the dimension [of subjectivity] concerns the quality of the speaker’s evidence for an epistemic evaluation: does (s)he have good, mathematically or formally reliable evidence (i.e., objectivity), or does (s)he have poor or vague, intuitive evidence (i.e., subjectivity). (ii) Alternatively, the dimension can be defined […] as follows: one pole involves the speaker’s indication that (s)he alone knows (or has access to) the evidence and draws conclusions from it; the other pole involves his/her indication that the evidence is known to (or accessible by) a larger group of people who share the same conclusion based on it. In the former case the speaker assumes strictly personal responsibility for the epistemic qualification, in the latter case (s)he assumes a shared responsibility among those who have access to the evidence and accept the conclusions from it (including him/herself). An important (but maybe not a necessary) element in how this distinction actually works may then be whether the interaction partner belongs to those sharing the evidence and the conclusions, or not (Nuyts 2001a: 393-394).49 49

Cappelli criticises that Nuyts (2001a) describes intersubjectivity as belonging to the evidential domain “because it seemed to be a category expressing shared commitment rather than evidential meaning” (Cappelli 2007: 130). According to her, subjectivity and intersubjectivity are two dimensions that are connected with the notion of commitment so that she proposes to rather speak of ‘individual commitment’ vs. ‘shared commitment’ (Cappelli 2007: 134).

79 In Epistemic Modality, Language, and Conceptualization. A CognitivePragmatic Perspective, where Nuyts (2001b) analyses the functional and structural characteristics of modal sentence adverbs, modal adjectives, modal auxiliaries and verbs of cognitive attitude in a comparative way concerning their expression of epistemic modality, he “reframe[s] the positions in the literature regarding the old sub/objectivity distinction in terms of the alternative analysis of (inter)subjectivity” (Nuyts 2001b: 64). Cornillie explains that a “statement is subjective when the evidence is restricted to the speaker’s realm, while a statement is called intersubjective when the speaker (assumes (s)he) shares it with other people” (Cornillie 2007b: 124) and that the concept of (inter)subjectivity has p[r]oven to be a useful addition. Since intersubjective statements have broader support than subjective ones, the former are seen as more reliable and are, hence, considered to express “near-factivity”, while, based on the speaker’s inference only, the latter lead to a “non-factive” interpretation (Cornillie 2007b: 126-127).

Despite drawing a distinction between objectivity and subjectivity (and not between subjectivity and intersubjectivity), or more precisely, between objective modality and subjective modality (cf. Lyons 1995: 329-334)50, Lyons also defines subjectivity as “the way in which natural languages, in their structure and their normal manner of operation, provide for the locutionary agent’s expression of himself and of his own attitudes and beliefs” (Lyons 1982: 102). Although it is assumed that the meaning of intersubjectivity and its ‘scope’ within linguistic research is ‘intersubjectively comprehensible’, it seems to depend on the study itself and the linguistic forms to be analysed whether to work with one opposition pair or with the other: for example, in his paper “Conceptual and constructional considerations on the subjectivity of English and Spanish modals”, Cornillie (2006) draws a distinction between epistemic modals belonging to the domain of subjectivity and deontic modals belonging to the domain of objectivity because “Spanish epistemic modals subjectively ground the event expressed by infinitive, whereas the deontic modals rather profile the force structure belonging to the objective scene” (Athanasiadou/Canakis/Cornillie 2006: 8). For Cor50

Cappelli – as me – rejects the existence of objective modality explaining that [i]f we consider the speaker to be the evaluator of a state of affairs, or of the propositional content p of a sentence, it seems hard to accept the existence of a type of modality which is totally objective: whatever is filtered through the speaker’s subjectivity becomes itself subjective (Cappelli 2007: 129). As there is no objective modality, she denies the opposition subjective vs. objective modality (cf. Cappelli 2007: 134).

80 nillie’s article “On conceptual semantics and discourse functions: The case of Spanish modal adverbs in informal conversation” (2010a) on the other hand, which also deals with discourse functions of Spanish epistemic adverbs/adverbial phrases, the notion of intersubjectivity – even though not mentioned explicitly – is of prime importance as he shows that a lo mejor and igual invite the co-participant to approve or reject the hypotheses put forward. They are involved in on-line planning to ensure a reaction/reply from the co-participant (via turn-taking) (Cornillie 2010a: 313).

To sum up, speakers use a lo mejor and igual to involve the addressee’s self, which is one feature that constitutes intersubjectivity (cf. Traugott 2010). Benveniste (1971 [1958]: 225) differentiates between subjectivité and objectivité with the help of the notions sujet d’enoncé (subject of the statement [p]) and sujet d’énonciation (subject who states that [p]). So the former represents the (objective) syntactic subject, while the latter represents the speaker (cf. also Athanasiadou/Canakis/Cornillie 2006: 1). Benveniste (1966) also associates subjectivity and its discourse-character: Le langage est la possibilité de la subjectivité, du fait qu’il contient toujours les formes linguistiques appropriées à son expression, et le discours provoque l’émergence de la subjectivité, du fait qu’il consiste en instances discrètes (Benveniste 1966: 263).

The association of subjectivity to the fact that it is provoked in discourse situation leads to the notion of intersubjectivity that “was first used in modern linguistics by Benveniste (1958) who regards the speaker-addressee interaction as the fundamental condition for linguistic communication” (Cornillie 2007b: 112). But the perhaps most influential line of research concerning (inter)subjectivity and (inter)subjectification is based on grammaticalisation theory and lead by Traugott, who – together with Dasher – also explains that “especially the discourse processes” (Traugott/Dasher 2002: 6) are involved in change that lead to meanings expressing subjectivity: “meanings tend to come to refer less to objective situations and more to subjective ones (including speaker point of view), less to the described situation and more to the discourse situation” (Traugott 1986: 540). Hence, Traugott/Dasher (2002) and Traugott (1986) – who firstly proposed her notion of subjectivity in 1982 – also establish a connection between the concept of subjectivity and its appearance in discourse, but it is strictly differentiated between the notions of (inter)subjectivity and (inter)subjectification:

81 I have called the diachronic process of semanticization “(inter)subjectification”, assuming that an important (though not rigid) distinction is to be made between –ity (synchronic state) and –ation (diachronic process) […] (Traugott 2010: 30).

In this connection Cuyckens/Davidse/Vandelanotte (2010: 4) explain: It has been a constant of Traugott’s thinking to define subjectification and intersubjectification as semanticization, which requires the new (inter)subjective meanings to be conventionally coded by the forms, with new form-meaning pairs as result.

So when “meanings tend to become increasingly situated in the speaker’s subjective belief state or attitude toward the proposition” (Traugott 1989: 31) we have to do with the diachronic process of subjectification, “the semasiological development of meanings associated with a form such that it comes to mark subjectivity explicitly” (Traugott 1999: 179). Such meanings she describes as enriched with the speaker’s perspective and thus pragmatically strengthened (cf. Traugott 1999: 188). In “The rhetoric of counter-expectation in semantic change. A study in subjectification” (1999) Traugott describes this process of pragmatic strengthening with the help of the semantic development of in fact, which is now used to emphasise subjectivity, having clearly gotten “a pragmatic function at the speech-act level itself” (Blank/Koch 1999: 3). Athanasiadou/Canakis/Cornillie (2006: 3) explain that especially lexical items or constructions which carry a ‘subjective moment’ with them are affected by this pragmatic strengthening and generalise: Therefore, if the meaning of a lexical item or construction is grounded in the sociophysical world of reference, it is likely that over time speakers will develop polysemies that are grounded in the speaker’s world, i.e., in reasoning, belief, or metatextual attitude to the discourse.

Consequently, it can be said that subjectification is a shift […] from meanings grounded in more or less objectively identifiable extralinguistic situations to meanings grounded in text-marking […] to meanings grounded in the speaker’s attitude to or belief about what is said (Traugott & König 1991: 189).

In Traugott (1988: 409-411) these early tion/subjectification are described in more detail:

stages

of

grammaticalisa-

82 Table 6: Traugott’s stages of grammaticalisation/subjectification (1988: 409-411) SEMANTIC-PRAGMATIC TENDENCY I “Meanings based in the external described situations Æ meanings based in the internal (evaluative/ perceptual/cognitive) situation” (1988: 409).

SEMANTIC-PRAGMATIC TENDENCY II “Meanings based in the described external or internal situation Æ meanings based in the textual situation” (1988: 409).

SEMANTIC-PRAGMATIC TENDENCY III “Meanings tend to become increasingly situated in the speaker’s subjective beliefstate/attitude toward the situation” (1988: 411).

Before outlining the concept of (inter)subjectivity I work with regarding the framework of this study and before discussing where (inter)subjectivity and evidentiality meet, the notion of intersubjectivity should be considered in greater detail: roughly speaking, in Traugott’s sense, intersubjectivity is represented in meanings that mark “attention to the addressee’s self-image” (2010: 29), that is, meanings involving “a communicative relationship between speaker and hearer” (Cuyckens/Davidse/Vandelanotte 2010: 13). So Traugott, who mentions certain uses of hedges such as sort of, well and perhaps as serving intersubjective functions (2010: 37), clearly applies her understanding of intersubjectivity to communicative situations, to speaker exchange. This is what Cornillie (2010a, 2010b) does as well: in his paper “An interactional approach to epistemic and evidential adverbs in Spanish conversation” he pays attention to the discourse functions of adverbs like quizá or evidentemente and shows that they all serve (different) functions in terms of organisation of turn-taking, thus serving more/other functions than (only) expressing the ‘classical’ epistemic and evidential qualifications, when analysed in conversational settings. Having analysed Dutch and German examples that contain modal adverbs as the following one Alle Sterne in einem solchen Sternhaufen sind sehr wahrscheinlich etwa gleichzeitig aus einer gemeinsamen großen Gaswolke entstanden ‘All stars in such a star cluster have very probably developed more or less simultaneously out of a common big gas cloud’ (Nuyts 2001b: 65; my emphasis; cf. also Nuyts 2001a: 389),

Nuyts concludes that the […] effect of intersubjectivity […] does not seem to be caused by the use of the modal adverb as such. It is rather a contextual effect, due to our general knowledge that there has been very substantial scientific research on space […] (Nuyts 2001b: 65).

At this point it has already become clear that the notion of intersubjectivity is used differently, due to the distinct frameworks it was the focus of. Nuyts himself explains:

83 The present notion of subjectivity […] is much narrower that [sic] Traugott’s, but they are linked in that the present dimension of (inter)subjectivity is very high on the scale of subjectivity in Traugott’s sense (Nuyts 2001b: 38-39).

While Traugott clearly addresses its discourse function in interactional settings, mentioning the speaker-hearer relationship, Nuyts – despite doing so as well – also relates the concept of intersubjectivity to notions as common (general) knowledge, mentioning that certain linguistic devices are used when the speaker assumes sharing knowledge with the hearer (cf. Nuyts 2001b: 37). So for him – but the justification is questionable – intersubjectivity and mirativity are ‘consequently’ linked because speaker and addressee share knowledge that is not new (cf. Nuyts 2001b: 37), although the term mirativity itself “refers to the linguistic marking of an utterance as conveying information which is new or unexpected to the speaker” (DeLancey 2001: 369-370). Nuyts (2001b: 307) means to show that “[e]pistemic adverbs are entirely neutral as to (inter)subjectivity”. And it even appears difficult if not downright impossible to combine them with separate subjectivity markers, such as the adverbial personally, or the prepositional phrases to me or in my opinion: cf. *personally they have probably run out of fuel, *to me they have probably run out of fuel, or even ??in my opinion they have probably run out of fuel. This contrasts with the verbal and adjectival epistemic expressions, which easily accept additional subjective forms […] (Nuyts 2001b: 307).

Cornillie has also shown for probablemente that the modal adverb is not used to invite the co-participant to give his/her opinion (2010a: 311), but the speaker himself uses the adverb to comment “upon a central idea that is already defined in the previous context” (Cornillie 2010a: 313). So probablemente is indeed not used intersubjectively, but while Nuyts refers to both subjectivity and intersubjectivity, Cornillie only refers explicitly to the latter. If probablemente is used for commenting upon something already mentioned, the subjective dimension is clearly present. The modal adverb evidentemente, in contrast, was shown to serve a special interactional function: it is used if the speaker wants to keep the turn; “by means of evidentemente the speaker lets the co-participant know that (s)he is still developing his/her idea” (Cornillie 2010b: 327). Therefore, Cornillie would not term epistemic adverbs as being entirely neutral as to (inter)subjectivity, as Nuyts does. Let us now summarise the (slightly) different uses of the notion of (inter)subjectivity: Cornillie, for instance, in analysing both lexical and grammatical uses of Spanish parecer constructions, has found out that parece que is used intersubjectively as “the access to the evidence is always intersubjective” (Cornillie 2007b: 127), while parecer + infinitive is only used subjectively, that is, it “explicitly encodes the [speaker’s/writer’s] point of view” (Traugott/Dasher 2002: 21-22). So in his paper “On the continuum between lexical and grammati-

84 cal evidentiality. Evidence from Spanish”, Cornillie defines intersubjectivity “in terms of shared status of the statement emitted”. “Shared evidence – or the assumption of shared evidence – leads to an intersubjective view of the state of affairs expressed by the speaker” (2007b: 113). He tests the evidentially used parecer constructions by adding the phrase pero yo no lo veo así or the questions ¿Quién lo dice? and ¿Tú crees?. If these contextualisations obtain, they are interpreted as an indication of an intersubjective dimension (Cornillie 2007b: 114). Even though it seems to be a legitimate testing procedure, the interactional setting – and thereby the addressee’s perspective – itself is not analysed, which Traugott/Dasher pay more attention to when they focus on “coded expression[s] of [speaker’s/writer’s] attention to the image or ‘self’ of [addressee/reader]” (2002: 22). In comparing subjectivity and intersubjectivity, Nuyts (2001b) defines the latter: […] does (s)he indicate that the evidence is known to (or accessible by) a larger group of people who share the conclusion based on it. […] (s)he assumes a shared responsibility for it (although (s)he remains coresponsible too, of course) (Nuyts 2001b: 34).

And then he (2001b: 34) adds that an “important element in how this distinction actually works may then be whether the interaction partner belongs to those sharing the evidence and the conclusions or not”, which is not only an important element for Traugott (and Dasher) but also the crucial point if speaking of intersubjectivity. Nuyts, additionally, seems to handle the notion of intersubjectivity in terms of a gradual concept: he speaks of intersubjectivity if the speaker “assumes a shared responsibility among those who have access to the evidence and accept the conclusions from it (including him/herself)” (Nuyts 2001a: 393) and names the fact whether the interaction partner belongs to those who share evidence and conclusion a “maybe not […] necessary” one (2001a: 394). Nuyts (2001a: 398) concludes that the “dimension of subjectivity, as it is generally assumed to be present in epistemic modal expressions, is actually (in principle) an independent evidential-like qualificational category”. This is what is similar to Cappelli’s line of argumentation. In analysing the epistemic/evidential use of English verbs of cognitive attitude, she does not only relate subjectivity to deixis but also implicitly evidentiality (and epistemic modality) to subjectivity: Subjectivity is a frequently used term and its sense varies within the different traditions of research. Stein (1995) identifies five major senses. […] third, subjectivity as a deictic for the speaking ego […] I use subjectivity in this sense, in a way akin to Stein’s (1995) third sense, as the expression of an ego and its pervasiveness in the evaluation, be it performed or reported. In this sense, verbs of cognitive attitude are means for the expression of subjectivity (Cappelli 2007: 131).

85 Precisely this ‘deictic understanding’ of subjectivity can be arranged with the concept of evidentiality: This view of subjectivity can be integrated into the evidential dimension. If the speaker can encode reference to the external source of information, why cannot he/she encode reference to the absence of such external evidence? […] The speaker can signal that he/she is uttering a personal judgement based on no other evidence but his/her own personal evaluation, his/her feeling or some other type of vague evidential information. This sort of evidence is here labelled “affective evidence”, and it is taken to include impressions, “irrational judgements” and any type of evidence depending on the ego of the evaluator or for which the evaluator cannot provide a precise definition (Cappelli 2007: 131-132).

So in expressing affective evidence subjectivity and evidentiality clearly meet. But a scalar relationship between the two concepts should be assumed as indicated at the start of this chapter: before an information source for the state of affairs is verbalised by the speaker, it was cognitively processed, evaluated and interpreted. Consequently, an utterance containing an evidential expression can never be totally objective, at most intersubjective if the information source and its interpretation is (assumed to be) shared with other speakers, including the addressee. Hence, a proposition that contains an evidential expression is always – at least a tiny little bit – subjectively marked. The reason is that the perception of the state of evidence is already by itself subjective (at most intersubjective). It is only possible to look through one’s own eyes and to hear with one’s own ears etc. My understanding of subjectivity markers is a pragmatic one. Although, for instance, in “the mental state predicates subjectivity is arguably part of the lexical semantics” (Nuyts 2001b: 270), there are other markers of subjectivity such as the evidential use of deber or the inferential use of the synthetic future form that increased “in pragmatic force” (Traugott 2003: 633) during the pragmaticisation process (cf. also Haßler 2003), which is a key concept for the present study. In their study of demonstratives, Hayashi/Yoon (2010) define pragmaticisation as a process whereby grammatical items evolve into pragmatic markers that serve specific discourse functions to such an extent that they display a number of phonological, morpho-syntactic, and/or semantic characteristics that diverge crucially from their original usage as grammatical items (Hayashi/Yoon 2010: 58).

This definition of pragmaticisation may be perfectly applied to the development of demonstratives but for the use of deber and the Spanish synthetic future form, for instance, it must be highlighted that these are markers of subjectivity that increased in pragmatic force. In contrast to the notions of “pragmatic strengthen-

86 ing” (Fried 2009: 262) or “pragmatic narrowing” (Vega Moreno 2007: 205), the evidential expressions discussed here should be considered under the heading of ‘pragmatic broadening’51. They fulfil different functions, and which function it is, depends on the respective pragmatic situation.52 So subjectivity is bound to language use and that is what it makes a notion belonging to the field of pragmatics. In the present study, the use of the notion of grammaticalisation is avoided if dealing with the Spanish forms analysed here53 because I doubt that it could be verified that these markers of subjectivity are systematic. Markers of subjectivity do not have grammatical meaning; they stand for pragmatic circumstances. Therefore, in analogy to Traugott’s semantic-pragmatic tendencies representing steps of the subjectification process, I would speak in terms of steps/stages of ‘context-free means of expression’. The synthetic future form, for example, (still) marks primarily future time, but beyond that it has acquired modal/evidential meaning. However, this is due to its (pragmatic) use. For this study, which deals with evidentiality, Cappelli’s words represent on the one hand a good summary about the relationship of evidentiality and subjectivity and reflect on the other hand my understanding of subjectivity in the context of the analysis of markers of subjectivity in use: Evidentiality and subjectivity have been traditionally seen as competing dimensions. On the one hand, there would be subjectivity, i.e. the evaluator’s point of view, the expression of his/her stance towards the state of affairs; on the other hand, there would be evidentiality, the evidence that can back an evaluation or an assertion. If no evidence is provided, subjectivity predominates, otherwise subjectivity decreases […] (Cappelli 2007: 132-133). [But their] relationship is much more complex […] [V]ery reliable evidence gives rise to the impression of a minor subjectivity and viceversa. The two dimensions meet in affective evidence, where subjective judgement itself is assumed to the role of evidence intended as a base for evaluation (Cappelli 2007: 133).

But what about intersubjectivity? Due to the text type which is worked with – online-news articles – the present study cannot be concerned with the question in which cases a certain linguistic device is used intersubjectively. It is simply 51 52 53

This term is also used by Vega Moreno (2007: 205). She, however, uses it in terms of idioms, transparency and pragmatic inference, as her study is about figurative speech. The construction puede que (cf. chapter 6.2) represents an exceptional case. It is not considered under the heading of ‘pragmatic broadening’. Again, puede que represents an exceptional case. Rodríguez Espiñeira/López Meirama (2008: 307) have shown that puede que developed from puede ser que and that the use of the more grammaticalised variant (puede que) is more subjective than the use of puede ser que.

87 not possible always unambiguously to determine whether a speaker uses a certain expression in order to reveal intersubjectivity. If a directly quoted speaker, for example, unequivocally indicates “that the evidence is known to (or accessible by) a larger group of people who share the conclusion based on it” (Nuyts 2001b: 34) and he clearly expresses to share responsibility for it (maybe even with the interviewer), the intersubjective dimension is obvious, and then it will be referred to that fact. But it is certainly impossible to show intersubjectively used expressions on the other level; it is considered (nearly) impossible to determine whether a journalist uses a certain means of expression to reveal intersubjectivity, although the past subjunctive is said to be such a means of expression conveying intersubjectivity (Lunn 1995). In journalistic discourse, there are subgenres that are characterized by very heavy Subjunctive use. In particular, the specialized press which addresses a knowledgeable readership (e.g. […] sports pages […]) uses the Past Subjunctive to mark information which can be assumed to be known to assiduous readers. (Lunn 1995: 432).

So the above mentioned use of the past subjunctive could be an example of how to express intersubjectivity. Lunn does not speak once about the concept of intersubjectivity, even though she mentions Guitart’s distinction between the use of the subjunctive on the pragmatic level and its use on the semantic level, whereby the former would coincide with intersubjectivity: Guitart (1991) has suggested a similar approach to the Spanish Subjunctive by distinguishing between pragmatic presupposition, based on speaker assumptions about hearer knowledge, and semantic presupposition, based on sentence-level truth conditions (Lunn 1995: 432).

Cases where speakers clearly express that they (assume to) share evidence with more people ‘than only with themselves’, are to be considered and analysed in terms of intersubjectivity. However, the restriction regarding the quoted speaker-level should be ‘intersubjectively comprehensible’. To close this chapter it should be kept in mind that evidentiality and subjectivity especially – but not exclusively – meet in affective evidence, where a subjective judgement represents the basis for evidential evaluation.

2.1.4 Where evidentiality and polyphony meet Comparing the lexemes ‘subjectivity’ and ‘polyphony’ one is about to think ‘while the former refers to the expression of a (very) personal viewpoint, the latter nearly literally expresses more than one voice; hence they exclude each other’. But thinking about the concepts of subjectivity and polyphony and relating them to the domain of evidentiality, for one thing, both seem to be related,

88 and for another thing, both can be related to evidentiality, albeit to distinct subcategories: subjectivity to – in Willett’s (1988: 57) terms – ‘Inferring’; polyphony, very roughly speaking, to ‘Reported’. While the relationship of polyphony and subjectivity to different evidential subcategories may have immediately become ‘intersubjectively comprehensible’, the relationship between polyphony and subjectivity should be explained more precisely: for Huarochirí texts – to first use an example of a language that possesses real evidentials – it could be shown that story tellers ‘manipulate evidentials’. As Aikhenvald (2004: 320) points out, in “Huarochirí texts – the oldest known source written in Quechua – manipulating evidentials is […] a stylistic device. The link between the narrative type and the evidential choice is transparent”. Otherwise put, in Huarochirí texts, when the narrator comments on what he himself has told, that means, when he takes an explicit position in a story, he does not use the reported evidential to indicate his explicit position, but the one that indicates direct evidentiality (Dedenbach-Salazar Sáenz 1997: 153-154). In this sense, the narrator manipulates the evidentials. In commenting on a story, the validational aspect comes in, and because of the fact that the commentator may be different from the narrator and the former “puts the story into the overall framework of the collection, and may be different again from a redactor who may have been responsible for putting the texts together and writing them down” (Dedenbach-Salazar Sáenz 1997: 164), the polyphonical aspect comes in as well. There are more voices heard than only the one by the narrator (Dedenbach-Salazar Sáenz 1997: 164). Hence, the consequence that there are more voices heard is due to the narrator’s wish to express a personal, a subjectively tinged comment. But while this manipulating use of evidentials results in the fact that the polyphonical aspect may be described as being due to a subjective comment, polyphony by itself is normally related to the evidential subcategory reported, as more than one voice can be heard. Then there are at least two voices: the current speaker’s/author’s and the quoted speaker’s/author’s. Thus, the notion of polyphony presupposes a “diversity of voices” (Bakhtin 2008: 300), whereby for this study it would be too simple applying the concept of dialogicity to all language use. Incorporating the notion of polyphony in a study of evidentiality in such a way would be too easy. Indeed, language use may always be dialogic (Bakhtin 2008), but relating dialogicity to evidentiality in this way (only), would be too superficial. However, before outlining how the concept of polyphony is integrated in the present study, we should first have a side glance at the conceptual origin of (literary) polyphony and on Bakhtin’s work on dialogicity and heteroglossia of language54: 54

For a summarising overview on Bakhtin’s notions ‘polyphony’, ‘dialogism’ and ‘heteroglossia’ see Pearce (2006).

89 According to Bakhtin, any act of speech is intrinsically dialogic, not only because it presupposes interaction with an addressee, but also because it includes a multiplicity of voices. Bakthin’s emphasis is on the heteroglossia and internal stratification of any language, on the dynamic process of taking and using another person’s words or thoughts as a natural element of communication, and on the different degrees of distance that one may assume as to one’s own discourse (Bondi 2009: 83).

Indeed, while Bakhtin focuses on heteroglossia in his essay on “Discourse in the Novel” (2008), he coined his concept of polyphony in Problems of Dostoevsky’s Poetics (1984), when he speaks about the authorial position in relation to its ‘autonomous’ characters: […] the new artistic position of the author with regard to the hero in Dostoevsky’s polyphonic novel is a fully-realized and thoroughly consistent dialogic position, one that affirms the independence, internal freedom, unfinalizability, and indeterminacy of the hero. For the author the hero is not ‘he’ and not ‘I’ but a fully valid ‘thou’, that is, another and autonomous ‘I’ (‘thou-art’) (Bakhtin 1984: 63).

With ‘heteroglossia’ he refers to the “diversity of social speech types (sometimes even diversity of languages) and a diversity of individual voices, artistically organized” by an author (Bakhtin 2008: 262). Due to the fact that both heteroglossia and polyphony are terms that refer to the multiplicity of voices or ‘tongues’ – if referring to the etymology of heteroglossia – in the present study the two notions are used synonymously. Doing so in literary studies would be fatal, but from a linguistic point of view, heteroglossia as well as polyphony implies ‘more than one voice’ and that is the point/meaning that is crucial for this study. In the present study, ‘more than one voice’ comprises not only the explicit mentioning of (an)other epistemic centre(s) but also the allusion of (an)other epistemic centre(s) because in both cases more than one voice is present. So ‘voice’ is to be understood as an epistemic centre, in which the representation of [p] originates/originated. In contrast to Volkmann’s notion of epistemic centre (2005: 453), in the present study ‘epistemic centre’ may also refer to inanimate entities because – as will be shown in later chapters – also reports55, for instance, may be quoted and thus represent another epistemic centre, in which [p] originated. However, in the present chapter – for the sake of simplicity – ‘voice’ is equivalent to ‘speaker’. In “Discourse in the Novel” Bakhtin notes that the use of words is always dialogic as a speaker automatically shares his words with other ‘users’ as long as he does not ‘individualise’ them because the word itself is dialogic (“internal dialogism of the word”; Bakhtin 2008: 279): 55

Obviously, reports are also written by ‘animate entities’, but – as will be shown later – sometimes the author of the report is not referred to rather the report as medium itself.

90 As a living, socio-ideological concrete thing, as heteroglot opinion, language, for the individual consciousness, lies on the borderline between oneself and the other. The word in language is half someone else’s. It becomes “one’s own” only when the speaker populates it with his own intention, his own accent, when he appropriates the word, adapting it to his own semantic and expressive intention (Bakhtin 2008: 293). […] many words stubbornly resist, others remain alien, sound foreign in the mouth of the one who appropriated them and who now speaks them; they cannot be assimilated into his context and fall out of it; it is as if they put themselves in quotation marks against the will of the speaker. Language is not a neutral medium that passes freely and easily into the private property of the speaker’s intentions; it is populated – overpopulated – with the intentions of others (Bakhtin 2008: 294).

In the same essay he studies the “heteroglot language” (Bakhtin 2008: 299) of novels such as in Dickens’ Little Dorrit. He observes that the novelist in general welcomes the heteroglossia and language diversity of the literary and extraliterary language into his own work not only not weakening them but even intensifying them (for he interacts with their particular self-consciousness). It is in fact out of this stratification of language, its speech diversity and even language diversity, that he constructs his style, while at the same time he maintains the unity of his own creative personality and the unity […] of his own style (Bakhtin 2008: 298).

Bondi’s words on Bakhtin’s remarks about the different degrees of distance that an author may express between his own discourse and another one (cf. Bondi 2009: 83), can be verified by the following text passage of “Discourse in the Novel”: The language of the prose writer deploys itself according to degrees of greater or lesser proximity to the author and to his ultimate semantic instantiation: certain aspects of language directly and unmediatedly express […] the semantic and expressive intentions of the author, others refract these intentions […] yet another group may stand even further from the author’s ultimate semantic instantiation, still more thoroughly refracting his intentions; and there are, finally, those words that are completely denied any authorial intentions: the author does not express himself in them (as the author of the word) – rather, he exhibits them as a unique speech-thing, they function for him as something completely reified (Bakhtin 2008: 299).

Although arguing against a ‘novelistic background’, the same degrees of proximity may apply to the relation between a quoting speaker/writer and quoted words the current speaker/writer denies any authorial intentions for. The other way round, if a speaker/writer quotes another person’s words from a joint discourse situation, in which both developed their ideas alternately, influencing each other, the current speaker/writer would not totally distance himself as he implicitly has had ‘authorial influence’. The degree of proximity would be lesser if, for instance, speaker A spoke to B, then B told C what A had told him and C

91 tells D what B had told him about A’s words. In that case, the current speaker C does not express himself in A’s words as he did not have ‘authorial influence’ on the development of A’s ideas during the discourse situation (only B had). So the degrees of greater or lesser proximity depend on the participation of a discourse situation. In this regard, the level distinction between author and narrator may be analogically applied to quoting speaker/writer and quoted person; the author lets the narrator speak; the current speaker/writer lets another person speak: We acutely sense two levels at each moment in the story; one, the level of the narrator, a belief system filled with his objects, meanings and emotional expressions, and the other, the level of the author, who speaks (albeit in a refracted way) by means of this story and through this story. The narrator himself, with his own discourse, enters into this authorial belief system along with what is actually being told. We puzzle out the author’s emphases that overlie the subject of the story, while we puzzle out the story itself and the figure of the narrator as he is revealed in the process of telling his tale. If one fails to sense this second level, the intentions and accents of the author himself, then one has failed to understand the work (Bakhtin 2008: 314).

Consequently, the two belief systems – that of the narrator and that of the author – are described as being in interaction so that a dialogic tension between the two languages as well as between the two belief systems emerges. This dialogic tension, in turn, permits the realisation of authorial intentions and causes that these authorial intentions can be sensed “at every point in the work” (Bakhtin 2008: 314). These subtly presented authorial intentions are what Bakhtin terms ‘heteroglossia’; it is another’s speech in another’s language, serving to express authorial intentions but in a refracted way. Such speech constitutes a special type of double-voiced discourse. It serves two speakers at the same time and expresses simultaneously two different intentions: the direct intention of the character who is speaking, and the refracted intention of the author. In such discourse there are two voices, two meanings and two expressions (Bakhtin 2008: 324).56

The heteroglossic component may also be applied to ‘real life’ quotations. For example, my “refracted intentions” are expressed in about 90% of the quotes used in this study. If another person’s words are not quoted for the purpose of disproving them later, we deal with so-called authority quotes (notion adopted

56

Although being aware of the distinction made in literary studies between the notions of heteroglossia and polyphony, both notions can be said to refer to the multiplicity of voices or tongues. Hence, the crucial point for the present study is that both concepts share the property of implying ‘more than one voice’ so that the distinction made in literary studies is not important for the present study.

92 but translated from Volkmann 2005: 12157), through which the author (me) expresses her personal authorial intentions, that is, my speech is to be found in another’s language. That is why I use other person’s words; they help me express myself – albeit in a refracted way. By using authority quotes the quoting speaker shares his point of view with the cited person, but utterances are also polyphonic if they involve “more than one point of view” (Gévaudan 2008: 1). That is where Ducrot (1980, 1984) comes in: up to now the concept of literary polyphony to non-literary language use was already applied, but Ducrot developed the concept and terminology of linguistic polyphony, whereby the concept “conceives polyphonic utterances as involving more than one point of view” (Gévaudan 2008: 1). In Le dire et le dit, in “Esquisse d’une théorie polyphonique de l’énonciation” (chapter 8), Ducrot divides the speaking subject (in general) into “locuteur” and “énonciateurs”. The “locuteur est un être de discours” (Ducrot 1984: 203) and the “énonciateurs” are êtres qui sont censés s’exprimer à travers l’énonciation sans que pour autant on leur attribue de mots précis; s’ils ‘parlent’ c’est seulement en ce sens que l’énonciation est vue comme exprimant leur point de vue, leur position, leur attitude mais pas au sens matériel du terme, leurs paroles (Ducrot 1984: 204).

This implies that Ducrot transferred Bakhtin’s author-narrator distinction in literary terms to the linguistic distinction between locuteur and énonciateur. Ducrot’s linguistic concept of polyphony was refined by Nølke (2001, 2009) and Nølke/Fløttum/Norén ([henceforth: ScaPoLine] 2004) in ScaPoLine. La théorie scandinave de la polyphonie linguistique (2004). In Le regard du locuteur Nølke (2001: 15-34) develops notional tools and an appropriate terminology to be able to handle polyphonic expressions on the sentence level. He – as well as Ducrot (1984) – differentiates between locuteur (the speaking person of the speech act; the current utterer) and énonciateur (the ‘creator’ of the original utterance), that is, he distinguishes between quoting speaker and quoted speaker (cf. Nølke 2001: 17). These are the notions that are worked with in the present study. I will 57

Volkmann speaks about an “Autoritätszitat” if, for example, professional literature or serious news services are quoted (Volkmann 2005: 121). Cornillie/Delbeque (2008: 39) also explain that “the value of reported knowledge is also associated with the authority attributed to the source of information”. The following text passage, for instance, contains an authority quote: Así pues, y por simple coherencia, deberíamos hacer nuestra, de todos, la definición del filósofo Herbert Marcuse en la que afirma que la cultura, entendida como proceso de humanización, solo puede ser posible, y cito textualmente, con la “exclusión de toda crueldad, fanatismo y violencia no-sublimada”. Solo así conseguiremos un progreso sin barbarie. (El País 16/08/2010)

93 differentiate between quoting speaker or current speaker and quoted speaker or quoted person, as I do not want to strictly distinguish the locuteur from the énonciateur for the following reason: the locuteur is defined as the current speaker, while the énonciateur, the ‘originator of the utterance’, is described as an “être discursif” (Nølke 2001: 17; cf. ScaPoLine 37-43; cf. Nølke 2009: 2225) who is also part of the current discourse, as he is somehow ‘incorporated’ by the quoting speaker. And this incorporation of the quoted person into a new discourse is not always – more precisely, seldom – done in a neutral way. The quoting person rarely cites another person only for quoting him. But this is what brings the terminology, the notional distinction between locuteur and énonciateur, with it. To my mind, the current speaker (nearly) always creates a new utterance, even if the main part of the utterance might have been created by another person, the quoted one. So the quoting speaker as well as the quoted speaker is an énonciateur as both create a (new) utterance. That is why I will stick to the alternative terms of quoting speaker and quoted speaker58. However, it should be taken into account that the quoted speaker does/did not have to necessarily exist physically, that is, he may also be imagined by the current speaker at the time of quoting. The quoted speaker may also be identical with the quoting one (cf. Nølke’s notion of “être discursif”). In this case by quoting another opinion is expressed. To sum up thus far, the ‘classic’ polyphonic terminology is not considered really appropriate.59 Locuteur and énonciateur are not differentiated between for the following reason: a locuteur is always (!) the creator of a new utterance – and thus an énonciateur, even if the main part of his statement is quoted or represents another opinion. Locuteur and énonciateur only distinguish from one another in the point that the quoted speaker is also part of the current discourse created by the quoting speaker, that is, the cited person is an être discursif. So the distinction by itself is necessary – that is what polyphony is about – but the ‘original’ terminology will not be used for the purpose of the present study. When a speaker cites, the quotation is usually subjectively tinged – at least by the verbum dicendi and normally by the context because a speaker always quotes for a particular purpose. Maldonado González observes for the use of the verba dicendi that a speaker, if not using the neutral verb decir ‘say’ to mark

58

59

However, synonyms such as ‘quoted person’ and ‘interviewed person’ for ‘quoted speaker’ or ‘current speaker’ for ‘quoting speaker’ will also be used. But, as explained, the terms énonciateur and locuteur are rejected. For the discussion of further terminological difficulties and proposals for translation into English see Kratschmer’s paper on “Polyphonietheoretische Formalisierung von historischen und Dritteperson-Diskursindividuen in narrativen Texten”.

94 reported speech, influences the addressee’s interpretation of the quoted words by his choice of the speech-accompanying verb: Todos los verba dicendi [...], excepto decir, aportan distintos tipos de información sobre el acto lingüístico efectuado, siendo muchos los que incluyen una información que condiciona directamente la manera en que el receptor interpretará el discurso citado e imponen, por tanto, una cierta lectura al destinatario (Maldonado González 1999: 3559).

Consequently, the current speaker originates a new – ‘new’ because it is subjectively tinged – utterance already only because of the speech-accompanying verb he chooses. Leaving the verba dicendi aside, Reyes explains that the mode of quoting itself may show how, or to which extent, the quoting speaker wants to subjectively tinge the cited words: [...] en [estilo directo] podemos “repetir” expresiones referenciales que no asumimos. Si alguien llama a otro “imbécil”, podemos citarlo en [estilo directo], repitiendo la palabra, sin arriesgar ninguna opinión nuestra [...]. Pero en el [estilo indirecto] podemos elegir la expresión; si decimos otra vez “imbécil” [...] la responsabilidad de la calificación pasa a ser nuestra (Reyes 2002: 20).

Maldonado González also explains that a certain interpretation latitude remains if the current speaker chooses to repeat someone else’s words indirectly: En la cita directa, dado su carácter de contexto opaco, el contenido es siempre responsabilidad del hablante de la situación de enunciación original. En la cita indirecta, en cambio, el contenido puede presentar ambigüedad entre una lectura opaca o transparente60 según las expresiones que en él aparezcan sean interpretadas por el oyente como responsabilidad del hablante de la situación de enunciación reproducida o del hablante de la situación de enunciación reproductora. Dicha responsabilidad, en la transparencia deíctica, depende del punto de anclado; en las transparencias ilocutiva y atributiva, en cambio, se asocian con el hablante (Maldonado González 1999: 3592).

60

The notions of lectura opaca (‘not transparent’) and lectura transparente (‘transparent’) are associated with the descriptions of lectura de dicto and lectura de re: El [estilo indirecto], donde el hablante reformula textos, suele tener una lectura de re, lo que significa que las expresiones referenciales se interpretan dando prioridad a su contenido, a su referencia al mundo, sin atender, al menos de manera explícita, al modo en que fueron enunciadas originalmente. La cita directa, en cambio, exige una lectura atributiva, llamada lectura de dicto, según la cual se atiende a la referencia al mundo pero también a la codificación lingüística misma, que, en estas construcciones citativas, debe coincidir con la original. En la lectura de dicto, la responsabilidad de la expresión (y con ella del punto de vista, valoración, etc.) se atribuye al hablante citado (Reyes 2002: 20).

95 Hence, the current speaker has the possibility to decide whether to assume responsibility for the qouted words or not by choosing on the one hand an appropriate verbum dicendi that serves his communicational purpose and on the other hand the mode of quoting (direct or indirect quoted speech). In this regard, also Gévaudan (2008: 3) observes: “Das Übernehmen und Delegieren von Verantwortung ist ein fundamentaler Aspekt der kommunikativen Motivation polyphoner Äußerungen”. In ScaPoLine (43-51) and in “La polyphonie de la ScaPoLine 2008” (Nølke 2009), the “liens énonciatifs” (Nølke 2009: 25-26) constitute one of the fundamental elements in terms of the responsibility-question: they “établissent des relations entre les différents [être discursifs] et les [points de vue]” (Nølke 2009: 25), that are “entités sémantiques porteuses d’une source qui es dite avoir le [point de vue]” (Nølke 2009: 19). The linguists of ScaPoLine differentiate primarily between “le lien de responsabilité” and the “liens de nonresponsabilité” (Nølke 2009: 25): Le lien de responsabilité est de loin le plus important pour l’analyse polyphonique. La justification de ce postulat réside dans le fait que pour chaque [point de vue] on doit se poser la question qui en est responsable? La détermination des liens fait partie des principes généraux qui dirigent le processus d’interprétation. La notion ‘être responsable de’, […] signifie ‘être la source de’ (ScaPoLine: 44).

Consequently, one of the resulting principles says that an être discursif is responsible for a certain point of view if and only if this ‘discursive being’ is the source of the point of view under question (cf. ScaPoLine: 44). The “êtres discursifs sont les êtres susceptibles d’être tenus responsables des points de vue exprimés” (Nølke 2001: 17). This implies that it is not the current speaker who creates or somehow modifies the point of view expressed in the quoted words. Considering the following examples, it becomes obvious that the quoting author simply lets the quoted person speak (directly), as decir is considered to be the only neutral verbum dicendi (cf. Maldonado González 1999: 3559): (15) “¿Miedo? Nacemos y morimos en este sitio. Estamos acostumbrados a las maniobras, las escaramuzas y todos los problemas. Seguimos con nuestras cosas como si nada”, dice Park, otro pescador cuya casa no resultó dañada en el ataque. La zona vive las últimas horas bajo los tambores de guerra. (El Mundo 26/11/2010)

The same quoted speaker, the same author, but – though using the same speechaccompanying verb – not totally the same neutral mode to indicate direct quoted speech because the journalist evaluates the speaker’s manner of saying; according to him, Park tells [p] ‘defiantly’ (desafiante): (16) El ataque del martes, el mayor sufrido por la aldea desde el final de la guerra coreana en 1953, no sería nada en comparación. “Esperaremos lo que venga en Yeonpyeong”, dice Park desafiante. (El Mundo 26/11/2010)

96 So an adverb describing the manner of saying may diminish/delete the neutral character of decir. The following three examples represent text passages where a person (17), various people (18) or a party (“PP” is the Partido Popular) (19) is indirectly quoted, whereby the author does not modify the point of view expressed in the quoted subordinate clauses. (17) [...] la noticia más leída es que la novia de Ronaldo, Irina algún apellido ruso impronunciable, dice que le han borrado con photoshop el ínfimo tanga color carne con que se hizo una sesión de fotos para la revista ‘GQ’, en cuya portada sale luciendo nalgas contra una rústica pared de Chinchón. (El Mundo 26/11/2010) (18) Los estudiantes dicen que la propuesta apartará a los más pobres de los estudios superiores. El Ejecutivo asegura que se trata de una medida inevitable. (El Mundo 25/11/2010) (19) El PP dice que, según sus cuentas, hay que añadir 17 millones más por cambios en las redes de agua y equipamiento sanitario. (El Mundo 26/11/2010)

In the following text passage (a part of a forum discussion), the person who obviously comments on a newspaper article or various ones indirectly quotes what is said about Maradona, very probably having rephrased the original statements about Maradona he heard: (20) Maradona siempre dijo que el nunca quiso ser un ejemplo. Por favor, en vez de criticarlo, dejarlo en paz. Dicen que quiere llamar la atención pero lo cierto es que es el periodismo el que lo busca constantantemente. Y la gente que lo critica lee las noticias acerca de él. Quién no quiera saber nada de él, que no lea estas noticias. (El Mundo 25/11/2010)

The next example, an interview, represents a use of dicen que ‘they say that’, whereby the persons the speaker claims to quote are indirectly quoted to such an extent that it could be assumed that the speaker expresses his own opinion about himself (and thus actually ‘quotes’ himself), only putting in dicen que for being seen as more modest. Indicating reported speech without truly quoting is forced by pragmatic factors. Incidentally, an interview is the form a journalist uses if he wants to quote himself and the interviewed person directly: (21) [...] he empleado la tira en algo y me ha salido rematadamente mal, y otras veces lo he ventilado con rapidez y ha sido igual de penoso. P[regunta] ¿Harán algún estreno del disco Nick Hornby y usted? R[espuesta] Daremos conciertos en EE UU y Londres. Y promocionaré el disco. Dicen que doy muy buenas entrevistas, así que es usted afortunado. P[regunta] ¿Dónde se conocieron? […] (El País 28/09/2010)

As already shown, polyphony can be marked explicitly (14, 15, 16), but also implicitly (17, 18, 19), or even ‘more implicitly’ (20, 21; cf. also the examples containing supuestamente and aparentemente in chapters 5.1.7 and 5.1.1, re-

97 spectively). So polyphony can be present to different degrees and that is why one speaks about distinct ‘degrees of explicitness’. The most explicit form of polyphony is direct quoted speech (cf. Gévaudan 2008: 4; cf Gévaudan 2009: 101-105). Gévaudan explains that the enunciator – in my terms: the quoted speaker – and his point of view may be present to different degrees: […] l’énonciateur peut être présent de manière plus ou moins explicite […]. Il est clair que l’introduction d’un locuteur s’inscrit au plus haut degré de l’expression explicite, alors que l’expression d’un point de vue dont l’énonciateur n’est pas ouvertement nommé est beaucoup moins explicite. […] on peut établir une échelle de l’expression explicite de la polyphonie (Gévaudan 2009: 103).

This scale, with different degrees of polyphonic explicitness, is associated with different types of quotation. While the first point represents the most explicit form of polyphony, the fifth point represents the most implicit form: 1. The quoted speaker’s point of view is expressed via direct quoted speech: Juan ha dicho: “María está enferma”. 2. The point of view is expressed by a subordinate clause, and hence a verbum dicendi, whereby the quoted person is the subject of the utterance/matrix clause: Juan ha dicho que María está enferma. 3. The quoted speaker and his point of view are explicitly mentioned, whereby the verbum dicendi is not the neutral decir ‘say’: Juan cree que María está enferma. Gévaudan (2008: 5) describes this form of quotation as a border case between indirect quoted speech and a less explicit form of polyphony: “Ein Grenzfall zwischen indirekter Rede und weniger expliziten Formen der Polyphonie bildet die Wiedergabe von Meinungen, Befürchtungen, Hoffnungen, etc.” of the quoted speaker. 4. The quoting speaker expresses the cited person’s point of view without mentioning him, that is, the originator of the utterance ‘Mary is sick’ is not explicitly mentioned: He oído que María está enferma. 5. The most implicit form of polyphony is to be found in sentences where an adverb expressing reportedness (such as allegedly, supposedly or reportedly) is used to quote a point of view that is not the one of the current speaker: María presuntamente está enferma (cf. Gévaudan 2009: 104).61 61

Actually, Gévaudan’s different degrees of polyphonic explicitness which are associated with the different types of quotation are very similar to Volkmann’s explanation of the different modes of quoting a person:

98 As will be shown in chapter 7.1 the condicional, which is not grammatically motivated as in the construction quoted speech-introducing verb in past form + que + subordinate clause containing the conditional form, can be termed an even more implicit form of polyphony so that a sixth degree to the scale of polyphonic explicitness presented by Gévaudan should be added. The ‘very implicit form’ of polyphony would also comprise the quotative use of the conditional form if no reference to the information source is given (cf. chapter 7.1). Actually, one could wonder why Gévaudan (2008; 2009) did not add a sixth degree to the scale of polyphonic explicitness as he himself speaks about “la polyphonie idiosyncrasique”, illustrating this kind of polyphony with the following example from Spanish: “Ayer, he encontrado la madre de Juan. Él esta [sic] enfermo”. Then he explains: L’effet polyphonique résulte ici de la phrase initiale qui introduit les conditions sous lesquelles le locuteur a appris la maladie de Jean et qui incitent à attribuer l’assertion de la seconde phrase au personnage de la mère. D’une part, l’énonciation ne présente aucune marque linguistique de polyphonie, d’autre part, aucune règle spécifique de la conversation y est appliquée (Gévaudan 2009: 121).

So there indeed is a sixth degree of polyphonic implicitness (where the reportive use of the conditional would belong to), and free indirect speech may be considered the seventh degree: in free indirect speech the current speaker adopts the words and the point of view of the speaker who is quoted without explicitly indicating that he quotes another person. In presenting the speech or thoughts of another person via free indirect speech the current speaker is able to retain the original idiomatic qualities, which may be considered a hint in order to identify the polyphonic character of the words. ‘The current speaker behaves as if it were not him who speaks but the other person’ (cf. Genette 1998: 123). Actually, Die Art, wie auf den [inaktuellen Sprecher (quoted speaker)] referiert wird, variiert stark. Es reicht von deutlicher Namensnennung und Markierung der Rolle als [inaktueller Sprecher] in der Form einer ausführlichen Redebegleitung (Ana dice…) über implizitiere Formen der Referenz mit Hilfe von Personaldeiktika, manchmal auch ohne explizite Redebegleitung (z.B. ein nicht auf den [aktuellen Sprecher] referierende[s] yo) bis hin zu noch impliziteren Markierungen einer fremden Sicht, beispielsweise mit Hilfe von Stilisierungsverfahren, wie stimmliche Nachahmung des fremden Sprechers im mündlichen Sprachgebrauch (Volkmann 2005: 33). These words, in turn, imply that it does not seem to be necessary to operate with the notion of polyphony if dealing with the evidential subcategory of quotation and the different modes of quoting. Nevertheless phrases as “implizitere Formen der Referenz mit Hilfe von Personaldeiktika” or “noch implizitere Markierungen einer fremden Sicht” could be replaced by more concrete terms such as ‘implicit polyphony’ and ‘very implicit polyphony’.

99 Gévaudan’s sample sentence could represent free indirect speech. Free indirect speech is also a stylistic device to express irony: Reyes (1996) even considers ‘ironic echoes’ cases of polyphony: La ironía es una cita porque el hablante repite o se hace eco de una proposición ajena, que proviene de un enunicado inmediato [...], o de un lugar común [...], o que representa lo que alguien suele decir, o lo que se suele decir en ciertas situaciones, o incluso lo que se podría decir: el hablante repite la proposición pero la aplica a un estado de cosas contrastante, volviéndola inadecuada, chocante (Reyes 1996: 51).

In case of referring to a particular statement when uttering an ironic statement, the form of polyphony is less implicit than if referring to an imaginative statement, i.e. to something that is usually expressed in certain situations (“lo que se suele decir en ciertas situaciones”). The more implicit the form of polyphony, the more important the consideration of the (extra-linguistic) context in the analysis of the respective polyphonic expression (cf. also Reyes 1984: 88-98). The most important fact for the present study is: while an explicit form of polyphony is to be found on the sentence level, an implicit form is only to be found on the textual level if the context is incorporated in the textual analysis (cf. Gévaudan 2008: 6). This is especially to keep in mind for expressions with quotative value. Besides the quotative conditional, an example for this is the Spanish adverbial locution al parecer ‘as it seems’, which was shown to be used to convey hearsay/quotative readings (cf. chapter 3.6 for a detailed view on Cornillie’s (2007a) analysis of al parecer). To sum up, the more implicit the form of polyphony or – to rephrase it – the more indirect the form of quotation, the more important becomes the context for the analysis of the utterances under consideration. While the different degrees of polyphonic explicitness belong to different forms of quotation and always refer to utterances that are not originated in the current speaker, the following opposition pair refers to the multiplicity of voices in another way: Nølke differentiates between “polyphonie interne” (2001: 23) and “polyphonie externe” (2001: 19). For the aim of the present study, especially for the analysis of inferential expressions, internal polyphony may be of importance as this form of polyphony describes expressions the current speaker uses to explicitly express that it is his point of view he utters. In terms of polyphony one could argue that the notion of internal polyphony is justified when the current speaker can be assumed to have paid attention to his inward voice. In his article “La dilution linguistique des responsabilités. Essai de description des marqueurs évidentiels il semble que et il paraît que” Nølke (1994) analyses the two constructions mentioned in the title of his paper regarding their polyphonic and evidential meaning. According to him, il paraît que [p] has the meaning “it is said/they say” (Nølke 1994: 93). Il semble que [p], on the other hand, is ana-

100 lysed to express inference (Nølke 1994: 93), which in turn represents one example of internal polyphony.62 Regarding the evidential subcategories of the expressions he says verbatim: L’analyse traditionnelle de ces deux marqueurs comme étant du type ouï-dire n’est satisfaisante que pour les emplois simples de il paraît que. Il semble que est d’un tout autre type, à savoir le type d’“inférences”, et même le marqueur il paraît que change complètement de valeurs dès qu’il est modifié morpho-syntaxiquement: il se transforme en un marqueur évidentiel du genre vision, en retrouvant ainsi sa valeur étymologique (Nølke 1994: 93).

He furthermore claims that the concept of polyphony was necessary to distinguish the different evidential meanings of the expressions he analysed (cf. Nølke 1994: 93). I wonder why this would not have been possible if having operated with the different subcategories of evidentiality as reported or inference, but as reported speech is marked by the feature ‘more than one voice is heard’ – a feature inference lacks off (despite one’s inward voice) it sometimes simply seems to be a matter of notional choice. But an analysis in terms of polyphony would only fail if being confronted with the other subtypes of evidentiality such as direct visual evidence. Nevertheless, not only the evidential subcategory of quotation can be arranged with the concept of polyphony but also the subcategory of inference, even though – as far as I can see – it is not really helpful to argue that inference represents internal polyphony. The analyses of inferential expressions would not turn out to be different, if operating with the notion of internal polyphony. Both the modal verb deber (de) and suponer for instance, can be used to express inference: (22) “[...] Creo que veía que podía correr peligro su segundo puesto en París”, dijo Contador, muy satisfecho con la experiencia. Casi tanto como Andy, que dijo: “Ha sido una guerra psicológica de la que salgo con más confianza. Contador debe de estar frustrado porque quería ponerse hoy de líder y no ha podido”. Quedan tres días de Pirineos. (El País 21/07/2010) (23) En declaraciones a la BBC Schumacher ha dado su punto de vista: “Como piloto tienes la capacidad de cambiar de dirección una vez, y eso es lo que hice. Obviamente había espacio suficiente para pasar (porque él lo hizo). No nos tocamos, así que supongo que dejé espacio suficiente para que pasara. Sabía que se estaba acercando, me fui hacia el interior e intenté dejarle claro que se fuera hacia el otro lado, donde había más espacio para él. No lo hizo y la cosa se apretó un poco”. (El País 04/08/2010) 62

The use of the subjunctive mood is therefore also analysed in terms of internal polyphony (cf. Nølke 1985, Nølke 2009), whereby this analysis is restricted to the use of the subjunctive as a means to express modality, more precisely, to express restricted responsibility for the content of the utterance (cf. Gévaudan 2008: 9).

101 (24) “Ya estamos trabajando sobre el próximo ejercicio. Más o menos con el mismo calendario que este año, por lo que supongo que los resultados estarán listos en junio”, dijo Giovanni Carosio, presidente del Comité Europeo de Supervisores Bancarios (CEBS), en una entrevista concedida a Reuters Insider. (El Mundo 18/11/2010) (25) [...] la patronal lleva “dos años trabajando en equipo, buscando soluciones” y ha afirmado que los “argumentos y esos consejos ya están hechos. Están escritos y están en blanco y negro y los tiene el presidente del Gobierno en su despacho, supongo que los tiene en su despacho, desde luego el Gobierno lo tiene. [...]” (El Mundo 25/11/2010)

Here, the underlined conjunctions or connectives such as porque63 (22), así que64 (23) o por lo que (24) and other phrases to indicate the reason for supposing that [p] such as obviamente había espacio suficiente para pasar (porque él lo hizo) ‘obviously there was sufficient space to pass (because he did it)’ (23) and desde luego el Gobierno lo tiene ‘recently, the government has it’ (25) indicate the reason for the inference expressed by deber (de) and suponer. But phrases alone may also establish a logic relation between the reason to draw an inference and the inference itself. Example (25), for instance, illustrates that supongo que introduces the inference [p] – ‘it is in his office’ – while the underlined phrase represents the reason for concluding [p]. For the present study it does not seem to be of help to speak of ‘internal polyphony’ if the same facts can be explained in terms of inference. However, the motivation for linguistically expressed polyphony (of any kind) is said to be the “dilution linguistique des responsabilités” (Nølke 2001: 15). Even though the fact that the speaker’s aim is always to ‘dilute the responsibility’ if quoting another person cannot be totally agreed with, it is obvious that not only the linguistic categories of polyphony and evidentiality are related but also polyphony and modality. More precisely, polyphony and the speaker’s manifestation of his validative attitude (notion adopted and translated from 63

64

Blühdorn/Reichmann (2010) investigate “Modal readings of sentence connectives in German and Portuguese”. They “use the term sentence connective for certain elements traditionally classified as conjunctions, adverbs and particles, which share the semantic function of encoding relations between sentences” (Blühdorn/Reichmann 2010: 15). They primarily analyse sentence connectives that operate at the epistemic (and deonticilloucutionary) level and show that “connections with porque […] usually have epistemic and/or deontic-illocutionary, but only rarely temporal readings” (Blühdorn/Reichmann 2010: 35). In example (22) Spanish porque highlights the inferential use of deber de as “inferential evidence-conclusion relations” (Blühdorn/Reichmann 2010: 20) are expressed. Nølke (2006), for instance, studies the French connector donc ‘so’ in terms of polyphony, as donc implies a relation between cause and effect.

102 Volkmann 2005: 5365) are related because – as mentioned above – usually, a speaker does not decide to quote without reason (excluding the communicative situation in which A asks B what C has told him, and then B quotes C only in order to answer A’s question). With the help of authority quotes a speaker (tries to) strengthen/s his own words. Another reason – especially in the context of newspaper articles – for attributing [p] to another voice clearly is the dilution of responsibility. The question, however, is whether contextual information gives hints about the current speaker’s validative attitude towards [p] and whether it is linguistically expressed. If no additional information is contextually provided, the current speaker/journalist’s attitude is not expressed. Generally, additional information is provided by the verbum dicendi, i.e. the verb which is chosen by the quoting speaker to accompany the words he quotes. This verb conveys the current speaker’s attitude at least to some extent. The author-reader dyad is also backgrounded by another pragmatic factor: in manuals on how to write journalistically, as in the following one, it is often recommended to use quotes because they ‘authenticate the news’: “Zitate verleihen Authentizität. […] Eine Story ohne Zitate, egal ob lang oder kurz, ist öde wie eine Mondlandschaft” (Cappon 2005: 121). So if the journalists are the quoting persons (and not the interviewed ones who may also quote while being interviewed) the instances of quotes are also due to the text type. Gévaudan observes: “Pour le genre article de presse, on peut constater que la polyphonie est un élément très important, vu que les informations y sont souvent présentées sous forme de citations plus ou moins explicites” (Gévaudan 2009: 117). And in her article “Polyphonie et dialogisme dans les genres journalistiques”, Weidenbusch (2009), for example, studies the linguistic expressions of polyphony and dialogism in journalistic texts. She wonders “quelles sont les possibilités analytiques des concepts de polyphonie et de dialogisme66 si on les applique à l’analyse d’un 65

Volkmann explains: Die validative Einstellung […] bezieht sich darauf, wie der Sprecher den Wahrheitswert der Proposition einschätzt. Der Sprecher kann einen Sachverhalt für wahr halten oder für falsch halten, bzw. eine beliebige Position zwischen diesen beiden Extremen validativer Einstellung einnehmen, z.B. einen Sachverhalt für sehr wahrscheinlich, für eher unwahrscheinlich halten oder seine Einstellung an die Existenz anderer Sachverhalte knüpfen […]. Zu einer Einstellungsbekundung kommt es, wenn der Sprecher seine Einstellung sprachlich ausdrückt, im Spanischen beispielsweise durch Verben wie creer, suponer, dudar (Volkmann 2005: 60; cf. also Volkmann 2005: 53-54, 59-61, 454, 456).

66

Weidenbusch’s notion of dialogism is adopted from Bres/Nowakowska (2006), who explain that “la parole de l’énonciateur en acte de langage interagit non seulement avec l’énonciataire mais avec d’autres voix qu’il ne peut manquer de rencontrer et qui le

103 genre journalistique tel que l’éditorial” (Weidenbusch 2009: 145). She works with the French journal Le Monde. In journalistic texts, Weidenbusch argues, different points of view are attributed to different sources “telles que les tiers individuels, les tiers collectifs hétérogènes ou homogènes” (Weidenbusch 2009: 146), that is, to different discourse beings (cf. ScaPoLine: 40). In this respect, Gévaudan analyses text passages containing a synthetic future form. He describes the use of French sera and Spanish deberá in the following examples as instances of implicit polyphony (Gévaudan 2009: 118; emphasis in bold is mine): Elle a dit au Monde qu’elle souhaitait corriger un système “excluant un nombre considérable d’étudiants des classes moyennes” de la possibilité de disposer d’une bourse. L’assiduité qui conditionne le versement des bourses sera mieux contrôlée. (Le Monde, 20/09/2007, p.1) El jefe del estado qiere que todos elles se equiparen al régimen general de la Función Pública. Este año el tesoro público deberá añadir casi 5.000 millones de euros para enjugar el déficit que generan las pensiones de estos ciudadanos. (El País, 20/09/2007, p.3)

In the underlined utterances the speaker is not explicitly mentioned (again), but the use of the future forms is said to lead the reader to the interpretation that the ‘implicitly quoted speaker’ is the speaker already mentioned in the context, namely ‘elle’ and ‘el jefe del estado’. But according to Squartini (2001: 321), in French and Spanish, the synthetic future form is exlusively used to express inference. Nevertheless, I could also find text passages containing será(n) where the synthetic future form appears together with phrases marking quoted speech. Será(n) seems to have the same function as the one explained by Gévaudan: (26) [...] arrestados, que serán puestos a disposición judicial en breve, entraron en la vivienda a primera hora de la mañana. La liberación se produjo pasada la media tarde con el asalto, por parte de los agentes especiales, de la vivienda. La mujer feuillent énonciativement” (Bres/Nowakowska 2006: 44). Hence, the concept of dialogism considers language production in both its cotext and context and is applied to discourse analysis (cf. Weidenbusch 2009: 148, 149, 159). To what extent the concept of polyphony differs from the concept of dialogism, Weidenbusch explains referring to Bakthin: while Bakthin explains that every language use is dialogic in the sense that every linguistic form contains traces of other ‘users’ and their uses of a certain form, “le dialogisme part de l’idée qu’il est inhérent au langage et que le dialogue se trouve partout” (Weidenbusch 2009: 151), that is, the notion also refers to the “relations que tout énoncé entretient avec les énoncés produits antérieurement ainsi qu’avec les énoncés à venir que pourraient produire ses destinataires” (Charaudeau/Maingueneau 2002: 175). However, both concepts are applied to the text level or discourse level (cf. Weidenbusch 2009: 150).

104 también será puesta a disposición judicial, según aseguraron las mismas fuentes. (El País 12/08/2010) (27) Según un informe que acaba de hacer público el Banco de Pagos Internacionales (BPI), Portugal y España son los dos países desarrollados donde este efecto será mayor en los próximos 40 años. (El País 12/08/2010)

While in Gévaudan’s text passages the speaker is mentioned once explicitly and is then implicitly quoted by será and deberá, the examples above seem to have an even more ‘implicit character’: example (27) in particular could be termed ‘more implicit’ because the exact information source is not known as the writer of the ‘report’ (informe) is unknown; nevertheless the use of the synthetic future form seems to go hand in hand with the phrase that indicates another voice. However, in chapter 7.1 it will be shown that the conditional alone (without any reference to the information source) may be used to indicate reported information, which is why a sixth degree to Gévaudan’s scale of polyphonic implicitness should be added. It surely has become obvious that when analysing examples of ‘polyphonic evidentiality’, [a] number of issues become relevant in the interpretation of textual voices […]: the voice introduced as the source, the ways in which the “original” message is presented, the explicit signals of reporting used, and the attitudes thus attributed to the reporting and the reported speaker/writer (Bondi 2009: 84).

If ‘the attitudes attributed to the reporting and the reported speaker/writer’ represent an issue relevant in the interpretation of polyphony, Bondi refers to the linguistic category of modality, more precisely to the reporting speaker’s/writer’s manifestation of his validative attitude concerning [p] or the source of [p]. Indeed, the quoting person can simultaneously express his attitude towards (the source of) [p] or his attitude may be pragmatically inferred from the communicative background. In journalistic articles, in particular, the indication of ‘other voices’ is motivated by an additional pragmatic factor: the author-addressee dyad somehow requires to mention other information sources. However, if, for instance, distance or the validation of (the source of) [p] are expressed simultaneously, that is, linguistically encoded, the examples will also be considered in terms of epistemic modality. Recently, another approach to polyphony has appeared in the cientific literature. Haßler (2010b, 2010c) employs a deictic approach to polyphony, explaining that the creation of multiple deictic centres in narrative texts automatically leads to a multiplicity of voices, that is, to polyphony (cf. also Volkmann 2009). In this sense, one could also argue that a journalist incorporates another deictic

105 centre (or various ones) and hence establishes another origo being the (at least) second zero point a certain [p] originates in.

2.2 Concluding thoughts or proposal for solution: speaker’s perspectivisation as superordinate category In this chapter, the relationship between evidentiality and other linguistic categories is not only summarised but also backed up by further studies that are concerned with the interrelationship of the categories which, in turn, are all related to evidentiality, e.g. modality and deixis or subjectivity and deixis. Hence, it will be argued here that the linguistic domains of evidentiality, epistemic modality, subjectivity, polyphony and deixis are actually subdomains of one superordinate category. Boye (200667, 2010a), for instance, has already shown that both epistemic modality and evidentiality should be considered subcategories of one and the same superordinate one, whereby in the present study it will be argued that not only evidentiality and epistemic modality should be subsumed below the superordinate category but also deixis, polyphony and subjectivity. So while the preceding chapters have shown that evidentiality and deixis, evidentiality and subjectivity, evidentiality and polyphony, and evidentiality and epistemic modality may overlap, in the following it will be shown that deixis, subjectivity, polyphony and epistemic modality may themselves overlap. While evidentiality and epistemic modality have the concept of a deictic source in common (Diewald/Smirnova 2010a: 87), it is also deixis that relates subjectivity and evidentiality: Cappelli (2007) argues in support of integrating ‘the deictic category of subjectivity’ into the understanding of evidentiality. Hence, she explains that evidentiality is deictic in terms of subjectivity although an ‘evidentialised’ proposition may be based on intersubjectively perceivable evidence. In her study on verbs of cognitive attitude she says the following about the relationship between deixis, subjectivity and evidentiality: Subjectivity is a frequently used term and its sense varies within the different traditions of research. Stein (1995) identifies five major senses. […] third, subjectivity as a deictic for the speaking ego […] I use subjectivity in this sense, in a way akin to Stein’s (1995) third sense, as the expression of an ego and its pervasiveness in the evaluation, be it performed or reported. In this sense, verbs of cognitive attitude are means for the expression of subjectivity (Cappelli 2007: 131).

67

I am greatly indepted to Kasper Boye, who provided me his dissertation manuscript. The page numbers are taken from the pdf-file. It was published in 2012: see Boye 2012.

106 This ‘deictic description’ of subjectivity can be arranged with evidentiality (cf. also chapter 2.1.3). So Cappelli further explains: This view of subjectivity can be integrated into the evidential dimension. If the speaker can encode reference to the external source of information, why cannot he/she encode reference to the absence of such external evidence? […] The speaker can signal that he/she is uttering a personal judgement based on no other evidence but his/her own personal evaluation, his/her feeling or some other type of vague evidential information. This sort of evidence is here labelled “affective evidence”, and it is taken to include impressions, “irrational judgements” and any type of evidence depending on the ego of the evaluator or for which the evaluator cannot provide a precise definition (Cappelli 2007: 131-132).

The last mentioned sort of evidence, affective evidence, is clearly the ‘most subjective’ sort of evidence a speaker may provide.68 A conclusion which is inferentially marked by the Spanish modal verb poder is also subjective: it is more subjective than a statement that is inferentially marked by deber but less subjective than a statement containing creo ‘I believe/think’. The evidential dimension that is present in the verb of cognitive attitude believe – and as it will be shown for its Spanish equivalent creer (cf. chapter 5.2.1) – is affective, that is, the most subjective sort of evidence that exists. Nuyts (2001b: 112) also contrasts believe and think with know, guess or suppose, for instance, which are said to belong “in the realm of the traditional evidential category of inferentiality”. Believe and think, in contrast “have evidential meanings in the realm of (inter)subjectivity” (Nuyts 2001b: 112). The modal verbs deber and poder are used to express inferential, but less subjective statements, whereby the former is even less subjective than the latter because marking a proposition inferentially by the use of poder signals that the speaker could have equally drawn another conclusion and he (subjectively) chose the one he expresses (cf. chapter 6). Hence, the speaker’s commitment to the likelihood that the proposition is true is not as strong as by the use of deber. Deber is generally used when “the speaker expresses a stronger commitment to the likelihood that the proposition is true” (Cornillie 2007a: 202). Nevertheless, by the use of both poder and deber “[…] it is the speaker who evaluates the different premises and decides on the final conclusion” (Cornillie 2007a: 192). Generally speaking, “poder expresses weak commitment

68

Cappelli (2007: 170) explains that affective evidence “is generally not included in the discussion of evidentiality” but that she uses the terms of ‘affectivity’ or ‘affective evidence’ to describe “some sort of subjective feeling or impression, [which] is generally opposed to evidentiality. […] the label ‘affective evidence’ is used in a technical term to indicate various types of vague or subjective evidential information” (Cappelli 2007: 319).

107 (possibility)” and “deber conveys intermediate69 speaker commitment (strong possibility)” (Cornillie 2007a: 254). Dendale/Tasmowski (1994) have intensively studied the French modals devoir and pouvoir and conclude that both modal verbs express subjective inferences, thus relating evidentiality and subjectivity: Les énoncés comprenant pouvoirE, tout comme ceux comprenant devoirE, sont l’indice d’un processus mental qui consiste à créer des prémisses, à en inférer des conclusions et à évaluer ces conclusions pour n’en retenir que partie. De telles conclusions sont toujours subjectives puisque c’est le locuteur qui décide des hypothèses à retenir. Alors qu’avec devoirE, le résultat est une sélection unique, pouvoirE indique, le plus souvent sur le mode implicite, que le locuteur reste a priori ouvert à d’autres éventualités. (Tasmowski/Dendale 1994: 55; emphasis in the original).

So evidentiality and subjectivity do not have to exclude each other. An affective use of creer, for instance, is more subjective, while a logical conclusion drawn out of evidences may be less subjective. In line with Cappelli (2007), the assumption that evidentiality and subjectivity are mutually exclusive is rejected.70 Cappelli considers the categories to be “two scalar dimensions with fuzzy boundaries” (Cappelli 2007: 133), whereby reliable evidence implies less subjectivity and the other way round. Consequently, subjectivity itself should be regarded as a ‘scalar-like notion’ because the presence of subjectivity depends on the evidentially used lexical item itself (whether the subjective dimension is already encoded by the lexical element, such as in the verbs of cognitive attitude, or not) and on further contextually provided information. In summary, subjectivity and evidentiality are clearly related categories (cf. also chapter 2.1.3). As Cappelli (2007: 131-132) has shown, deixis relates these categories. Nuyts (2001a: 383) even argues that subjectivity is “a separate evidential qualification”. This implies that he subordinates subjectivity to evidentiality. And subjectivity, being an evidential qualification, “can (but need not) be ex69

70

The strongest speaker commitment is conveyed by the use of tener que, which correlates with the fact that it is used to express deductive inferences (cf. Cornillie 2007a: 155). This is explained to be in contrast to the traditional view: Evidentiality and subjectivity have been traditionally seen as competing dimensions. On the one hand, there would be subjectivity, i.e. the evaluator’s point of view, the expression of his/her stance towards the state of affairs; on the other hand, there would be evidentiality, the evidence that can back an evaluation or an assertion (Cappelli 2007: 132-133).

108 pressed jointly with the epistemic one”, he states (Nuyts 2001a: 384). So the dimension of subjectivity is not a distinction within the epistemic domain; it is said to be “an independent evidential-like qualificational category” (Nuyts 2001a: 398). This, in turn, means that Nuyts does not need to deal with two types of epistemic modality, namely objective and subjective epistemic modality – whereby he even rejects Lyon’s (1977, 1995) notion of objective modality – “but with an interaction of an epistemic with an evidential qualification” (Nuyts 2001a: 386). According to him, which supports Cappelli’s (2007) line of argumentation, “mental state predicates systematically express subjectivity” (Nuyts 2001a: 390). But modal adjectives and modal auxiliaries are also shown to (possibly) involve the dimension of subjectivity (cf. Nuyts 2001a: 389, 390, 392). In my view, Nuyts’ (2001a) argumentation for subjectivity only being an evidential qualification – and not also an epistemic one – is not absolutely convincing (cf. also Pietrandrea 2005 and Diewald/Smirnova 2010a). Since his paper “The modal confusion: on terminology and the concepts behind it” (2005) it is clear that even Nuyts himself does not endorse this view any longer. There he explains that both epistemic modality and evidentiality “feature a structural distinction between what may be called ‘subjective’ and ‘intersubjective’ uses” (Nuyts 2005: 26). “In the epistemic system”, he (2005: 26) explains, “this meaning [of subjectivity and/or intersubjectivity] is […] realised by choosing a form type which allows an impersonal subject (indicating intersubjectivity) or a first person subject (indicating subjectivity)”. So subjectivity is not any longer considered only an evidential qualification. Indeed, epistemic modality is also said to be a subjective category (cf. Palmer 1986: 16; Palmer 2001: 33; Coates 1995: 59; Pietrandrea 2005: 36-39; Hoye 2008: 160). Although it was already pointed out that evidentiality and subjectivity are clearly related categories, I would like to focus on the notional diversity of ‘subjectivity’ addressed by Pietrandrea (2005) in Epistemic Modality. Functional properties and the Italian system. It is observed that “[a]s far as the use of subjectivity in the characterization of modality is concerned, it is possible to recognize three different meanings of them” (Pietrandrea 2005: 36). Subjectivity is often synonymously used with ‘performativity’. In this case it is synonymous with the indexical expression of the speaker (Pietrandrea 2005: 36) and refers to “the self-expression a speaker performs when using a modal expression” (Pietrandrea 2005: 37). Subjectivity may also be synonymous with ‘metapropositionality’ (Pietrandrea 2005: 36): the epistemic subjective meanings of modal expressions “express the attitude of the speaker towards the propositional content” (Pietrandrea 2005: 37; emphasis in the original). So modal expressions are said to fulfil the function of propositional operators because “they semantically modify an entire proposition and not just a part of it” (Pietrandrea 2005:

109 37). The third synonym for subjectivity is ‘genuine epistemicity’ (Pietrandrea 2005: 38). Genuine epistemicity contrasts with objectivity of inferentiality (Pietrandrea 2005: 36). This third kind of understanding of subjectivity is considered highly similar to performativity, or rather, to a “more general sense of ‘performativity’” because it also refers to the expression of the speaker’s belief (Pietrandrea 2005: 39). Consequently, Pietrandrea (2005: 39) develops the following working definition: modality is “the performative category expressing the speaker’s genuine opinion towards the modalized proposition”, thus combining all three synonyms that are all subjective in character. So it seems to be beyond question that epistemic modality is highly subjective.71 Lyons (1977: 847) explains that epistemic modality represents one aspect “of the expressive, or indexical, function of language. […] it is an indication of the speaker’s opinion or judgement that is involved”. By speaking of an ‘indexical function’ and of ‘the speaker’s opinion or judgement’ Lyons seems to relate deixis as well as subjectivity to the notion of epistemic modality. Indeed, he explains that only “few of our everyday statements have […] dispassionate, totally non-subjective character” (Lyons 1995: 331). By ‘subjectivity’ he, inter alia, understands “the property (or set of properties) of being […] a subject of consciousness (i.e., of cognition, feeling and perception)”, explaining that subjectivity “denotes the property of being […] a ‘thinking entity’”, which is to identify “with the self or the ego” (Lyons 1995: 337). So in explaining subjective modality, which for him “is much more common than objective modality in most everyday uses of language” (Lyons 1995: 330), he makes use of the terms ‘self’ (cf. Frawley 1992) and ‘ego’ (Cappelli 2007: 131), that is, using terms which are related to the concept of deixis. When Lyons (1995: 331) explains that “[s]ubjective epistemic modality is nothing other than […] a locutionary agent’s qualification of his or her epistemic commitment” this is reminiscent of Diewald’s (1999) argumentation that by the deictic use of a modal verb a proposition is qualified by a factuality value coming from the origo. The speaker, as 71

Coates (2003), analysing epistemic expressions in English everyday spoken interaction, also emphasises the subjectivity of epistemic modality: in everyday spoken interaction, epistemic modality is used to convey the speaker’s attitude to the proposition, not to convey some objective truth. It is also important to understand that epistemic modal forms can express the whole range of attitudes from confidence to doubt. In other words, an utterance such as I’m sure they’ll come, is as modalized as an utterance such as maybe they’ll come. The first conveys the speaker’s confidence in the proposition “they will come”, the second conveys the speaker’s doubt, but in both cases the speaker chooses to mitigate the force of the proposition (Coates 2003: 331).

110 Lyons (1995: 253-254) also argues, expresses a particular attitude towards a proposition, and Lyons calls this attitude ‘epistemic commitment’. The kind of deixis that, according to him, is very close to subjective modality is ‘secondary deixis’ (Lyons 1995: 311). Logically, there is also ‘primary deixis’. Primary deixis is what is usually known as pure deixis (cf. Eguren 1999): “[p]rimary deixis is of the kind that can be accounted for in terms of gestural reference within the framework of the deictic context” (Lyons 1995: 310). Secondary deixis, in contrast, is said to involve “displacement or reinterpretation of the spatio-temporal dimensions of the primary deictic context” (Lyons 1995: 310). He illustrates this rather abstract statement by the following example, simultaneously relating secondary deixis to subjectivity: […] if speakers are holding something in the hand they will normally use ‘this’, rather than ‘that’, to refer to it (by virtue of its spatio-temporal proximity). If they say What’s that? in such circumstances, their use of ‘that’ will be indicative of their dislike or aversion: they will be distancing themselves emotionally or attitudinally from whatever they are referring to. This is but one example of one kind of secondary deixis (Lyons 1995: 310-311).

Hence, epistemic modality is not only subjective but also deictic – according to Lyons, ‘secondarily deictic’. Even though it was already indicated above that epistemic modality and deixis are related categories, the notion of secondary deixis represents, to my mind, a ‘dilution’ or rather ‘overexpansion’ of the notion of deixis. With this understanding of deixis expressions that are clearly neither deictic nor used deictically would be allowed to be labelled as ‘deictic’. Hoye (2008: 160), who studies evidentiality in discourse, also considers epistemic modality as “fundamentally subjective” and even rejects the assumption “that epistemic modality may aspire to objectivity” (Hoye 2008: 160) because, according to him, “in natural language and discourse, objectivity in any absolute, epistemological sense, is unsustainable” (Hoye 2008: 160; cf. chapter 2.1.1 of the present study). In the following it will briefly be summarised to what extent polyphony and evidentiality may be said to be related categories: Ducrot (1984) defines polyphony as the presence of more than one point of view, attitudes or positions in an enunciation, whereby the (original) enunciators are responsible for their point of view – not the current speaker who refers to those points of view (cf. Ducrot 1984: 204, 205, 214). The speaker is only responsible for his own enunciation and for the way of having arranged the different points of view: “Le locuteur, responsable de l’énoncé, donne existence, au moyen de celui-ci, à des énonciateurs dont il organise les points de vue et les attitudes” (Ducrot 1984: 205). Dendale (1992) labels polyphony an epistemic notion (!):

111 La valeur épistémique […] se laisse décrire […] comme une constellation de plusieurs attitudes d’énonciateurs sur l’information véhiculée par l’énoncé. Cette pluralité d’attitudes de divers énonciateurs est exactement ce à quoi réfère aussi le terme de polyphonie (Dendale 1992: 105; emphasis in the original). […] la polyphonie est la présence simultanée de plusieurs voix dans le même énoncé (Dendale 1992: 106; emphasis in the original).

It is – in my view – not comprehensible why an epistemic value (of a certain expression) can be described as ‘a constellation of several attitudes/points of view’. But it is this “pluralité d’attitudes” where Dendale sees the relation to polyphony so that “le concept de polyphonie peut recevoir, entre autres, une interprétation spécifiquement épistémique” (Dendale 1992: 119). An epistemic marker is said to have two properties: Un marqueur épistémique est un élément linguistique qui indique (1o) la fiabilité d’une information et/ou (2o) le type de garanties (ou d’evidence) dont quelqu’un dispose pour étayer son affirmation (Dendale 1992: 105).

With this definition Dendale applies the property of evidential markers to epistemic markers. This becomes even clearer when he mentions the conditionnel (journalistique) as an example “de marqueur à fonction épistémique d’Emprunt” (Dendale 1992: 111), whereby he applies an epistemic function to the evidential subcategory reported. This is true for the conditional but it should be nevertheless notionally distinguished between epistemic modality and evidentiality. As in the present study it is notionally differentiated between epistemic modality and evidentiality a notional overlap between the evidential subcategory reported and polyphony is assumed – not a general overlap between epistemic modality and polyphony. The reason is the following: as long as the current speaker/journalist (only) introduces (an)other point(s) of view or attitude(s) he does not indicate his own validative attitude with regard to the point(s) of view mentioned. If no additional information is contextually provided, the current speaker/journalist’s attitude is not expressed. One exception – and perhaps it is the only one – indeed is the journalistic conditional: here, evidentiality, polyphony and epistemic modality overlap (cf. chapter 7.1). Dendale (1992) treats “La polyphonie comme notion épistémique” only because he subsumes expressions that indicate reported evidence below the heading of “marqueur épistémique” (cf. Dendale 1992: 105, 109). Even though this subordination is not considered correct, it is tolerable in the case of the journalistic conditional. Nevertheless, it is doubtful that this line of argumentation could be likewise applied to other polyphonic expressions. That polyphony and evidentiality must have something in common show different studies by Sullet-Nylander, who analyses newspaper headlines (that

112 contain the conditionnel journalistique or other evidential/polyphonic expressions), in terms of polyphony (2002) as well as in terms of evidentiality (2006). The polyphonic approach and the evidential approach to the analysis of reported information are related because within both approaches the different points of view which the speaker/journalist introduces may be described as different deictic centres the original and then reported statement originated in. Haßler (2010b, 2010c), for instance, employs a deictic approach to polyphony. According to her, the creation of multiple deictic centres results in a multiplicity of voices, i.e. in polyphony. Hence, deixis and polyphony may also be considered related categories. In “Point de vue et évidentialite” Dendale/Coltier (2003) examine possible connections between the notion of evidentiality and the notion of “point de vue”. They draw the conclusion that certain linguistic forms are well analysed in both terms – in terms of evidentiality as well as in terms of polyphony: Ce que la comparaison de ces deux notions et de ces deux phénomènes linguistiques a clairement montré […] c’est que les domaines de recherches que constituent l’étude de l’évidentialité et la théorie de la polyphonie – qui comprend la notion auxiliaire de [point de vue] – sont en intersection: certains éléments langagiers, comme le conditionnel épistémique [ou “journalistique”] ou devoir inférentiel par exemple, se prêtent aussi bien à une analyse en termes de polyphonie et de points de vue, qu’à une analyse, tout autre, en termes d’évidentialité (Dendale/Coltier 2003: 126).

In the present study it is preferred to analyse the linguistic expressions that are the subject under discussion in terms of evidentiality as evidentiality as a descriptive category provides a more extensive set of technical terms. It could be nevertheless useful to refer to the different degrees of polyphonic explicitness if dealing, for instance, with the reportive use of the condicional. After reading the following statement by Diewald/Smirnova (2010a: 76): “We argue that both categories [epistemic modality and evidentiality] are in principle independent of each other, though closely related”, one gets the impression that they align themselves with many other scholars (cf. chapter 2.1.1). But then Diewald/Smirnova make clear that for them epistemic modality and evidentiality are categories that, on the one hand, share certain features but, on the other hand, are separated by other characteristics. They list in detail the following characteristics that evidentiality and epistemic modality share (a-c), the ones that distinguish evidentiality (e) from epistemic modality (d) as well as the feature that represents the essential contrast between both categories (f): (a) Epistemic modality and evidentiality are semantic-cognitive domains which are concerned with the speaker-based evaluation of the ontological status (the kind of existence) of a described event/situation.

113 (b) Both are deictic categories, i.e. they make a speaker-based evaluation of the proposition concerning each dimension’s particular semantic domain. (c) Both share the concept of a deictic source, the origo. (d) Epistemic modality concerns the linguistic encoding of a deictic factuality judgment of the speaker concerning the described event. (e) Evidentiality concerns the linguistic encoding of the information source, i.e. the type of source the speaker adduces for the described event. (f) The essential contrast between them is the fact that the modal source in epistemics is not explicitly expressed, while it is encoded in the evidentials as the secondary predication (Diewald/Smirnova 2010a: 87).

So Diewald/Smirnova (2010a: 81) see the “central distinctive feature between evidentials and epistemics […] in the existence of a secondary event (the evidential source) in the former”. The feature that evidentiality and epistemic modality clearly share, in contrast, is that a modalised proposition as well as an ‘evidentialised’ proposition contains a speaker-based evaluation of the ontological status of an event (cf. Diewald/Smirnova 2010a: 87). Hence, there is not only a deictic use of modals (cf. Diewald 1999) but also a deictic use of evidential expressions, which was already shown by Jakobson: Jakobson (1971) explains that evidentials also need an origo as reference point. For him (1971: 135-136) evidentials are ‘shifters’, whereby “the general meaning of a shifter cannot be defined without a reference to the message” (Jakobson 1971: 131; cf. also 134). So shifters are linguistic signs that must be related to an origo; they need to have a reference point in order to be correctly understood. The most striking point of Diewald/Smirnova’s (2010a) line of argumentation is that they argue that epistemic modality as well as evidentiality is a deictic category, while Haßler (2010a, 2010d) and Wachtmeister Bermúdez (2005) argue that it is only evidentiality that can be termed a deictic category. They even use deixis to distinguish between epistemic modality and evidentiality. The point is that a speaker-based evaluation of a proposition – be the evaluation epistemic or evidential – originates in the origo and both kinds of evaluation have their goal in the proposition. Thus, both categories indeed share the concept of a deictic source. In the following it will briefly be outlined why a few linguists use deixis to distinguish between evidentiality and epistemic modality and why other linguists explain that both categories share the concept of a deictic source. Diewald/Smirnova (2010a) and Diewald (1999) describe a modal or evidential expression as ‘deictic’ because it originates in the speaker’s origo and has its goal (or reference point) in a proposition that is modalised or ‘evidentialised’. Haßler (2010a, 2010d), in contrast, only describes evidential expressions as deictic because their meaning depends on the speaker. Haßler’s understanding of

114 deixis is more ‘classical’ than Diewald’s and Diewald/Smirnova’s. So the reason for the different ‘deictic understandings’ of epistemic modality and evidentiality is due to the different understanding of deixis itself. Hence, epistemic modality and evidentiality are clearly related categories. In languages that do not possess a grammatical system of evidentials the ‘degree of relatedness’ depends on contextually provided information. It is, nevertheless, interesting how often the relationship between epistemic modality and evidentiality is dealt with, without assuming a superordinate category. As Boye observes, only “[…] a couple of works take evidentiality and epistemic modality to be separate categories that are, however, both subcategories of the same superordinate category […]” (Boye 2010a: 15; my emphasis). In his paper on “Semantic Maps and the Identification of Cross-Linguistic Generic Categories: Evidentiality and its Relation to Epistemic Modality” Boye (2010a) explains that the different claims about evidentiality and epistemic modality concerning their relation are “partly a result of the lack of empirically based criteria for identifying cross-linguistic generic categories” (Boye 2010a: 15). In other words, as long as there is no superordinate category for evidentiality and epistemic modality, how can Aikhenvald (2004) claim that evidentiality is a category in its own right, how can others claim that evidentiality includes epistemic modality (cf. Papafragou 2000) or that epistemic modality includes evidentiality (cf. Palmer 1986), or how can van der Auwera/Plungian (1998) and Diewald/Smirnova (2010a) explain that both categories are separate but sometimes overlapping? Boye emphasises the fact that the range of meaning regions such as “direct evidence, indirect evidence, certainty, probability and epistemic possibility can all be defined in terms of a notion like the philosophers’ ‘justificatory support’” (Boye 2010a: 16; cf. Boye 2006). Thus, he outlines that epistemic modality and evidentiality are comprised by a superordinate category and that both categories only then can be described as overlapping, separate etc. So according to Boye (2006: 1), “[t]he area of linguistic phenomena covered by the terms ‘modal’, ‘evidential’ and ‘epistemic’ is in need of both terminological and substantial clarification”. He therefore argues for the introduction of “a third conceptually coherent meaning domain, an epistemic meaning domain, and a third corresponding descriptive category, a category of epistemicity” (Boye 2006: 2) in order to clarify the area that is covered by the three terms. Within the epistemic meaning domain he differentiates between two types of meanings, whereby the one type of meaning “can be narrowly and precisely defined in terms of a concept of ‘epistemic support’” and the other one as ‘epistemic justification’ (Boye 2006: 2). The concept of epistemic support comprises meanings of epistemic moods, but in Boye’s terms, ‘epistemic support meanings’, while the concept of ‘epistemic justification’ comprises ‘evidential mean-

115 ings’ (cf. Boye 2006: 2). In both concepts the epistemic component is of high importance to distinguish epistemic support meanings and epistemic justification meanings – or evidential meanings – from kinds of support or justification that are not epistemic. The concept of epistemic support excludes “meanings that can be characterized in terms of ‘moral support’ or another kind of ‘non-epistemic support’” (Boye 2006: 17). The concept of epistemic justification excludes “meanings that can be characterized in terms of ‘moral justification’ or another kind of ‘non-epistemic justification’” (Boye 2006: 20). So Boye subsumes both the domain of epistemic modality and the domain of evidentiality below a superordinate domain, i.e. the epistemic meaning domain. He then introduces the concept of ‘justificatory support’, which is said to represent a generalisation of the concept of ‘epistemic support’ and the concept of ‘epistemic justification’: All these meanings [the epistemic support meanings and the evidential meanings], and thus the epistemic meaning domain and the descriptive category of epistemicity, can be narrowly and precisely defined in terms of a concept of ‘justificatory support’, in so far as the concept is intended as a generalization of the concepts of ‘epistemic support’ and ‘epistemic justification’ (Boye 2006: 2).

Boye’s epistemic meaning domain with its corresponding descriptive category and the different meanings it comprises together with the all-comprising concept of ‘justificatory support’ can be schematised as follows:

Epistemic Meaning Domain / Category of Epistemicity

Meaning1

Meaning2

Epistemic Support

Epistemic Justification

Epistemic – support Meanings

JUSTIFICATORY SUPPORT Fig. 6: Boye’s epistemic meaning domain (my scheme)

Evidential Meanings

116 The epistemic meaning domain with its descriptive category of epistemicity is thus characterised by two properties: 1. It is conceptually coherent in that the linguistic meanings it comprises can all be defined in terms of a concept of justificatory support. 2. It is conceptually complex in that the linguistic meanings it comprises fall into one of two conceptually coherent subdomains: 1) a domain which comprises linguistic meanings that can be defined in terms of a concept of epistemic support, and 2) a domain which comprises linguistic meanings that can be defined in terms of a concept of epistemic justification (Boye 2006: 2).

The concept of ‘epistemic support’ is defined as lacking the subjective component of degree of speaker commitment and as neutral “with respect to the ‘certainty-uncertainty’ distinction” (Boye 2006: 16). The degrees of commitment being excluded, the degrees of epistemic support are said to specify different epistemic-support meanings: Boye firstly distinguishes between “the meaning of full epistemic support” and the meaning of “less than full epistemic support”. Meanings of full epistemic support convey full commitment or full certainty or rather the “maximum degree of commitment/certainty”. Meanings of less than full epistemic support consequently “correspond to what may be characterized in terms of ‘less than full commitment/certainty’ or ‘non-maximum degree of commitment/certainty’” (Boye 2006: 17). If the meaning of a certain element conveys less than full epistemic support it can again be differentiated between different degrees, namely between “partial epistemic support and absent (positive as well as negative) epistemic support” (Boye 2006: 17). This is how the (less than) full epistemic-support meanings are defined. Evidential meanings are said to constitute an evidential subdomain of epistemic meanings (Boye 2006: 21). Evidential meanings that convey ‘epistemic justification’ – but no kind of justification that is ‘non-epistemic’ – are characterised by common types of direct and indirect evidence, whereby Boye excludes the subcategory of quotation from his study: he explains that ‘reported’ or ‘hearsay’ evidentials convey the meaning of reportive evidence, while ‘quotative’ evidentials differ from other evidentials in so far as “[t]hey do not take simply a ‘proposition’ or a piece of information in their scope, but a whole verbatim speech act” (Boye 2006: 20).72 That is why Boye considers genuine quotatives and quotative meaning non-evidential. However, I (2012a) have shown elsewhere that quotations may be evidentially used to apologetically support another statement. A quotation may consequently take a proposition in its scope.

72

I would like to draw attention to the fact that the use of quotatives or quotations does not presuppose a real speech act that is quoted only for the purpose of quoting.

117 In my opinion, there are to be found quotations that function in the same way as, for instance, modal verbs. Based on Diewald’s (1999) relational structure to illustrate the deictic use of modal verbs that apply a modal-deictic value to a proposition and on Diewald/Smirnova’s (2010a) scheme to capture the deictic nature of evidential expressions, it will be illustrated that a speaker can use a quotation to apply an evidential value to a proposition, that is, using the quotation to ‘evidentialise’ a proposition. This will be shown with the help of contextually provided information. Sometimes a proposition [p] is to be found in context, which is ‘affected’, i.e. ‘evidentialised’, by this quotation. Although it is reasonable to assume that a speaker never quotes another speaker for no reason and that (most) quotations are used to support another statement, in the following I will stick to examples where the linguistic context provides information to clarify the fact that quotations should be analysed in terms of ‘justificatory support’. Two examples which are taken from Spanish detective novels will be offered. So the language data provided should be described in terms of ‘feigned oral speech’. As they are dialogic in nature, they well illustrate that – in speech – quotations may be used to apologetically support a certain proposition. The ‘evidentialised’ proposition, that is, the proposition that is affected by the quotation is marked by a postpositioned ‘[p]’. The following two examples represent instances of quotations that are used for a justificatory support [p]: (28) - Era un chico alto y rubio. Parecía extranjero, pero no lo era. Tenía acento, aunque no creo que fuera andaluz o murciano. [...] Tal vez era manchego, pero de la parte del sur. [...] - ¿Vivía aquí? - No. Estaba de paso [p]. Me dijo que había trabajo en Holanda. En la Philips de La Haya. Es todo lo que sé. (Tatuaje 39)

The quotative part introduced by Me dijo que ‘He told me that’ does support [p] because the speaker uses the indirectly quoted words to attest his proposition No [vivía aquí]. Estaba de paso ‘No [he did not live here]. He was passing through’. This represents the proposition [p] that is ‘evidentialised’ by means of the quotation. In the next dialogue passage speaker A seems to use the quotative part to support his own proposition [p2] as well as the proposition [p1] uttered by his interlocutor (speaker B) using the hearsay expression se habló de que ‘it was told that’: (29) [A] - [...] Negocio redondo. Eso le dio los primeros millones, que empleó para poner en marcha todo lo demás. Poner en marcha es un decir, porque tuvo la habilidad de asociarse con gente emprendedora[...]

118 [B] -¿Qué socios? [A] - Básicamente dos: Planas y el marqués de Munt. [B] - Eso suena a mucho dinero [p1]. [A] - A mucho dinero y a muy buenos padrinos [p2]. Durante un tiempo se habló de que el alcalde estaba detrás de ellos. (Los Mares del Sur 39)

As examples (28) and (29) show, quotations can be said to support [p]. Contextually provided information illustrates that speakers have a special purpose in quoting someone (particular or non-particular) – a purpose that goes beyond the (pure) relativisation onto another epistemic centre and the purpose of communication associated with this relativisation (cf. Volkmann 2005). I agree with Boye (2006) in that quotations may take more than a proposition in their scope but from a functional perspective they nevertheless fulfil an evidential function: they provide justificatory support for (a) certain proposition(s); they have a “función probatoria o ‘evidencial’” (Reyes 1996: 10). So by adopting a functional point of view it was shown that quotations do not have to be excluded from a study of evidential meaning, whether or not they simply take a proposition or a piece of information in their scope. As already mentioned (in chapter 2.1.1), Chafe (1986) – dealing with evidentiality in its broadest sense – describes the linguistic encoding of evidence as one of “a range of epistemological considerations” (Chafe 1986: 262). As Chafe, certainly more unconsciously than consciously, hereby imposes a superordinate on evidentiality, Boye (2006: 26) considers Chafe’s study a precursor of his own study. Another study that describes evidentiality and epistemic modality as subcategories of the same superordinate category is “Layers and operators in Functional Grammar” by Hengeveld (1989, 200573). If it now seems as if I were losing the focus on the superordinate-category problem, it is due to the Functional Grammar perspective Hengeveld adopts and Functional Grammar terminology he consequently uses. The main goal of Hengeveld’s paper is to refine the clause model which is proposed by Dik (1978)74 and used in Functional Grammar. According to Hengeveld, it “should be adapted in such a way that a number of different layers can be distinguished” so that “a proper treatment of modality” is made possible (2005: 1). The basic idea he represents 73

74

Hengeveld’s paper was originally published in the Journal of Linguistics 25 (1989) and then again in Crucial Readings in Functional Grammar, edited by Anstey/Lachlan Mackenzie (2005). The work Hengeveld mainly refers to is: Dik, Simon C. (1978): Functional Grammar. Amsterdam: North Holland. As I did not consult Functional Grammar itself, this bibliographical reference will not show up in the bibliography of the present study.

119 is that every utterance can be analyzed at two levels: the representational (Bühler 1934) and the interpersonal (Halliday 1970) level. At the representational level a SoA [state of affairs] is described in such a way that the addressee is able to understand what real or hypothesized situation is referred to. At the interpersonal level this situation is presented in such a way that the addressee is able to recognize the communicative intention of the speaker. Thus the representational level is concerned with the narrated event, the interpersonal level with the speech event […] (Hengeveld 2005: 2).

Usually in Functional Grammar predications are said to have a descriptive function and a content function. Predications are said to not only give a description of an external situation the speaker tells about but to also represent “the propositional content or message unit processed within that speech act” (Hengeveld 2005: 1). Hence, what Hengeveld proposes is to substitute the terms ‘descriptive’ and ‘content function’ by ‘representational’ and ‘interpersonal function’. Arguing from a Functional Grammar point of view, different classes of operators that operate on the representational or interpersonal level of an utterance are distinguished and said to comprise certain grammatical means: 1. Predicate operators specify additional properties of the set of states of affairs designated by a bare predication (Hengeveld 2005: 6). 2. Predication operators “locate the [states of affairs] in a real or imaginary world” (Hengeveld 2005: 6), “without affecting the properties of the [states of affairs] designated by a predication” (Hengeveld 2005: 8). 3. With proposition operators “the speaker specifies his attitude towards the (truth of the) proposition he puts forward for consideration” (Hengeveld 2005: 6). So these operators express the speaker attitude towards the content of his speech act; they are not concerned “with the occurrence of the event to which reference is made within that speech act” (Hengeveld 2005: 7). Instead, by using proposition operators, the speaker specifies “to what degree he feels committed to the truth of the content he transmits” (Hengeveld 2005: 14). 4. And with illocution operators “the speaker modifies the force of the basic illocution of a linguistic expression” (Hengeveld 2005: 6), that is, the speaker makes it serve his communicative strategy “in view of the possible perlocutionary effects of his speech act” (Hengeveld 2005: 14). While predicate and predication operators operate at the representational level (Hengeveld 2005: 7), proposition and illocution operators operate at the interpersonal one (Hengeveld 2005: 13). In his table on the classification of the operators, Hengeveld seems to use a superordinate term he imposes on evidentiality and epistemic modality. Although never referring explicitly to the category problem, as the goal of his pa-

120 per is the refinement of the treatment of modality from a Functional Grammar perspective, he somehow ‘solves the problem’ as a closer look on his table shows (cf. Hengeveld 2005: 6; my emphasis): Table 7: Hengeveld’s classification of the operators (2005: 6) OPERATORS (classification) PREDICATE OPERATORS

PREDICATION OPERATORS

PROPOSITION OPERATORS ILLOCUTION OPERATORS

SEMANTIC DOMAIN - internal temporal constituency - presence or absence of property or relation expressed by predicate - time of occurrence - frequency of occurrence - actuality of occurrence - source of proposition - commitment to proposition - weakening strategy - strengthening strategy

GRAMMATICAL CATEGORY - imperfect/perfective, phasal aspect - predicate negation - tense - quantificational aspect - objective mood/polarity - evidential mood - subjective mood - mitigating mode - reinforcing mode

Hengeveld uses the term epistemological modality to describe proposition operators, that is, to describe “those linguistic means through which the speaker expresses his commitment with regard to the truth of a proposition” (Hengelveld 2005: 14). He then distinguishes two subtypes of epistemological modality – and here we are: subjective modality, through which the speaker specifies the kind and degree of his commitment; and evidentials […], through which the speaker specifies how the proposition came to his knowledge. What both subtypes have in common is the relevance of the source of the information contained in a proposition (Hengeveld 2005: 14).

That means that Boye (2010a, 2006) and Hengeveld (1989, 2005) argue in the same direction: both impose a superordinate category on evidentiality and epistemic modality, or subjective modality, which in turn represent subcategories to Boye’s notion of “justificatory support” (2010a: 16) or Hengeveld’s “epistemological modality” (2005: 14). In my view, both terms are a little misleading in so far as the former notion sounds ‘too evidential’ and the latter ‘too epistemic’. Furthermore, as a superordinate category that not only comprises epistemic modality and evidentiality but also other categories such as subjectivity, for instance, should be found, I would refuse – for the purpose of the present study – ‘too evidential’ and ‘too epistemic’ notions.75 That is why Faller’s notion pro75

Bednarek (2006) deals below the heading of “epistemological positioning” with questions such as ‘Who is the source of information?’, ‘What is the basis of someone’s (the

121 posed in her study on Semantics and Pragmatics of Evidentials in Cuzco Quechua (2002) must also be rejected. Faller distinguishes between evidentials and modals, but proposes to “call an element that expresses evidentiality and epistemic modality simultaneously […] epistential” (Faller 2002: 87). Furthermore, she explains: Rather than having the concepts narrow epistemic modality and narrow evidentiality be included in either the wide concept of evidentiality or the wide concept of modality, I prefer to maintain the narrow definitions, and possibly find a third term which can serve as a superterm for the two concepts. A good candidate for such a term is perhaps qualification (Faller 2002: 88).

Therefore, qualifications cover evidentials, modals and epistentials. SimonVandenbergen/Aijmer see English as a language having modals and (lexical) evidentials as well as epistentials – “devices which express modality and evidentiality simultaneously” (2007: 31).76 However, the equivalent noun to epistential, that is ‘epistentiality’, cannot serve as an appropriate representation of the superordinate category that is searched for. It does not seem to be neutral enough or all-encompassing because also polyphony, subjectivity and deixis must be accounted for. What should now be ‘concluded from the conclusion’ is that all categories – evidentiality, epistemic modality, subjectivity, deixis and polyphony – represent related categories at least in Spanish and quite likely in other languages as well. That is why these linguistic categories should be subsumed below one encompassing notion, that is, below one superordinate category. It only makes sense to argue whether, for instance, epistemic modality and evidentiality are overlapping notions or not if and only if they are subsumed below a superordinate cate-

76

writer’s or a third party’s) knowledge?’, ‘How certain is their knowledge?’ etc., and then explains that the notion of epistemological positioning “partly overlaps with the notion of evidentiality” (Bednarek 2006: 635; my emphasis). I would not say that both notions overlap. Evidentiality seems to be subordinated to ‘epistemological positioning’ instead. However, for the purpose of the present study this notion also sounds ‘too epistemic’. Even in languages where evidentiality is part of the grammatical system epistemic modality and evidentiality may overlap: An example showing that evidential markers assume the pragmatic function of marking the uncertainty of the speaker even in languages with pure evidentials is the use of the suffixes –mi/–shi/–chi in the Huallaga dialect of Quechua, which allows the speaker to assume or reject responsibility for the content of what is said. With –mi the speaker assumes responsibility, with –shi he defers it to someone else, with –chi he indicates that no responsibility can be taken for this type of content (Haßler 2010d: 97).

122 gory. I should put emphasis on the fact that the following ‘categorical map’ is developed from the speaker’s perspective only: the categories are understood as bringing to the fore the speaker’s perspective. That is why evidentiality may overlap with epistemic modality, polyphony, deixis and subjectivity; epistemic modality with evidentiality, subjectivity, polyphony and deixis and polyphony may overlap with evidentiality, epistemic modality, subjectivity and deixis: if a speaker introduces at least one other voice for evidential purposes, he establishes another origo and thus another deictic centre. Polyphony may overlap with subjectivity because as long as the speaker only refers or even explicitly quotes another origo, he does not necessarily simultaneously convey his subjective viewpoint regarding the quoted words, but he may do so. And he may simultaneously convey his validative attitude: this is, for instance, the case of the journalistic conditional (cf. chapter 7.1). Being a polyphonic structure it is used to indicate reported information, whereby the speaker simultaneously expresses his validative attitude so that the journalistic conditional is also subjective and epistemicmodal. In summary, all categories may overlap in a particular linguistic expression being part of a particular context. It should again be emphasised that all categories may overlap as it is logically not correct to assume that they overlap at any time in any linguistic expression. The following examples shall illustrate that different categories overlap in different linguistic expressions. Examples (30) and (31) contain verbs of cognitive attitude, pensar and suponer. Think, the English equivalent of pensar, is considered a purely epistemic verb (cf. Capelli 2007: 185). But additionally it is explained: Where evidence is provided by the context, think is construed as the output of a computational process, whereas where the context does not make explicit reference to the reasons for the evaluator’s attitude, think only indicates that what follows is the subjective judgement of the speaker, his/her opinion (Cappelli 2007: 184).

Example (30) contains an instance of pensar to which an inferential reading is applied from contextually provided information: (30) 31.mar.2011 | 22:15 #30 A mi me parece lógico que la gente se vaya a otro pais para tener una vida mejor, eso es lógico y lícito. Pero ¿estaba España preparada para acoger a tantísima gente, que se ha multiplicado con la reagrupación familiar? pues pienso que no, pues por cada uno que trabaja, o trabajaba, se han venido cuatro más. La mayoría de ellos trabaja en la economía sumergida y no contribuye a la Seguridad Social y el gobierno es incapaz de encauzar esto [...] (El Mundo 01/04/2011)

The speaker asks whether Spain was prepared to host so many people (¿estaba España preparada para acoger a tantísima gente […]?). He then infers that it was not prepared to do so by evaluating the current working situation. Hence, epistemic modality, evidentiality, deixis and subjectivity overlap here. The epis-

123 temic/evidential value applied to [p] originates in the speaker’s origo, whereby this evaluation is subjective as verbs of cognitive attitude – in its qualificational non-descriptive construal – generally give voice to the speaker’s subjective, cognitive stance. The following use of suponer can be analysed in similar terms as the same categories overlap here: (31) Eso sí, la independencia sería inmediata, con lo cual tendrían que pasar unos añitos hasta que fueran miembros de la UE, etc., y, supongo, que esas cuentas que se echan de la famosa balanza fiscal variaría porque, claro, ahora cuentan, por ejemplo, los beneficios que tiene, por ejemplo, La Caixa en el resto de España y, si se independizan, esos impuestos se pagarían en España [...] (El Mundo 29/12/2010)

The speaker clearly provides more information on why he supposes that [p]. Cappelli also differentiates between uses of English suppose where contextual information is available which proof that the reading of suppose is inferential and uses where “the inferential process is not overtly provided” so that “the verb is interpreted as signalling the speaker’s subjective and tentative hypothesis” (Cappelli 2007: 217). Example (a) represents the former use, while example (b) represents the latter: (a) I suppose that, Holland being so flat, all Dutch houses needed a piano nobile from which to get a better view […]. (b) Boy looked through the glass of the living room window, or at the glass of the television screen, with the same fierce attention that he had stared through the show windows. If you had asked him, I suppose that he would not have been able to give a very convincing answer as to why he was walking, or looking, or watching […] (Cappelli 2007: 217).

The next two examples contain the modal adverb probablemente, whereby the adverb in example (32) is represented in its ‘canonical’ use but in example (33) in its ‘non-canonical’ use. That means, the former is epistemically used to (simply) express the probability that [p], while the latter, in contrast, conveys an inferential reading: the speaker infers that 3% represent only the tip of the iceberg because up to 30% of the Europeans are not able to control their expenses. Hence, in the former use the categories epistemic modality, deixis and subjectivity overlap but in the latter, epistemic modality, subjectivity, deixis and evidentiality (inferentiality) because the semantic aspects encoded by the adverb remain epistemic, whereas the inferential meaning aspect is contributed by contextually provided information: (32) El ‘trading de alta frecuencia’ no puede sobrevivir sin este tipo de sistemas, por lo que su demanda es cada día mayor. Así, el hedge fund Renaissance Technologies, probablemente el más rentable del mundo, tiene 300 empleados, y ni un solo economista. (El Mundo 31/12/2010)

124 (33) Según los datos que se manejan, el porcentaje de adictos a las compras que llegan a esta situación patológica y de pérdida de control es del 3% aproximadamente, “aunque probablemente eso sólo es la punta del iceberg porque hasta el 30% de los europeos tiene problemas de descontrol en el gasto”. Igual entre hombres y mujeres A pesar de la idea de que son las mujeres las mayores adictas a las compras [...] (El Mundo 31/12/2010)

In the use of supuestamente the categories evidentiality, deixis, subjectivity and polyphony overlap. Supuestamente is used to indicate reported information, whereby at least one other voice is introduced so that it may be termed polyphonic. At least one other deictic centre is established or at least referred to where the reported information originated in. The current speaker/journalist does not indicate his own validative attitude towards [p]. That is why the use of the adverb is in no way modal. The evidential value that is applied to [p] with the use of supuestamente originates in the speaker’s origo so that its use is also subjective. Furthermore, at least one other subjective point of view is introduced by the reference to another voice: (34) En una de esas conversaciones, mantenida el 24 de febrero de 2009, Crespo que fue secretario de Organización del PP gallego- ordenaba supuestamente a López Rubal la “ejecución” de los fondos que, según los informes policiales, la red habría ocultado en Suiza. (El Mundo 30/12/2010)

The last two examples contain instances of the journalistic conditional. Additionally, the source of information is contextually provided. In the journalistic conditional, all categories overlap. Its use could be explained in similar terms as the one of supuestamente above, whereby in example (35) it is referred to more than one other voice (varios expertos ‘various experts’) and in example (36) to a report, but the journalist’s validative attitude is conveyed as well because the same information could have been transmitted by using the indicative (cf. chapter 7): (35) Otras las opciones que baraja el Ejecutivo japonés y la compañía eléctrica Tepco es la construcción de un sarcófago que cubra los reactores dañados de la central. Una técnica que ya se utilizó en Chernóbil y que tampoco ha sido descartada, aunque sería la última alternativa a tomar, según han declarado varios expertos. (El Mundo 31/03/2011) (36) Una de las consecuencias directas de la aplicación de la ‘hipoteca a ala americana’ sería, según un informe de la Agencia Negociadora, sería que los tipos de interés reales se duplicarían, haciendo más difícil el acceso a una vivienda. (El Mundo 31/03/2011)

A further fact I should put emphasis on is that a subjective expression is not automatically evidential, polyphonic or epistemic and that a deictic expression is not automatically evidential, polyphonic or epistemic.

125 The categories of evidentiality, epistemic modality, subjectivity, deixis and polyphony are encompassed by the notion of speaker’s perspectivisation because they all bring to the fore the speaker’s perspective of the narrated event – even in the case of polyphony, where the speaker shows that he is able to include further perspectives. Speaker’s perspectivisation is the superordinate category, while the strategy of perspectivising a certain [p] represents the process. The outcome of this process consequently is the verbalised speaker’s perspective.

126

Fig. 7: The model of speaker’s perspectivisation

The notion of perspectivisation with regard to evidentiality and related topics is also used by Mayer (1990), Mushin (2001: 190), Tátrai/Csontos (2009), by Diewald (1999: 211) in German, by Wachtmeister Bermúdez (2005: 28) in Spanish and in a slightly different way by Haßler (2010a: 225), whereby Traugott/Dasher (2002: 21-22) also stress the point that, for instance, deixis and modality are subjective because they explicitly encode the speaker’s/writer’s point of view, which may be seen as a synonym for the speaker’s perspective. Rooryck (2001a) also explains that the notion of ‘source of information’ is closely linked, if not identical, to that of ‘speaker perspective’ and ‘point of view’, which are often argued to play a role in other areas of the grammar, from long distance binding and logophoricity […] (Rooryck 2001a: 127).

Haßler (2011a: 3) also uses the French synonym perspective du locuteur in order to explain the relationship between modality and polyphony: […] dans la modalisation, une relativisation de la proposition se fait à l’égard de la conscience et de la perspective du locuteur actuel, dans la polyphonie, la relativisation se produit pour indiquer une perspective étrangère (Haßler 2011a: 12).

127 Nevertheless, the notion of perspectivisation has not been used so far to refer to more categories than evidentiality and epistemic modality, even though Haßler (2011a) already includes the notion of polyphony. Furthermore, as far as I know, the notion has not been used so far to serve as a superordinate category.77 Hence, evidentiality is to be defined as one category of speaker’s perspectivisation, whereby evidentiality may overlap with other linguistic categories that are also subordinate to the category of speaker’s perspectivisation and it does overlap with deixis and subjectivity. In languages that do not possess a grammatical system of evidentiality, evidentiality is expressed by evidential expressions which may be lexical or grammatical in nature. Whether an expression is evidential or evidentially used is determined by meaning aspects that are encoded by a particular expression and possibly additionally by meaning aspects that are contributed by contextually provided information. An expression is evidential or evidentially used if it expresses the source of information, i.e. evidence, for the transmitted information. So evidentiality should be defined as one category of speaker’s perspectivisation and as the category of speaker’s perspectivisation that is defined in terms of the notion of evidence78.

77

78

If the notion of perspectivisation with respect to evidentiality and related categories has been used by more linguists than I came across, I apologise for not having mentioned them above. I subscribe to Boye’s (2010b) definition of evidentiality in terms of the notion of evidence, despite the fact that I would add ‘quotative’ as well: That is, it is defined as covering meanings – and expressions with meanings – that can be described in terms of familiar distinctions between ‘direct’ and ‘indirect evidence’, subsequent distinctions between ‘visual’ and ‘non-visual direct evidence’ and between ‘reportive’ and ‘inferential evidence’, and so on […] (Boye 2010b: 291-292).

128

3 What has been done so far concerning evidentiality in Spanish, other Romance languages and recently in general79 “In about one-quarter of the world’s languages, every statement must specify the type of source on which it is based, choosing from a system of evidentials” (Dixon 2010: 13; my emphasis). And Haßler (2002, 2004, 2010a, 2010d)80, Große (2000, 2011), Volkmann (2005), Cornillie (2007a, 2007b, 2010a, 2010b), Wachtmeister Bermúdez (2004, 2005), Squartini (2001, 2004, 2007, 2010), Dendale/Van Bogaert (2007) and Diewald/Smirnova (2010a, 2010b) show that in other languages a speaker uttering a statement can specify the type of source on which it is based, choosing from a system of evidential expressions because “[i]rrespective of the existence of a grammaticalized system of evidentials, every language has a variety of non-grammaticalized expressions and constructions carrying evidential meanings” (Diewald/Smirnova 2010a: 41). All the linguists (among others) mentioned above understand evidentiality as a functional category. Consequently, “its investigation should not be restricted to inflectional morphology but consider all types of formal realization of the functional category” (Diewald/Smirnova 2010a: 51). Studies that investigate other types of formal realisation of evidentiality but inflectional morphology will be compiled here. So this is a review chapter, where an overview about studies treating evidentiality in Spanish as well as in other Romance languages will be given, even though not all studies can be discussed here at length. Recently published papers and studies that are of theoretical and/or methodological importance for the present study will also be reviewed, although neither Spanish in particular nor with Romance languages in general is dealt with.

79

80

The whole state of research of evidentiality from its beginnings up to ‘more or less now’ is perfectly summarised in Volkmann’s Weltsicht und Sprache. Epistemische Relativierung am Beispiel des Spanischen (2005: 75-85), but see also Plungian’s history of linguistic studies of evidentiality (2010: 23-28). I, in contrast, will concentrate on a detailed review of papers and books since 2001. Those published after March 2011 cannot be considered here. So in the following the most important studies of the last decade will be reviewed. The papers by Haßler will not be reviewed here as they are relied on elsewhere, that is, in preceding chapters. The same holds for the papers by de Haan.

129

3.1 Evidentiality and the grammaticalisation of German evidential expressions The first study considered briefly is the extensive monograph about evidentiality in German by Diewald/Smirnova (2010a). Their corpus-based investigation of evidentials in German is synchronic as well as diachronic in nature, that is, they have a look at four evidential constructions and their development from Old High German up to present-day German. The evidential constructions they focus on are (cf. Diewald/Smirnova 2010a: 1): a) werden (‘become’)

+ infinitive

b) scheinen (‘seem’)

+ zu-infinitive

c) versprechen (‘promise’)

+ zu-inifinitve

d) drohen (‘threaten’)

+ zu-infinitive,

as the corresponding examples illustrate (cf. Diewald/Smirnova 2010a: 9): a) Wenn also Franz Beckenbauer und andere der Stiftung Warentest vorhalten, sie wolle sich über die Weltmeisterschaft profilieren, werden sie ziemlich schnell feststellen, dass… ‘So, if Franz Beckenbauer and others reproach the Stiftung Warentest for trying to promote themselves via the World Cup, they are bound to realize very quickly that…’ b) …und die die Freude am Risiko scheint sein zwanzigjähriger Sohn Hans-Joachim geerbt zu haben. ‘…and his twenty-year-old son Hans-Joachim seems to have inherited his readiness to take risks.’ c) …und nach längerem Suchen und nachdem er ihn mit den Zähnen getestet hatte, fand Kürenberg einen Reis, der körnig zu kochen versprach. ‘…and after a lengthy search and after testing it with his teeth, Kürenberg found a type of rice which promised to boil grainy.’ d) Sitz gerade auf dem Fahrrad, Ilja! Führe den Lenker nach links, wenn Du nach rechts zu kippen drohst! ‘Sit straight on the bike, Ilja! Turn the handlebars to the left, when you find yourself falling off to the right.’

According to Diewald/Smirnova (2010a: 1) these four verbal periphrastic constructions + infinitive have been about to develop into grammatical markers of evidentiality, which shows that in German a new grammatical category is emerging and explains why they call these constructions “evidentials”.

130 Their empirical study of synchronic and diachronic language data, based on different text types and/or registers, unifies the approaches of construction grammar, functional theories on language change, typological research and grammaticalisation theory, whereby their leading hypothesis is that the development of evidential verbal constructions in German constitutes a grammaticalization process in the course of which evidential distinctions are integrated as grammatical distinctions into the verbal system of German (Diewald/Smirnova 2010a: 1).

They argue in support of the acceptance of the simple fact that grammatical systems (always) develop: because of the fact that they are always found to be in progress, they may “give rise to systems which are not yet fully grammaticalized”; grammatical systems “display varying degrees and stages of a grammaticalization process” (Diewald/Smirnova 2010a: 50). This, in turn, leads them to deal with the question “to what degree an evidential construction is grammaticalized” (Diewald/Smirnova 2010a: 3). The second chapter of their study deals with definitions and delimitations of evidentiality. Here, they also address “[t]he dispute on the status of evidential expressions in European languages” (Diewald/Smirnova 2010a: 50), taking up Aikhenvald’s (2004) arguments against the existence of evidentials in European languages. The authors convincingly argue that the four German verbal periphrastic constructions + infinitive are evidential – even though not fully grammaticalised – because they fulfil the basic semantic requirement to be classified as evidentials, namely ‘source marking’ (cf. Diewald/Smirnova 2010a: 52). While the third chapter relates to the relationship between epistemic modality and evidentiality, where the ‘deictic nature’ of evidentiality is also dealt with – whereby the assumption that evidential markers are deictic signs is also formulated earlier (cf. Diewald/Smirnova 2010a: 10-11, 53-54) and again later (Diewald/Smirnova 2010a: 90; cf. also chapter 2.2 of the present study) – the two following chapters are devoted to the process of grammaticalisation in general and Lehmann’s grammaticalisation parameters in particular (see, for instance, Lehmann 2002). Subsequently, chapters six to nine, which represent the empirical main part, analyse the synchronic status as well as the diachronic development of the four constructions in question. Their leading hypothesis “that the German language has acquired a new grammatical category […], which is integrated into the growing system of verbal periphrastic paradigms in German” (Diewald 2010a: 324) could be verified. The four verbal periphrastic constructions + infinitive that have been in question can indeed be termed evidential markers in present-day German, as the indication of evidence makes up their primary meaning, whereby they belong to the

131 indirect evidential domain (Diewald/Smirnova 2010a: 325, 4). Furthermore, the four constructions are grammaticalised to a relatively high degree, albeit displaying different degrees of grammaticalisation when compared to each other: versprechen + zu-infinitive is the most retarded verbal periphrastic construction, whereas the construction werden + infinitive represents the most progressed form (Diewald/Smirnova 2010a: 325). Evidentiality in German. Linguistic Realization and Regularities in Grammaticalization represents an excellent study, where theoretical profundity, a well developed methodological approach and an empirical investigation that takes synchronic as well as diachronic data into account are combined.

3.2 Evidentiality as a kind of epistemic relativisation – exemplified by Spanish Volkmann’s extensive monograph Weltsicht und Sprache. Epistemische Relativierung am Beispiel des Spanischen (2005) treats epistemic modality, evidentiality and quotation as different types of ‘epistemic relativisation’ (for this notion see also Haßler 2001: 170). By ‘epistemic relativisation’ the linguistically encoded reference to an epistemic centre is meant, i.e., the description of a state of affairs from the viewpoint of a certain epistemic centre. The epistemic relativisation onto an external source is distinguished from the relativisation onto the internal – the speaker’s – source (cf. Volkmann 2005: 453). This theoretical-notional background and approach to epistemic modality and evidentiality in Spanish of Volkmann’s study imply two facts: first, epistemic modality, evidentiality and quotation are given a ‘superordinate notion’, namely ‘epistemic relativisation’. Volkmann terms epistemic relativisation a semantic domain on its own (cf. Volkmann 2005: 413). Secondly, as Volkmann opposes relativisation onto an external source to relativisation onto the internal source, she adopts Frawley’s (1992: 413) deictic categorisation and scaling of epistemic modality – although it should be better termed as the deictic categorisation of evidentiality: ‘from the self’ is contrasted with ‘from the other’. Volkmann herself explains that she adopts Plungian’s (2001) differentiation between personal evidence and mediated evidence, whereby relativisation onto the internal source would apply to the former, relativisation onto an external source to the latter (cf. Volkmann 2005: 83). Against the background of epistemic relativisation Volkmann investigates linguistic expressions in Spanish that a speaker, in describing a certain state of affairs, uses to relativise onto him or onto another source. Thus, linguistic means such as verbs of cognitive attitude or modal verbs are within the scope of her research but also other (graphic) indicators that signal the differentiation be-

132 tween distinct epistemic centres like quotation marks or change in typeface. Furthermore, the analysis of the syntactic variety of the relativised proposition with respect to the relativising expression, as well as the analysis of paralinguistic devices such as facial expressions and gestures is included in her study. The empirical data are retrieved from fictional and non-fictional written texts of Contemporary Spanish (Volkmann 2005: 9). As Volkmann’s findings on the epistemic future and the modal verbs will be referred to in later chapters (7.2 and 6, respectively), I will only briefly summarise her remarks on indicators which signal the differentiation between distinct origos because it is in that chapter where she introduces her notion of doppelte Deixis ‘doubled deixis’: the term ‘origo-differentiating indicators’ (cf. Volkmann 2005: 231) is applied to linguistic elements which indicate a change of perspective, that is, a change from one origo to another epistemic centre. Prominent linguistic devices that fulfil this function are deictic expressions such as pronouns or temporal adverbs. Bühler’s notion of ‘deixis at phantasma’, by which a second origo – a fictional or remembered epistemic centre – is established, represents the theoretical as well as notional background for Volkmann to develop her notion of ‘doubled deixis’ (cf. Volkmann 2005: 233).81 If in one and the same utterance or in a text different deictic expressions that are related to different epistemic centres are found, a change of perspective is indicated. The change of perspective may be further emphasised by graphic indicators as example (a) illustrates: (a) Ana me dijo: “El próximo jueves tengo examen práctico de conducir” (Volkmann 2005: 233).

While in example (a) the change of perspective is obvious, example (b) illustrates an overlap of perspectives because the graphic indicators are missing but the different deictic expressions are to be interpreted as an indicator of different perspectives, of different epistemic centres: (b) Mañana tenía Ana el examen práctico de conducir (Volkmann 2005: 234).

In example (b) the temporal adverb mañana – an expression for post-temporaneity – and the imperfecto form tenía – to indicate pre-temporaneity – do, at first sight, not match. But while mañana refers to the speaking origo, the current speaker, tenía refers to a quoted speaker. So tenía represents the deixis-at-phantasma part of this utterance as it refers to another statement. It could be trans81

The term doppelte Deixis is guided by Reyes’ (1990) term doble deixis. She, in contrast to Volkmann, uses the term only to refer to overlaps of perspective, while Volkmann also integrates the change of perspective (cf. Volkmann 2005: 234; cf. also Volkmann 2009).

133 lated as phrases like alguien me dijo ‘someone told me’ (cf. Volkmann 2005: 234; cf. also the example provided by Reyes 1996: 12). In examples (a) and (b) are thus established to different deictic centres, i.e. the deixis ‘is doubled’. The same examples could also have been analysed in terms of polyphony, while example (a) would belong to the most explicit form and example (b) to the most implicit form of polyphony (cf. Gévaudan’s examples in chapter 2.1.4). An analysis in terms of polyphony would have also led to the conclusion that in both examples two distinct voices are found so that different epistemic centres are established. Volkmann’s study of (linguistic) means that signal a relativisation onto the internal or onto an external source should be seen as a highly important contribution to the research field of evidentiality, epistemic modality and quotation not only for Spanish linguistics but also for other languages, as it provides a methodological basis for the analysis of relativising expressions in other languages as well. Epistemic relativisation is seen as a universal phenomenon (Volkmann 2005: 422) since every language provides linguistic means by which a speaker can relativise a certain proposition.

3.3 Evidentiality and the processes of subjectification and modalisation The Belgian Journal of Linguistics that comprises papers on Topics in Subjectification and Modalization, edited by Cornillie and Delbecque (2006), deals with the expression of the speaker’s stance to the proposition (or part of it). [The papers] shed new light on two kinds of shifts: (i) the shift from an originally lexical reading toward propositional or procedural ones and (ii) the shift from direct speech to various forms of indirect speech (Cornillie/Delbecque 2006: vii).

In the following I will briefly outline the papers by Vandelanotte (2006) and Byloo/Kastein/Nuyts (2006), and consequently have sometimes a sideglance at the studies they are influenced by. While Byloo/Kastein/Nuyts’ paper is to be found in the first part of the journal, Vandelanotte’s study belongs to the second part. In the first part “the patterns lying behind the grammticalization and subjectification of adverbs and verbs” are discussed (Cornillie/Delbecque 2006: vii). The papers of the other part, in contrast, “are concerned with how tense and mood interact with degrees of grammaticalization and subjectification in speech and thought representation” (Cornillie/Delbecque 2006: ix). In “Speech or Thought Representation and Subjectification, or on the need to think twice” Vandelanotte (2006) studies different structural types of speech or thought representation and demonstrates that indirect speech or thought (IST)

134 and distancing indirect speech or thought (DIST) “lend themselves to subjectification” (Vandelanotte 2006: 163). In IST, he claims, “the reported clause […] does not re-enact the original speech event, but gives rather the current speaker’s perspective on the original sayer or cognizant’s words or thoughts” (Vandelanotte 2006: 139). After offering the example With a slight Russian accent, he asked me who I was and what I wanted – in contrast to directly addressing the question where the asked person would be referred to as you – he explains: “it is, in indirect speech or thought, the speaker’s deictic centre from which both the reporting and the reported clauses are deictically and expressively construed” (Vandelanotte 2006: 139). DIST, in contrast to IST, is said to be even more current-speaker centred as “both the reported and the reporting clause are construed entirely from the perspective of the current speaker” – not even a “separate sayer’s consciousness is truly conjured up” (Vandelanotte 2006: 141). In examples such as Who am I and what do I want, he asked me, the original sayer’s discourse is “echoed in what is deictically and expressively fully a speaker’s utterance” (Vandelanotte 2006: 141). However, it should not be claimed that another speaker’s consciousness is not conjured up, whereby “truly” may be the code word here: it is not truly conjured up; nevertheless it is conjured up, namely by the personal pronoun “he” in “he asked me”. Within the framework of his study, he takes up the distinction that is often made “between sentences in which the predicate (such as think) appears in the so-called main clause, and those in which it appears in a ‘syntactically parenthetical’ (Ifantidou 2001: 120) clause” (Vandelanotte 2006: 137). Vandelanotte demonstrates that the two different types of speech or thought representation differ regarding their syntactic possibilities “and hence instantiate two different types of structure” (Vandelanotte 2006: 138). At this, he is highly influenced or rather inspired by Thompson’s study on ‘object complements’ (2002). In her paper, based on a corpus of conversational English, Thompson argues “that the standard view of complements as subordinate clauses in a grammatical relation with a complement-taking predicate (CTP) is not supported by the data” (Thompson 2002: 125). For predicates such as forget, guess, (be) possible or think (Thompson 2002: 138) that take complements she rejects their analysis “under the heading of complementation” (Thompson 2002: 125), as what is usually analysed as ‘complements’ firstly are not objects of CTPs and secondly are not subordinate to CTPs (Thompson 2002: 136). In her view, these CTPs and their subjects should be analysed “in terms of epistemic/evidential/evaluative formulaic fragments expressing speaker stance toward the content of a clause” (Thompson 2002: 125), which she summarises under the heading of ‘stance expression’, i.e. the expression of epistemic stance, the providing of evidence or the expression of an evaluation (Thompson 2002: 140). She shows that the vast majority of the CTPs,

135 which may precede, interrupt or follow their associated utterances (Thompson 2002: 142-143), “are epistemic/evidential/evaluative formulas, performing stance work, that is, providing epistemic, evidential, or evaluative framing for the utterances they go with” (Thompson 2002: 141). She summarises: […] CTP-phrases are stored and retrieved as schematic epistemic/evidential/evaluative (e/e/e) fragments. An e/e/e fragment provides an epistemic, evidential, or evaluative perspective on the finite indicative utterance it goes with; which part of those utterances it is construed as providing the e/e/e frame for is a matter of inference and negotiation (Thompson 2002: 146). […] the grammar of complementation, with a main clause containing a subject and an utterance-cognition verb together with a clausal object (or subject), may be better understood as a combination of an epistemic/evidential phrase together with a declarative or interrogative clause. I have argued that the facts of everyday language use support this view, suggesting that actual conversational English provides no evidence that speakers operate with a category of “complement”, that the “complements” are best described as being in a grammatical relation with the CTP-phrase, or that they are in any sense “subordinate” to the CTP-phrase. Instead, what the data show is the very general and frequent use of a schema consisting of an e/e/e phrasal fragment and a clause, where the action the speaker is accomplishing is being done by that clause (Thompson 2002: 155).82

Mithun (2009) argues in the same direction: the news, the main information, the speaker wants to transmit is not the act of believing or hearing but the information expressed by the complement. So the analysis of such utterances under the heading of complementation indeed seems to be a misleading term, as it does not ‘only’ complement the matrix clause (cf. Mithun 2009: 63). Main clauses like pienso que or creo que are better understood as epistemic or evidential phrases as will also be shown in chapter 5.2. Byloo/Kastein/Nuyts (2006) compare the meanings and uses of the English adverb certainly and the Dutch one zeker. Harry Potter and the Chamber of Secrets (Rowling 1998) and its Dutch translation (Rowling 1999) represent the corpus which the occurrences of the adverbs are retrieved from (2006: 47). They list examples like the following one 82

In their paper “Complement-taking predicates: Usage and linguistic structure” Boye/Harder (2007) criticise Thompson’s paper (2002) from a usage-based syntax-point of view. They show “how a picture that is fully committed to maintaining the role of structural (including structural-semantic) subordination can simultaneously remain fully faithful to principles of usage-based linguistics” (Boye/Harder 2007: 569). However, Thompson’s line of argumentation is reasonable as she addresses a problem primarily terminological in nature: usually, the matrix clause is seen as central and not the complement clause. The structure consisting of CTP + subordinate clause is indeed not appropriately analysed in terms of ‘main clause’ + ‘complement clause’.

136 [Dobby begon] zichzelf met Harry’s bureaulamp op het hoofd te slaan. Beneden viel een stilte. Twee tellen later hoorde Harry [...] oom Herman de gang opstormen. ‘Dirk heeft zeker de televisie laten aanstaan, de kleine rakker!’ riep hij. (PD: 17) ‘[Dobby] seized Harry’s desk lamp and started beating himself around the head [...]. A sudden silence fell downstairs. Two seconds later Harry [...] heard Uncle Vernon coming into the hall, calling, ‘Dudley must have left his television on again, the little tyke!’’ (Byloo/Kastein/Nuyts 2006: 52)

“where the evidence leading to the epistemically modified conclusion is mentioned in the text; hence the inferential meaning is strongly invited” (Byloo/Kastein/Nuyts 2006: 53). The inferential meaning is underlined by the fact that the original text passage in English contains must. The Spanish equivalent of zeker, seguramente, will also be shown to convey an inferential reading in certain contexts (cf. chapter 5.1.6).

3.4 Evidential adverbs and their grammatical/lexical status In “Evidentiality: Linguistic categories and grammaticalization”, Boye/Harder (2009) also deal with sentence adverbs, more precisely, with their grammatical/lexical status because they are mainly concerned with the question of “what a linguistic category is, and what it means for linguistic elements to have grammatical rather than lexical status” (Boye/Harder 2009: 9). The two questions are dealt with regard to evidentiality. The ‘peculiarity’ of their paper is that sentence adverbs such as apparently are considered grammatical elements – and not lexical ones – because they are said to involve secondary predication (when a meaning is coded as being secondary information). Main predication, in contrast, is represented by lexical elements. Secondary predication is thus associated with grammatical elements that have to be seen in relation to another element as the grammatical status of elements can (only) be considered as having arisen when compared to another element: […] Secondariness is the fundamental property associated with grammatical status: grammaticalization occurs in all and only those cases where an element becomes coded as secondary in relation to another, thereby creating both a new, less prominent element, and a dependency relation with the associated primary element. (Boye/Harder 2009: 38).

Boye/Harder’s approach – although arguing in another paper (2007) against Thompson’s line of argumentation – could also be applied to Thompson’s analysis (2002) or Mithun’s (2009). In a sentence having the structure CTP (or e/e/e fragment) + complement clause it is, for instance, not primarily the act of

137 believing or thinking that represents the main information the speaker wants to transmit but the information that is conveyed by the subordinate clause. So the CTP-phrase could be described as secondary in relation to the subordinate clause. Nevertheless, in the present study sentence adverbs will later on be named ‘lexical evidential expressions’. Later studies should be concerned with the grammatical status of CTP-phrases.

3.5 Evidentiality between lexicon and grammar in various languages The articles in the Italian Journal of Linguistics (2007), edited by Squartini, study lexical and grammatical means to express evidentiality in various (mostly European) languages, e.g. Cornillie (2007b) in Spanish, Olbertz (2007) in Mexican Spanish, Pietrandrea (2007) in Italian, Giacalone Ramat/Topadze (2007) in Italian and Georgian and Dendale/Van Bogaert (2007) in French. In the introductory part of the journal Squartini explains the purpose of the Italian Journal of Linguistics: Between the two opposite endpoints represented by languages with a fully-fledged grammatical paradigm of evidential forms, and languages which totally attribute these functions to the lexicon, we should admit languages showing a more mixed nature. These languages demonstrate that evidentiality shows up as a semantic development of grammatical forms (evidentiality strategies in Aikhenvald’s terms), but is also expressed by semi-grammatical markers as well as by the interplay of lexical and grammatical phenomena (Squartini 2007: 5-6).

That is why the journal contains studies dealing, among others, with Romance languages which, in the following, will be reviewed in more detail. In “A semantic description of French lexical evidential markers and the classification of evidentials” Dendale/Van Bogaert’s (2007) aim is to ‘reanalyse’ a few French lexical markers (trouver, avoir l’impression, penser, croire, paraître and sembler) and two grammatical markers (devoir and the future form) in terms of evidentiality because already before the “evidentiality era” a selection of these markers was “described in modal terms or in non-technical evidential-like terms, comprising such notions as conclusion, reasoning, calculus, deduction, supposition, assumption, etc. […]” (Dendale/Van Bogaert 2007: 66). Their reanalysis is said to show that the semantic differences between the markers can be grasped in terms of evidentiality, whereby the common classification of evidentials is not considered “fine-grained enough to cover all the fine distinctions between semantically similar lexical expressions” (Dendale/Van Bogaert 2007: 65).

138 In this review – and against the contents of the present study – I will leave out their analysis of paraître and sembler and will focus on their study of the verbs of cognitive attitude, devoir and the future instead. The first part of their paper represents a study of the evidential uses of verbs of cognitive attitude, i.e. je trouve que ‘I think/judge that, lit. I find (out) that’, j’ai l’impression que ‘I have the impression that’, je pense que ‘I think that’ and je crois que ‘I believe/think that’. The authors remark that these verbs have rarely been analysed against the background of evidentiality (cf. Dendale/Van Bogaert 2007: 67), which is also true for their Spanish equivalents. To my mind, the first part of their “semantic in-depth analysis” (Dendale/Van Bogaert 2007: 66) is a little bit problematic as they work with introspectively gained language data or data that were introspectively gained by Ducrot (1975, 1980). To explain, for instance, the difference between je pense que/je crois que and j’ai l’impression que they picture the following situation: [The] situation, adapted from Ducrot, illustrates the semantic differences between je pense que/je crois que on the one hand and j’ai l’impression que on the other hand. Suppose a friend has told you, with the authority of an expert, that the MP3-players of brand X are of a rather bad quality and that you had better buy an MP3-player of another brand. Without any additional knowledge, you can pass on the advice to another friend by saying: Je crois que ces appareils ne sont pas solides. ‘I think that those players are not reliable.’ In the same situation, the use of j’ai l’impression que would be inappropriate […] (Dendale/Van Bogaert 2007: 69; emphasis in bold is mine).

Does this mean that French je crois que is commonly used to indicate reported information? If speaker A tells speaker B what speaker C had told him, why should speaker A tell speaker B C’s information by using je crois que? In the described situation, phrases that indicate reported information such as ‘C has told me that/I was told by C that’ or an even more indirect ‘I heard that’ are much more likely to be used than the phrase proposed by Dendale/Van Bogaert. Nevertheless it is true that j’ai l’impression would still less fit into the example sentence and it is certainly true that “je crois que can be used as a marker of non-personal experience by the speaker; j’ai l’impression que, by contrast, is necessarily a marker of personal experience by the speaker” (Dendale/Van Bogaert 2007: 70).

139 Concerning the difference between croire and penser, the authors mention that this difference is not sufficiently discussed by Ducrot (1980) and then quote Martin’s (1988)83 view of the difference between these two verbs: […] penser is a verb of judgment, always implying a construction of the mind, a conclusion formed on the basis of evidence, incompatible with direct perception. Croire, on the other hand, is a verb of (uncertain) knowledge for Martin (1988: 548) signalling that the speaker has some information that can make him take responsibility for p as a plausible proposition (Dendale/Van Bogaert 2007: 71).

Unfortunately, Dendale/Van Bogaert do not work out how penser and croire distinguish from each other with the help of ‘non-introspectively’ gained data. Nonetheless, they conclude that “je pense que would be a better example of evidential inference marker, than je crois que, because it expresses a construction of the mind” (Dendale/Van Bogaert 2007: 75), but in certain contexts both expressions can be considered inferential markers (Dendale/Van Bogaert 2007: 85). The difference between je crois que/je pense que and je trouve que, in contrast, is – at least partly, namely for trouver – thoroughly worked out with the help of empirical data, among others from the corpus programme Frantext. The former pair is said to indicate indirect evidence and the latter to indicate direct evidence, even though it is “unmarked as for the specific type of direct evidence” (Dendale/Van Bogaert 2007: 73). J’ai l’impression is, in line with je pense que (and je crois que?), considered a marker of indirect evidence (Dendale/Van Bogaert 2007: 76). Their study of devoir and the future form has revealed that the former can be used to express an inference (cf. Dendale/Van Bogaert 2007: 83) because it can be replaced by adverbials such as certainement (cf. also the translation of English must by Dutch zeker in Byloo/Kastein/Nuyts’ paper), while the latter is denied to be a ‘real’ evidential marker (this is in line with Pietrandrea’s (2005, 2007) statements about the future): It is a [sic] more a kind of assertion marker or alethic marker (saying something about the truth value of the proposition), but one that can be used in an evidential strategy (Dendale/Van Bogaert 2007: 83).

83

The studies by Ducrot (1975, 1980) and Martin (1988) that are referred to by Dendale/Van Bogaert are: Ducrot, Oswald (1975): “Je trouve que”. In: Semantikos 1 (1), 63-88; Ducrot, Oswald (1980): “Je trouve que”, in: Oswald Ducrot et al. (eds.), Les mots du discours, Paris: Minuit, 57-92; and Martin, Robert (1988): “Croire que p/penser que p”, in: Jean-Louis Bezenech et al. (eds.), Hommage à Bernard Pottier, Paris: Klincksieck, 547-554. As I myself did not work with these studies, they do not show up in the bibliography of the present study.

140 The inferential or purely speculative reading it generally gives rise to stems from the fact that any information created by the speaker but which he cannot verify at the moment of speaking, must be inferred information (Dendale/Van Bogaert 2007: 83).

This brings up the question whether empirical data from Frantext, for instance, would have confirmed that the future “is not an evidential marker” (Dendale/Van Bogaert 2007: 83). In “The coding of evidentiality: a comparative look at Georgian and Italian” Giacalone Ramat/Topadze (2007) compare Italian and Georgian – two geographically and typologically distant languages – regarding their way of the expression of evidentiality. Although, according to traditional views, Romance languages lack grammatical evidentiality, whereas Georgian is part of the Caucasian language area, where evidentiality is grammaticalised (Giacalone Ramat/Topadze 2007: 11), the authors show that both languages share a few similarities despite the fact that “[n]either language has developed a morphological evidential category” (Gicacalone Ramat/Topadze 2007: 7). For example, [i]n both languages some verb forms, associated with various temporal and aspectual values, have taken on evidential meanings in specific contexts. (Giacalone Ramat/Topadze 2007: 7). […] in both languages the grammatical means of [evidentiality] coding derive from the verbal system either from tense/aspect morphemes, as in the case of the Georgian perfect, or from temporal/modal ones, as in the case of the Italian conditional (Giacalone Ramat/Topadze 2007: 12).

They analyse the use of the Georgian perfect and Italian imperfect, the use of the future as well as the conditional. The authors also include the analysis of the use of modal verbs and further adverbial expressions. While in Georgian the perfect is used to express indirect knowledge about an event (Giacalone Ramat/Topadze 2007: 12-13), in Italian it is the imperfect that may show an evidential extension in certain contexts (Giacalone Ramat/Topadze 2007: 24). So the Georgian perfect is used to show that “the speaker has not witnessed the event but has acquired information about it from indirect sources by inference, hearsay (second-hand, third-hand information), etc” (Giacalone Ramat/Topadze 2007: 13), whereas the Italian imperfect may take on a reportive value as was also shown for the Spanish imperfecto (see, for instance, Reyes 1996 or Volkmann 2005; cf. also Dessì Schmid 2010 for the ‘atypical’ modal meaning of the Italian imperfetto). Concerning the Italian future they somehow criticise that the future has received the label ‘epistemic future’: they “prefer to highlight the inferential basis for the speaker’s confidence and to use the label inferential future, in line with Squartini (2001)” (Giacalone Ramat/Topadze 2007: 25). I would not only ‘pre-

141 fer’ to highlight the fact that the synthetic future is often used inferentially, I would even exclude a pure modal use. If not used to express future time, then it is used to express an inference84. Giacalone Ramat/Topadze conclude: The epistemic future is inferential in nature. It is based on the speaker’s inference, whose source is not specified: it can be either external or internal, based on either hearsay or reasoning. […] The future can express judgments based upon an inferential process, which may be not explicit […] (Giacalone Ramat/Topadze 2007: 26).

That is why they also disagree with Pietrandrea (2005), who says regarding the use of the Italian future: “[…] unlike the modals, the future seems not to condense any inferential process, but appears to function as the form expressing the speaker’s genuine opinion about the propositional content” (Pietrandrea 2005: 93). Disagreeing with Pietrandrea, Giacalone Ramat/Topadze show that the Georgian future – just as the Italian future – may be used to express inference (2007: 18). Comparing their results on Italian with the use of the Spanish imperfect and the synthetic future, it can be concluded that both Romance languages show similar tendencies in use (cf. chapter 7 for the use of the imperfecto). Giacalone Ramat/Topadze also show that Italian and Georgian possess lexical expressions of evidentiality “or intermediate stages between lexicon and grammar” (2007: 12). The expressions are said to range from periphrastic constructions with modal verbs to a variety of adverbial expressions (Giacalone Ramat/Topadze 2007: 12). With respect to the periphrastic constructions with modal verbs, they have a closer look at the Italian modal verbs potere and dovere as well as on the Georgian invariable modal unda, ‘must, need’. The meaning of dovere and potere is said to be based on evidential sources. The present indicative deve and the conditional dovrebbe are shown to “indicate that the speaker draws his/her conclusions on the basis of, respectively, a strong vs. weak type of inference” (Giacalone Ramat/Topadze 2007: 22; cf. also Pietrandrea 2005: 72-76). The Georgian modal particle unda is also shown to be used to express inference (Giacalone Ramat/Topadze 2007: 18). The adverbial expressions they analyse are Georgian turme ‘apparently’, which has a reportive or inferential value (Giacalone Ramat/Topadze 2007: 13). Taking different discourse conventions into account, they illustrate that the indication of the information source in Georgian news-reporting differs from other narrative genres. In news, reported information is usually expressed by lexical expressions (Giacalone Ramat/Topadze 2007: 17), while in traditional narrative 84

However, if it is used to express a (very) weak inference, more precisely, an assumption, the modal component or rather the epistemic meaning are more prominent than the meaning of inferential evidence (cf. also Boye 2006: 66).

142 genres more direct quotations and thus reportive particles are to be found (Giacalone Ramat/Topadze 2007: 18). Georgian has, for instance, the reported speech markers -metki and -tko, “which result from the grammaticalization of the verb tkma ‘say’” (Giacalone Ramat/Topadze 2007. 16). The former “derives from the sequence me vtkvi ‘I said’” and is exclusively used to mark quotations “in the first person singular when the speaker reports an utterance s/he had already made or reflected on in the past”, whilst the latter “is used when the addressee is a mediator between the speaker and a third person” (Giacalone Ramat/Topadze 2007: 16). The particle -o, in contrast, is used to mark quotations in the second or third person, and “[w]hen the information source is not specified, -o means ‘it is said’, ‘they said’” (Giacalone Ramat/Topadze 2007: 16). The overall conclusion they consequently draw is that, comparing Italian and Georgian, inferential and reportive evidentiality are expressed differently. While the two evidential subcategories “cluster together in the Georgian perfect”, they are distinguished in Italian (Giacalone Ramat/Topadze 2007: 7). A further conclusion of this study is that the means to express evidentiality are a domain in movement, which admits several intermediate stages, as is manifested by the presence of grammaticalization processes involving lexical items and the increasing use of adverbial constructions (Giacalone Ramat/Topadze 2007: 7).

Such a linguistic device that is on its way towards grammaticalisation is the Italian third person singular form dice. While the impersonal construction si dice che ‘one says that/it is said that’, which is used to express reported evidence, functions as a main clause that is followed by a complement clause introduced by the complementiser che, the related form on the grammaticalisation path is “morphologically invariable and positionally mobile” (Giacalone Ramat/Topadze 2007: 27). In the following I will briefly have a closer look at the Spanish expressions that are equivalent to the Italian ones and will test Giacalone Ramat/Topadze’s findings. The following four examples (37-40) contain instances of Spanish se dice (que). The first three instances of se dice are used with the complementiser que and fulfil the same function as si dice che analysed by Giacalone Ramat/Topadze. It is used to express reported evidence: (37) En el papiro egipcio, Chester Beatty, se dice que el libro es el medio más seguro para alcanzar la inmortalidad. (El País 25/04/2010) (38) El agujero que se dice que tienen las dos es más grande que el de los bolsillos de Carpanta y tal vez en sus vueltas uno sólo podría encontrar, como en las viñetas de Escobar, unas telas de araña con su insecto dormitando displicente. (El País 28/04/2010)

143 (39) En el texto presentado ayer en Pamplona por varios miembros conocidos de la ilegalizada Batasuna se dice (con otras palabras) que ETA, la violencia terrorista, es ahora un obstáculo para la realización del proyecto independentista. (El País 27/04/2010)

The following instance of se dice is also used to indicate that the transmitted information does not originate (exclusively) in the current speaker: (40) El velo es una manifestación de discriminación y opresión de la mujer, se dice mayoritariamente. (El País 28/04/2010)

The syntactic position of se dice indicates that this form is positionally mobile, as the equivalent Italian form, because it is posposed to [p]. This is clearly in contrast to se dice que, which – from a grammatical point of view – functions as a main clause requiring a complement clause. A similar morphologically invariable and positionally mobile form is Mexican Spanish dizque ‘he/she says that’, which is dealt with by Olbertz (2007). According to her, dizque is not only used to express reported evidentiality but also “subjective distance between the speaker and the communicated content” (Olbertz 2007: 151). Hence, her paper demonstrates that evidentiality – usually exclusively associated with objectivity – and subjectivity are not two categories that have to exclude each other (cf. chapter 2.1.3 and Cappelli 2007): dizque, an adverbial expression of reportive evidentiality, “expresses an objective distance between the speaker and the content she/he communicates”. But at the same time it “may be interpreted as an indication that the speaker doubts the reliability of this information”, thus creating “an implicature of subjective distance” between the speaker and the content that is communicated (Olbertz 2007: 151). So the paper analyses the different functions of dizque that range “from reported speech to the expression of the falseness of a given communicative content from a semantic and a syntactic point of view” (Olbertz 2007: 152). The analysis of naturally occurring data reveals that the semantics of dizque and its syntactic position – and hence its scope – are strictly interrelated: [A]s the size of the linguistic expression modified by dizque decreases, the subjective relation between the communicated content and the speaker (i.e. the parameter ‘speaker attitude’) becomes gradually more prominent, while the objective relation (i.e. the two parameters referring to information source) becomes increasingly irrelevant. (Olbertz 2007: 165-166).

The subjective relation between the communicated content and the speaker is described as “the implicature of a negative speaker attitude”. The implicature becomes so prominent “that it may become part of the meaning of dizque” (Olbertz 2007: 151). This implies that the degree of subjectification of dizque’s semantics “is inversely proportional to its scope” (Olbertz 2007: 151). This fact, in

144 turn, shows that dizque can be considered a counterexample to the view that subjectification is/has to be connected to an increase in scope. If its scope decreases, the reportive function is of little relevance and it is used to serve “the speaker above all to dissociate him/herself from the appropriateness of the property or relation described and, as a consequence, from the truth of the corresponding proposition” (Olbertz 2007: 164). The discussed data are presented along a syntactic and a semantic line. Semantically, Olbertz differentiates between cases in which dizque is used to convey reported meaning and cases in which it is used to convey hearsay meaning. In the first case the source of the proposition can be identified, in the second one the source cannot be identified. “The syntactic line concerns the scope of dizque, i.e. the size of the linguistic element which dizque modifies” (Olbertz 2007: 154155). The different syntactic constructions she considers are: main clauses, finite and non-finite subordinate clauses, non-verbal adverbial phrases and noun phrases (cf. Olbertz 2007: 155-160). This means that Olbertz considers modifications at clause level as well as at constituent level (cf. 2007: 155). For example, for the instances of dizque with finite subordinate clauses she concludes that none of these is entirely neutrally used “with respect to the speaker’s evaluation in terms of truth” (Olbertz 2007: 157). The following text passage is an example containing dizque in a non-finite subordinate clause, where the speaker’s disbelief is made explicit in the quoted argument: Pachita, la muy chambona, por más que le rogamos, se emperró en quedarse en el hotel, dizque porque uno de sus dientes de cera se estaba ablandando con el calor, pero yo creo que fue por miedo que no la dejaran entrar. (González 1999) ‘The very clumsy Pachita, however we urged her, insisted on staying at the hotel, supposedly because one of her wax teeth was getting soft through the heat, but I think it was for fear of not being admitted’ (Olbertz 2007: 157).

For non-verbal adverbial phrases, to pose another example, Olbertz lists a few cases of adverbial modification “in which the proposition can be attributed with a high degree of probability to a specific source” (Olbertz 2007: 158), even though at the constituent level reported speech “is less prone to occur than at the level of the clause” (Olbertz 2007: 158). The most important point she makes is that dizque in non-verbal adverbial phrases almost always transmits a negative truth commitment on the part of the speaker (Olbertz 2007: 158), as the following example illustrates: Andrés estaba rojo dizque del coraje, pero era del brandy. (Mastretta 1990) ‘Andrés had turned red supposedly from anger, but it was from brandy’ (Olbertz 2007: 158).

145 Therefore it is to conclude that dizque at clause and constituent level “almost always carries an implicature of doubt or even outright rejection of the truth of the proposition by the speaker”, besides its reportive function (Olbertz 2007: 160). Hence, at clause and constituent level it conveys an evidential as well as a modal reading. Below constituent level, “dizque no longer modifies the communicative content as a whole, but the ascription of properties or relations within a given propositional content […]” (Olbertz 2007: 160). The items she considers are adjectives/attributive adjectives, verbs and predicative adjuncts and nouns (cf. Olbertz 2007: 160-164). Concerning the modification of the former the author finds out that dizque is primarily used to indicate that the property the adjective expresses “does not contribute to conveying a truthful picture of the reality as conceived by the speaker” (Olbertz 2007: 161). So the meaning that is more prominent is the modal one – not the evidential one. The following example is considered very special: Mira güey, desde secundaria teníamos un grupo dizque de beneficencia y esas jaladas. (Martín 1976) ‘Look mate, from secondary school onwards we had a group supposedly of charity and that kind of bullshit’ (Olbertz 2007: 161).

Because of the fact that the source of the propositional content is a first person speaker, dizque cannot be read as a marker of reported information: it is instead used to mark that the attribute de beneficencia, according to the speaker, does not convey a truthful picture of the reality. If verbs and predicative adjuncts are modified by dizque, epistemic modality is the one and only function fulfilled by the Mexican Spanish adverbial expression (cf. Olbertz 2007: 161). When dizque modifies a noun, it can no longer be seen as an adverb because it “occupies an adjectival slot” (Olbertz 2007: 163), even though it cannot be termed a ‘real’ adjective either, since it is not congruent with the nouns in number. Therefore Olbertz (2007: 163) concludes that, when modifying nouns, it must be a particle. This leads to the questions whether particles are more grammaticalised than adjectives and whether dizque, when used with nouns, should be really termed a particle (cf. Olbertz 2007: 167). In answering these questions, she seems to differentiate between less and higher grammaticalised elements, whereby dizque is said to belong to a less grammaticalised category of particles as “grammatical elements cannot normally be modified by adverbs” (Olbertz 2007: 168). Because of the fact that Olbertz (2007: 168) is able to provide an example in which dizque is modified by an adverb, dizque is considered less grammaticalised. As “particles are understood as comprising all invariable elements which are not prepositions, conjunctions or ad-

146 verbs” (Hartmann 1994: 2953; quoted in Olbertz 2007: 163), the author considers dizque a particle only when used to modify nouns. In summary, both the objective and the subjective relations play a role at the level of the main clause; the subjective relation becomes more prominent when dizque modifies subordinate clauses and constituents, and it becomes the central semantic value in the case of the modification of predicates (Olbertz 2007: 166).

But how could it be explained that the subjective reading increases when dizque’s scope decreases? Olbertz explains that information source marking is less probable, the smaller the linguistic item is. It must be a pragmatically motivated choice when the speaker reports smaller sections of utterances instead of speech reports of at least one utterance, which are more common (Olbertz 2007: 166). If the meaning of a particular linguistic element is conceived as a junction of lexical semantics and pragmatic meanings arising from its use, “one may say that the change from reportative evidentiality to the expression of negative truth commitment is a semantic shift towards a meaning component that was already present, although covertly” (Olbertz 2007: 167). So if a speaker reports a small section of an utterance “there might be something beyond what is being literally expressed” (Olbertz 2007: 166): this is the subjective meaning component beyond the evidential meaning. In contrast to Olbertz’s line of argumentation, Company Company (2006) analyses dizque in line with dale!, tate!, ándale! and sepa! as a discourse marker. Discourse markers are defined by an increase of their scope which makes them typical items to express subjectivity. Olbertz has convincingly shown that 1. the property of subjectification must not be tied to an increase of a lexical element’s scope and 2. dizque should not be termed a discourse marker.85 The Italian equivalent dice che is analysed by Pietrandrea (2007) as an evidential adverbial construction. 85

Travis’ (2006) and Babel’s (2009) articles confirm that dizque is a grammaticalised marker of evidentiality. While Travis (2006) studies the use of dizque in both spoken and written Spanish in Colombia, showing that dizque “is used both as a strict evidentiality strategy and as an epistemic marker” (Travis 2006: 1272), Babel (2009) analyses the use of diz(que) in a contact variety of Spanish in central Bolivia. She shows that it “has undergone influence from Quechua to become a systematic reportative evidential marker in this variety of Bolivian Spanish” (Babel 2009: 487) and that it acts as a reportive evidential signalling hearsay (Babel 2009: 500-501). What Babel’s study differentiates from the studies by Olbertz (2007), Company Company (2006) and Travis (2006) is that it highlights the analysis of evidential markers against the background of sociolinguistics since, according to Babel, the “linguistic information encoded in reportative evidentials cannot be cleanly separated from social influences” (2009: 487).

147 In her paper “The grammatical nature of some epistemic-evidential adverbs in spoken Italian” Pietrandrea (2007) shows that the epistemic adverbs forse ‘perhaps’ and magari ‘perhaps/I wish’, the adverbial construction secondo me ‘in my view/in my opinion’ as well as the evidential adverbial construction dice che ‘says that’ “can be compared to grammatical forms, at least to an extent” (Pietrandrea 2007: 39), even though they are usually considered as fully lexical. The adverbs and adverbial constructions are considered “partially grammaticalized forms” in contrast to “more grammaticalized epistemic markers (such as modal verbs and epistemic future)” (Pietrandrea 2007: 40). So her study is two-fold: on the one hand it is shown that lexical expressions can be termed as being partially grammaticalised, and on the other hand it is shown with the help of the above mentioned adverbs and adverbial constructions that the “interaction of epistemic modality and evidentiality is particularly complex in Italian” (Pietrandrea 2007: 40). The study is highly context-sensitive or rather discourse-sensitive. Pietrandrea explains: An intuition underlying the present work is that the exact semantic characterization of metapropositional level units such as the epistemic adverbs cannot be obtained through the simple observation of their occurrence in the clause, but that in order to precisely grasp their meanings it is necessary to define the place these units occupy in larger portions of discourse. (Pietrandrea 2007: 45-46).

Her discourse-sensitive analysis of magari and forse leads her to refine the semantics of the adverbs, which are usually defined as generic hedges: magari is considered a non factuality marker and forse a possibility marker (cf. Pietrandrea 2007: 52): Magari emerges as a genuine non factual marker serving the function of putting forward as fictious the content it modifies. It can be employed in epistemic contexts, but it has not the evaluation of the truth of propositional content as its core meaning. Forse is an epistemic possibility marker. (Pietrandrea 2007: 58).

While magari and forse are considered purely modal, secondo me is said to represent the boundary between evidentiality and epistemicity (cf. Pietrandrea 2007: 52). In using secondo me the speaker is said to do “two different things at the same time: he expresses his own opinion and he refers to himself as the source of evidence for what he is asserting” (Pietrandrea 2007: 52). The following example is said to illustrate this: Secondo me Barbara e tornata. ‘In my opinion Barbara is back.’

So according to Pietrandrea (2007: 52), secondo me is an epistemic marker as well as an evidential marker; the speaker himself is the source of evidence.

148 Thus, secondo me is described as fulfilling a double function: [p] is simultaneously marked evidentially and epistemically. To my way of thinking, this raises the question whether it could be said that all speakers refer to themselves as a source of evidence when using I believe/think/guess/know etc., as long as there is an I (or a me/my). Or, asked in another way, is not the speaker always the source of information for the statement he utters as long as he does not explicitly state that he has another information source for the statement? Concerning secondo me, Pietrandrea finally concludes that the adverbial construction is a true evidential (rather than epistemic) marker as it has a ‘source of information’ as its core meaning (Pietrandrea 2007: 54). She underpins this fact by a test, employing secondo me in purely conjectural contexts. She explains: Let us imagine that a friend has lost his watch and that he asks us where we think that the watch can be. If we do not know anything about that watch, nor on how our friend normally uses it, we can only legitimately use a purely epistemic marker and try: L’avrai lasciato a casa ‘You may have left it at home.’ More awkwardly we would say: ?

Secondo me l’hai lasciato a casa

‘In my opinion you left it at home’ (Pietrandrea 2007: 54).

After having termed the adverbial construction an evidential marker, Pietrandrea raises the question of what kind of evidential it is. To answer this question, Pietrandrea borrows a test from Squartini (2008)86. She tests whether secondo me is compatible with reportive contexts, with contexts of direct evidence and contexts in which a circumstantial inferential process is expressed. Secondo me is neither compatible with the first context nor with the other two (cf. Pietrandrea 2007: 54-55). It can only be used in inferential generic contexts, where previous experience or general world knowledge is the inferential basis. Hence, secondo me is finally said to be “a generic inference marker stressing the speaker’s opinion or knowledge as the source of inference” (Pietrandrea 2007: 55). The last construction she considers is a pure evidential: the Italian reportive form dice che. It is a pure evidential as it does not imply “any qualification of the speaker’s commitment towards the propositional content” (cf. Pietrandrea 86

Pietrandrea (2007) had access to Squartini’s paper before it was published. She refers to Squartini’s paper as ‘forthcoming’. It was in 2008 when it was published.

149 2007: 55). While dice che is used as a reportive evidential, the parenthetical dice – more used in colloquial Italian – is used as a quotative, i.e., it marks direct speech (cf. Pietrandrea 2007: 56, 57). Giacalone Ramat/Topadze (2007) call dice a form that “is on its way towards grammaticalization” because it is mostly used “as a marker of direct speech, but also of indirect speech, and is morphologically invariable and positionally mobile” (Giacalone Ramat/Topadze 2007: 27). Pietrandrea furthermore shows that dice che “may sometimes acquire an epistemic overtone by means of a pragmatic implicature” (Pietrandrea 2007: 57; cf. Olbertz 2007 on dizque), although epistemic modality does not represent a meaning aspect that is encoded by the construction. In contrast to Mexican Spanish dizque, which has a core epistemic meaning when it modifies linguistic elements below constituent level (see above), dice che “always has a clausal scope” and “epistemic modality is by no means embedded in its meaning”, whereby it may have “epistemic overtones” (Pietrandrea 2007: 57). The discourse-sensitive analysis of the epistemic and/or evidential use of forse, magari, secondo me and dice (che) represents one side of Pietrandrea’s paper. The other question it is concerned with is how the grammatical status of the adverbs and adverbial constructions can be explained. Referring to Lehmann’s grammaticalisation criteria (see, for instance, Lehmann 2002), Pietrandrea explains that ‘reduced autonomy’ represents the condition for considering them as grammatical. This reduced autonomy “may simply reside in their integration in a closed paradigm of options and in their reduced syntactic mobility” (Pietrandrea 2007: 60). The adverbs and adverbial construction form a closed class of homogeneous and complementary options, as they convey grammatical meaning and show reduced syntactic mobility, and as such they are considered as “weakly grammaticalized epistemic forms, rather than fully lexical” (Pietrandrea 2007: 60). Using ‘evidence from Spanish’, Cornillie’s paper (2007b) deals with the continuum between lexical and grammatical evidentiality, more precisely, it analyses grammatical as well as lexical uses of Spanish parecer.87 He emphasises that [b]y doing so, the approach differs from Aikhenvald’s (2004) claim that evidentiality should refer to an exclusively grammatical, obligatory category of a language. Following this view would entail that evidentiality can not be considered from a 87

Comparing the different functions of seem, depending on the construction it is part of (seem with different complementations structures, seem in impersonal constructions, seem in combinations with an experiencer), Aijmer (2009) regards English seem to as the most grammaticalised form and as an evidential marker which is considered similar to the modal auxiliaries.

150 functional perspective and would imply that most European languages do not have an evidential category (Cornillie 2007b: 108).

For parece que and parece ser que he summarises that both hearsay and inference readings can be conveyed (Cornillie 2007b: 123-124), which correlates with the results in his work (Cornillie 2007a) that is considered next. Although both constructions yield either hearsay or an inferential reading, concerning the type of inference, parece ser que and parece que differ as the former “is not provoked by visual or auditory evidence” (Cornillie 2007b: 119). Hence, only parece que can rely on inference from visual or auditory evidence (Cornillie 2007b: 115-116). “[T]he more grammaticalized parecer + infinitive is restricted to one evidential type only”, that is, to inference from direct evidence or reasoning (Cornillie 2007b: 114). The me/le parece que construction is said to be found in the medium position of the continuum between lexical and grammatical evidentiality (Cornillie 2007b: 116). This construction “involves belief or relies on inference, but can also be a mere subjectifier without evidential qualification whatsoever” (Cornillie 2007b: 118). Según parece, al parecer, me parece and por lo que parece are considered constructions belonging to “parenthetical parecer” (Cornillie 2007b: 121) and are said to convey various readings (Cornillie 2007b: 124). With regard to the al parecer construction, the readings listed differ slightly from the table in Cornillie’s study on (semi-)auxiliaries (2007a: 36). There, al parecer is said only to convey a hearsay reading. This study will be reviewed in the following section.

3.6 The relation between evidentiality and epistemic modality in Spanish In Evidentiality and Epistemic Modality in Spanish (Semi-)Auxiliaries. A Cognitive-Functional Approach Cornillie (2007a) analyses the evidential/epistemic use of different parecer constructions, of the modal verbs poder, deber and tener que as well as of resultar, prometer and amenazar. As his work on the modal verbs is going to be considered in more detail anyway (cf. chapter 6) and for reasons of space, I will only briefly summarise his main results on the parecer constructions, paying special attention to al parecer – also in terms of polyphony. Cornillie shows that the different parecer constructions correlate with various combinations of evidential values, whereby al parecer is a hearsay reading applied to:

151 Table 8: The parecer constructions and their evidential values (cf. Cornillie 2007a: 36) BELIEF Me/le parece que me parece

INFERENCE HEARSAY parecer + que-clause parece ser + que-clause parecer + infinitive según parece

al parecer

Consequently – and as already indicated above (cf. chapter 2.1.4) – al parecer can be said to represent an implicit form of polyphony, as this adverbial locution + [p] can represent a statement that did not originate in the current speaker/writer. In the following I will provide a few examples which illustrate that al parecer – just as the journalistic conditional – may belong to the sixth point of the scale of polyphonic implicitness: (41) [...] Finalmente, la mala noticia la anuncia Semana en su portada. No todo podía ser tan bonito este verano y, al parecer, Marta Sánchez está pasando por una delicada crisis en su matrimonio con Jesús Cabanas. La mala situación personal de la pareja se hizo pública cuando la cantante acudió sola a la boda de Lolita [...] (El País 07/08/2010) (42) [...] detectaron que sobre el sujeto existía una orden de la Interpol de búsqueda, captura y extradición cursada por Pakistán. El sospechoso es un desconocido en la ciudad: llegó a Tortosa hace unos dos o tres meses, señalan fuentes municipales. Al parecer, se había instalado en un domicilio que compartía junto a otros compatriotas. Las autoridades pakistaníes disponen ahora de 40 días para remitir al juez la documentación en la que sustentan la petición de extradición. (El País 07/08/2010) (43) El otro día me espantó leer un reportaje sobre la violencia y corrupción de la policía argentina. Al parecer comete secuestros, extorsiones, asesinatos. Un 61% de los argentinos cree que los policías están involucrados en el crimen, y el 15,6% ha tenido que pagarles sobornos. (El País 10/09/2010) (44) Según informa The Guardian, al final de ese día Morlock le señaló a Holmes que el asesinato tenía por objetivo la diversión y le amenazó si se lo decía a alguien. La segunda víctima, Marach Agha, fue disparada y asesinada el mes siguiente. Al parecer, Gibbs le disparó y colocó un fusil cerca del cuerpo para justificar su muerte. En mayo, Mulá Adadhdad fue asesinado después de ser disparado y atacado con una granada. (El País 12/09/2010) (45) Según fuentes próximas al caso, sus testimonios han sido valiosos, y todavía prometen serlo más los de los dos últimos (el responsable de un club y un agente) que se acaban de ofrecer a participar. Al parecer, detrás de la decisión de uno de ellos subyace el enfrentamiento con otro de los principales imputados por una cuestión sentimental. (El País 15/09/2010)

152 Indeed, what examples (41-45) illustrate is that an implicit form of polyphony is only to be found on the textual level, also considering the context (cf. Gévaudan 2008: 6). Analysing al parecer + [p] only, would not have led to a succesful interpretation of al parecer being used quotatively; it is the context which shows this. The underlined phrases that represent in each case sources of [p], which all contain al parecer, illustrate that [p] can be attributed to the source already mentioned and that al parecer is used to express that [p] belongs to the other voice – and not to the one of the journalist. It seems to be even more difficult to see that al parecer in the following example is also used to convey hearsay readings, as it is not the ‘explicit context’ that helps to interpret the use of the adverbial locution. In text passage (46) al parecer somehow ‘has to’ convey a hearsay reading as – by applying the process of elimination to the possible meanings – no other logical meaning that fits in this context remains. How could the journalist know that Mike Edwards ‘died instantenously’, if not from someone else? (46) El músico británico Mike Edwards, fundador del grupo de rock Electric Light Orchestra (ELO), ha muerto a los 62 años al ser arrollada por una bala de heno de 600 kilos la furgoneta que conducía por una carretera del condado de Devon, Reino Unido. Al parecer el músico, que tocaba el chelo en la formación original de la banda, murió de manera instantánea. (El País 08/09/2010)

In contrast to Cornillie, who has shown that the only evidential use of al parecer is the hearsay one (cf. Cornillie 2007a: 36) – which is also in contrast to Cornillie’s paper (2007b: 123) where he says that the al parecer construction may be based either on reported information or on an inference – it is to highlight that the adverbial locution can really convey an inferential reading: (47) [...] se modifican cupos energéticos en renovables o se toma el peor camino para defender -aunque sea justamente-, la protección medioambiental de Doñana. Y Andalucía sigue sin ser noticia, porque al parecer es más importante discutir una transferencia al País Vasco o el encaje legal del Consejo del Poder Judicial de Cataluña, que las políticas de vivienda, energéticas o de empleo. (El País 06/09/2010)

Example (47) contains an affective, ironical conclusion that Andalusia is still without news because – as it seems – other things are more important. There is no reason to apply a hearsay reading to this use of al parecer. The next example is ambivalent; even the broader context does not help to determine whether al parecer is used quotatively or inferentially. In example (48) two synthetic future forms are underlined, which show that [p] is (not) going to happen in future time. On the one hand, one could argue that the future forms are indices to interpret the adverbial locution as being quotatively used because how could the journalist otherwise know what is (not) going to happen, if not from someone else? On the other hand, if the synthetic future form is used here to express in-

153 ference, this would not match with a hearsay reading of al parecer. Following this line of argumentation, al parecer must also be used inferentially. So two interpretations are possible: (48) El próximo domingo, los Reyes ofrecerán en su residencia estival un almuerzo en honor a la primera dama estadounidense, Michelle Obama, que les visitará tras pasar cuatro días en Marbella. Quien al parecer no asistirá a la comida será Sasha, la hija menor de los Obama, que viaja con su madre. (El País 07/08/2010)

In example (49), the evidential subcategories hearsay and inference seem to slightly overlap, although it seems more appropriate to assume that the journalist uses al parecer quotatively. Nevertheless, the quoted speaker(s) from whom he got the information must have inferred [p] which the journalist then transmits. If the driver was found in the wood beside the freeway Portocolom-Manacor, it surely was reasonable to assume that he was on the way from the one place to the other. Hence, the journalist quotatively transmits information that was inferred. Another possible interpretation is that the quoted speaker(s) the journalist got the information from interrogated the driver’s family/friends. Thus, al parecer is quotatively somehow ‘double-marked’. However, its use is clearly evidential: (49) El coche, un Subaru Impreza de alta cilindrada, era propiedad del joven, quien al parecer conducía desde Portocolom y se dirigía a la localidad mallorquina de Manacor. En el accidente, en el que no se han visto implicados otros vehículos, el coche se ha adentrado 150 metros en el bosque llevándose por delante varios árboles [...] (El País 13/09/2010)

In sum, Cornillie’s cognitive-functional approach to Spanish (semi-)auxiliaries shows how to combine a well theoretically funded study of lexical and grammatical means that convey evidential and/or epistemic meaning with a finegrained linguistic analysis of the different uses of (semi-)auxiliaries.

3.7 Evidentials and modal meanings in Guaraní Giacalone Ramat/Topadze (2007) analyse Italian (si) dice che, Company Company (2006) and Olbertz (2007) study Mexican Spanish dizque and Pietrandrea (2007) the Italian evidential construction dice que. Große (2011)88, in contrast, is concerned with Guaraní ndaje, mostly translated as Spanish dicen que ‘they say that’ or se dice que ‘it is said that’ (Große 2011: 242). Guaraní provides markers having a modal as well as an evidential meaning; they are usually described as only being modal, even though they primarily encode two types of evidentiality: 88

Große’s study from 2000 is concerned with the analysis of linguistic means that serve to express evidentiality in Brasilian Portuguese.

154 direct (witnessed) and hearsay (Große 2011: 239). Ndaje thus represents a hearsay marker, although Ayala (1993) remarks that this marker may also be used if a particular person is quoted: En su significación y uso “ndaje” no es del todo equivalente al impersonal castellano “dicen”. “Dicen” es del todo impersonal y anónimo, mientras que “ndaje” se puede usar aún en el caso de que quien lo dijo sea una persona determinada y conocida; lo que tiene de propio la partícula “ndaje” es que el hablante se limita a transmitir lo dicho, eludiendo toda responsabilidad con respecto a la verdad de lo dicho (Ayala 1993: 215; quoted in Große 2011: 242-243).

Ayala’s words additionally show how short the distance is in Guaraní between epistemic modality and evidentiality: regardless of whether the words of a known person are marked by ndaje or whether the marker is used to convey hearsay reading, the speaker, at all events, indicates that he is not/does not want to be responsible for the transmitted information (cf. chapter 7.1 on the use of the journalistic conditional). Because of the language contact between Guaraní and the Spanish language variety spoken in Paraguay, ndaje is also used by those speakers of Spanish who command both languages (Große 2011: 244). For her analysis, Große consulted Paraguayan daily newspapers (online) in a two-year period, distinguishing between a newspaper of the standard language and a journal being morphosyntactically more influenced by Guaraní (cf. Große 2011: 244). Subsequently Große offers a variety of interesting examples like the following one: Después de presentarse radicalmente tuneada en TV, Benigna Leguizamón, la segunda pichucha de Fernando Lugo, ndaje recibió propuestas de 100 candidatos, y todos para casamiento (Große 2011: 246).

On the one hand, this example confirms that ndaje is used as a hearsay marker and on the other, it illustrates very well the language contact on the morphosyntactic level. Even more interesting are the following two examples because ndaje is used in connection with aparentemente and supuestamente, that is, with adverbs that are also shown to convey a hearsay reading (cf. chapters 5.1.1 and 5.1.7 of the present study): El muchachón tenía aparentemente un “ojo reventado” y aseguró a los caquis que malandros lo atacaron queriendo robarle, pero que él se defendió de los mismos a puñetazo limpio ndaje. Pero esta versión de lo que le ocurrió no convenció para nada a los caquis, como avei a los funcionarios del hospital de la calle General Santos, quienes contaron que le golpearon supuestamente cuando fue pillado por algún titular en off side (posición adelantada) con alguna dama ndaje (Große 2011: 249).

155 Große analyses these uses of ndaje as follows: especially when used in connection with further elements that function as an evidential marker, ndaje serves the primary function of conveying distancing from the transmitted information and doubt towards the truth of the proposition (cf. Große 2011: 249). Otherwise the information would be evidentially double-marked. In my opinion, the underlined phrases (my emphasis) in the second text passage emphasise that ndaje modalises – instead of evidentially marks – the information because a hearsay marker would not make any sense here: if ‘some officials told that [p]’, why then mark the same proposition with a hearsay marker? Große furthermore observes a relation between the meaning of ndaje and its syntactic position: the modal reading seems to be favoured if the marker is to be found at the end of the information (cf. Große 2011: 249). These interesting results leads Große to tentatively suggest that ndaje is en example of grammaticalisation at the discourse level because the particular use of ndaje where it is posposed to [p] is bound to newspaper style. The language contact between Paraguayan Spanish and Guaraní may have initiated ‘an alteration of the grammatical system of evidentiality that is due to language contact’ (Große 2011: 252).

3.8 Evidentiality and its grammatical/lexical realisations in various languages Recently, Diewald/Smirnova published – besides their extensive study on Evidentiality in German (2010a) – the collected volume Linguistic Realization of Evidentiality in European Languages (2010b), in which the papers by Alcázar (2010), Haßler (2010a) and Cornillie (2010b) deal with – roughly speaking – evidentiality in Romance languages. In the parallel corpus study on information source in Spanish and Basque, Alcázar (2010) analyses, in a contrastive way, evidential markers in Basque, which are grammatical in nature, and their non-grammaticalised equivalents in Spanish. “The focus of the study is omen, an alleged hearsay evidential marker” (Alcázar 2010: 131) and its lexical translations in Spanish. He describes the use of omen as grammatical evidentiality for several cogent reasons: omen is a particle (a member of a closed set), it is dedicated to the expression of hearsay, and it is a proclitic. In effect, omen may be considered part of verbal inflection (Alcázar 2010: 133).

For the particle dedicated to the expression of hearsay he lists several Spanish evidential strategies to translate omen, such as al parecer, aparentemente, supuestamente or dice/n (Alcázar 2010: 137), that is, several lexical means to express hearsay evidence. Thus,

156 [t]he corpus data reveals that Spanish resorts to a diverse set of lexical means, or evidentiality strategies in the sense of Aikhenvald 2004, to express meanings that are reduced to the use of omen in Basque […] (Alcázar 2010: 133).

So he shows that Basque “fit[s] into typologically acknowledged evidential systems” (Diewald/Smirnova 2010b: 7) and concludes that “Spanish and Basque […] represent two different evidential systems, broadly conceived as lexical and grammatical, respectively” (Alcázar 2010: 131). Despite discussing the expression of evidentiality and epistemic modality in Spanish, Italian, Portuguese and French, with her paper “Epistemic modality and evidentiality and their determination on a deictic basis: the case of Romance languages” Haßler (2010a) shows that the categories of epistemic modality and evidentiality may be disentangled with the help of the application of deictic characteristics. She describes evidentiality as a deictic phenomenon because “evidentiality markers refer to elements outside the linguistic context, namely to the source of information and to the speaker who has access to it” (Haßler 2010a: 237). After analysing several papers that treat evidentiality more or less explicitly as a deictic phenomenon (cf. Haßler 2010a: 227-232), she raises some problems in distinguishing between epistemic modality and evidentiality in the case of some Romance adverbs (cf. Haßler 2010a: 232-237) so that the introduction of deixis as the category to distinguish between evidentially and epistemically used expressions seems to be justified: Deixis implies a perspectivation from the standpoint of the speaker, which the hearer, outgoing from his/her own discursive position, has to reinterpret. This concept of implicit perspectivation in deixis allows a differentiation of epistemic modality and evidentiality. […] While the epistemic modality monologically contributes the epistemic stance of the speaker or the author, evidentiality can also require the hearer or reader to produce a reference to the source of information, which means a necessary adaptation with respect to his/her identity, individuality and epistemic stance toward the sources (Haßler 2010a: 239).

Hence, evidentiality is not only perceived as a reference to an information source, but also as a deictic phenomenon “which refers to the speaker who conceived the utterance and to his complex relationship to the hearer, to information and its sources” (Haßler 2010a: 243).89 Cornillie (2010b) analyses the use of epistemic and evidential adverbs in Spanish conversation, that is, taking an interactional approach to Spanish epistemic adverbials and adverbs such as a lo mejor, tal vez, aparentemente, evi89

See also chapter 2.1.2.1 of the present study: for Diewald (1999) and Diewald/Smirnova (2010a) epistemic modality is also conceived a deictic category.

157 dentemente or obviamente. Analysed in conversation, the adverbs are shown to have – despite adding an epistemic and/or evidential value to the proposition they go with – certain interactional functions. For evidentemente, for instance, it is shown that […] the evidential adverb [evidentemente] mainly involves inferences from reasoning. Yet, the reasoning is not so much triggered by sensorial, visual or auditive stimuli but is part of the discourse planning. It can be argued that there is a semantic shift from a purely evidential reading (i.e. something that can be clearly perceived) to a discourse related reading (i.e. something that is clear according to discourse patterns) (Cornillie 2010b: 327).

From the conversational perspective, the closer look at a lo mejor – to give a further example – has revealed that it […] is not only used to introduce hypotheses with new information, but also invites the interlocutor(s) to confirm or reject the view of the state of affairs (or part of it) presented, that is, the adverb plays a role in the turntaking process (Cornillie 2010b: 328).

So in this paper Cornillie goes beyond the traditional description of Spanish adverb(ial)s and their epistemic/evidential qualifications and illustrates their different roles in the organisation of the turn taking process in interactions.90 90

Also Capone’s study (2001) is, roughly speaking, concerned with the interactional functions, even though it does not focus on turn taking processes etc. The study analyses the discourse-pragmatic functions of some English adverbs. Capone adopts a neo-Gricean framework to “discuss a range of adverbs, such as obviously, certainly and really” (Capone 2001: 13). Grice’s notion of conversational implicature – standing for “messages which are conveyed without being actually said” (Capone 2001: 15) – is used to find out in which type of context(s) the linguistic expressions under discussion are most likely to occur (cf. Capone 2001: 13). He consequently focuses “on the relationship between the meanings of certain modal expressions, their contexts of use and the functions they are intended to convey in those same contexts of use” (Capone 2001: 13). The application of Grice’s notion of conversational implicature to the analysis of the semantics and pragmatics of the adverbs and to their pragmatic/semantic inferences “associated with the use of certain modal expressions” (Capone 2001: 23) forces Capone to analyse the expressions in question in dialogues. In the essay’s part on semantic and pragmatic considerations concerning obviously Capone explains: “By saying Obviously p one will conversationally implicate that some doubt is in the air or that the speaker is engaging in a discourse-related activity (either arguing or conceding a point, among other things)” (Capone 2001: 50-51). Capone further explains that – by applying Grice’s maxim “that one should not vocalize an assumption that is obvious or taken for granted” (Capone 2001: 51) to the analysis of an utterance like Obviously [p] – the speaker signals that he is willing to cooperate at a deeper level, that is, he is willing to engage in a discourserelated activity with the interlocutor (cf. Capone 2001: 51). For certainly Capone (2001:

158

3.9 Evidentiality and modal certainty in English adverbs Obviously and certainly are, among other adverbs, not only dealt with by Capone (2001) but also by Simon-Vandenbergen/Aijmer ([henceforth: S-V/A] 2007). They themselves describe their extensive monograph The Semantic Field91 of Modal Certainty. A Corpus-Based Study of English Adverbs as “a modest attempt to take up the challenge of constructing a semantic field of cer-

91

76) ascertains that the adverb is especially used in situations where speakers express their uncertainty about a particular situation (cf. also Simon-Vandenbergen/Aijmer 2007). In other words, the truth-conditional meaning of the utterance containing certainly clashes with the (real) situation of the utterance (cf. Capone 2001: 76). This leads Capone to the following conclusion concerning the pragmatic component of the adverbs discussed: they “will be used in situations where doubt is envisaged or when one is engaging in an argumentative act” (Capone 2001: 78). Furthermore, he describes modal adverbs as being affected by the semantic bleaching process, describing this ‘affectedness’ as being due to the adverbs’ pragmatic implicatures: “[…] modal adverbs are subject to semantic erosion due to their pragmatic implicatures which might lead to systematic change and replacement of the relevant lexemes belonging to this area of the lexicon” (Capone 2001: 87). In reviewing Capone’s essay, Hoye (2005: 1485; see also Hoye 1997: 208-212) sees his argument supported that “in certain instances, modal adverbs can be treated as ‘modal particles’” and thus subscribes to Capone’s theory of ‘semantic erosion’. But, especially concerning obviously and certainly, speaking about ‘semantic erosion’ should be avoided, rather dealing with the adverbs in terms of ‘pragmatic broadening’. As, for instance, obviously does not only mean ‘something is obvious’ but can also convey inferential readings (cf. chapter 5.1.3 on obviamente), its meaning should not be termed as ‘semantically eroded’ but as ‘pragmatically broadened’. An adverb may fulfil functions which go beyond the function that is bound to the meaning aspect that is encoded by it. To allege another example: the adverb evidentemente may be used to express what is evident. In that case it fulfils the function which is bound to the meaning aspect that is encoded by the adverb. But the adverb may also be used to convey an inferential reading; then it fulfils a function that goes beyond the function that is bound to the meaning aspect which is encoded by evidentemente (cf. chapter 5.1.2). To sum up, one should rather speak of ‘pragmatic broadening’ instead of ‘semantic erosion’ because the adverbs have acquired (or are acquiring?) new pragmatic (and discourse) functions. They explain that they use the term ‘semantic field’ to refer to a more general, less specific notion than ‘semantic map’. It simply denotes that the adverbs of modal certainty belong together semantically, without specifying in what ways they relate and without suggesting any type of structuring which has been associated with semantic fields in different studies (S-V/A 2007: 23).

159 tainty as expressed through modal adverbs in English” (S-V/A 2007: 1). The adverbs they discuss have been roughly divided into adverbs which the speaker uses to indicate that he has evidence and adverbs which signal that the speaker does not have evidence. Clearly, for instance, represents an example of the former group, certainly, in contrast, belongs to the latter (cf. S-V/A 2007: 3). The adverbs then have been further classified “as epistemic, evidential, expectation and speech act adverbs depending on their properties” (S-V/A 2007: 24). But their study shows that this classification only makes sense on the lexical-semantic level: As “[t]he same adverbs may be used for several functions” and as “there are cross-overs from one category to another”, the classes cannot be considered categories with strict boundaries (S-V/A 2007: 24). By their interactive approach to modal adverbs “which takes into account the dialogical nature of all verbal communication” (S-V/A 2007: 4) they show that modal adverbs are not only used to express the speaker’s commitment to the truth value of the proposition but also to serve interactive, conversational functions. So the function an adverb serves is context-dependent or situation-dependent. Simon-Vandenbergen/Aijmer thus assume that adverbs have both conceptual or contentful as well as procedural meaning: An adverb has a specific evidential or epistemic modal meaning. We will refer to this as the conceptual meaning of the adverbs and distinguish it from the procedural meaning. For example, certainly is contentful in that it means epistemic certainty and procedural when looked upon from the perspective of indexing the speaker’s or writer’s stance to the text or one of the participants (S-V/A 2007: 54).

For their analysis of adverbs, they rely on three types of language data: monolingual corpora, translation corpora and informant testing to get complementary information (S-V/A 2007: 323). In order to find out how speakers and writers use the adverbs in question for different purposes they examine language data of corpora of present-day English (mainly the International Corpus of English; British English). The British National Corpus was sometimes consulted for the purpose of comparison (S-V/A 2007: 6). They used translations to analyse the multifunctionality of the adverbs and to establish meaning relations between them (S-V/A 2007: 21), whereby they base the translational approach on many different languages. The informant tests represent the third part of their methodology because the corpora are indeed of limited use when two or more adverbs are very close in meaning and seem to occur in exactly the same environments. It is also interesting to test whether a sentence is acceptable when we exchange one adverb for another (SV/A 2007: 8).

160 The informant tests contained substitution tests where one adverb was substituted by another. In this way they were able to elicit “subtle functional differences between closely related adverbs” based on the intuitions of native speakers (S-V/A 2007: 323). Altogether, a total of 1816 tokens of different modal adverbs such as certainly, definitely, no doubt, surely, indeed, evidently, clearly, of course, obviously or necessarily is analysed (S-V/A 2007: 69-70). The overall aim of their study is to create a semantic map for the adverbs of modal certainty, whereby I would prefer the term ‘semantic-pragmatic map’ as “[t]the dimensions and features which structure the field are discovered on the basis of the syntagmatic analysis and by comparing adverbs of certainty which are close to each other in meaning”. (S-V/A 2007: 13). The syntagmatic analysis has, for instance, revealed that certainly occurs frequently with a following butclause which implies that this adverb usually conveys a concessive reading (cf. S-V/A 2007: 21-22). However, the meaning of one and the same adverb highly depends on the ‘subject variable’. “The interpretation of particular adverbs will in some cases depend on whether the subject of the clause refers to the speaker, the addressee or a third person” (S-V/A 2007: 22). For their analysis SimonVandenbergen/Aijmer adopt a ‘bottom-up approach’, that is, they start with looking within the clause at the position and scope of the adverb, then go beyond the clause and look at the sentence level to finally end with the adverbs’ analysis in the larger co-text (S-V/A 2007: 81-82). The last step implies that they also focus on the interactive, communicative, roughly speaking, pragmatic, functions the adverbs (may) fulfil: Finally, examining the syntagmatic behaviour of the adverbs beyond the clause involves looking at the larger co-text, preceding and following utterances. This means that we have to go beyond the sentence to look for clues regarding interpersonal relations and information structuring factors which have prompted the use of the adverbs. Logical and rhetorical relations which are expressed within the sentence in one case may indeed be expressed between sentences in other cases (S-V/A 2007: 82).

The semantic relation between certain adverbs is consequently double-checked: a syntagmatic approach combined with an approach that is based on translations provides evidence for the adverbs’ relations they have or do not have from two independent sides. Studying modal adverbs, Simon-Vandenbergen/Aijmer also see themselves being confronted – besides the term ‘modality’ – with the notions of evidentiality, subjectivity and deixis: the term ‘modality’ they use to refer to the meaning that is encoded by all (!) adverbs they analyse because

161 [t]hey all express particular degrees of speaker certitude concerning the truth of the proposition. These degrees are, however, not constant in any particular adverb but may vary according to the context […] (S-V/A 2007: 38).

The adverbs may also convey a subjective or intersubjective reading (in Nuyts’ terminology), which are clearly context-dependent. A few adverbs such as obviously, clearly or evidently are considered evidential (cf. S-V/A 2007: 38). Addressing the above mentioned third term, deixis, Simon-Vandenbergen/Aijmer adopt Traugott/Dasher’s (2002) argument that epistemic modality has something in common with deixis: as epistemic modality indicates the distance between the speaker and the actual world he refers to in his proposition, modal adverbs do so as well (cf. also Diewald 1999 and Diewald/Smirnova 2010a). “[I]f an adverb expresses uncertainty, this indicates a greater distance from the actual world than in the corresponding sentence which is not modalized” (S-V/A 2007: 44). Their ‘three-folded’ approach has, for instance, revealed that the adverbs no doubt and surely carry the meaning of ‘less than absolute certainty’ and that they are usually used when the speaker wishes to convey doubt, simultaneously asking his interlocutor(s) for agreement. At this, surely and no doubt clearly differ from certainly, definitely and indeed (cf. S-V/A 2007: 145). Another example is that obviously has been shown to be ‘really’ evidential because some French translations gave proof for this analysis: the most frequent French translations of obviously were évidemment, de toute évidence or à l’évidence – clearly being evidential expressions (cf. S-V/A 2007: 158) and thus confirming the inherent evidential status of obviously. But obviously is not alone: clearly is very close to this adverb – semantically as well as pragmatically (S-V/A 2007: 166). The only difference they could find is concerning their usage in written or oral speech: “[…] clearly turns out to be more frequent in writing than in speech, and this is in sharp contrast with the closely related obviously” (S-V/A 2007: 197). And […] obviously has developed further interactional functions in conveying solidarity, while clearly has stayed within its evidential domain and conveys less speaker involvement: it is more an adverb of intellectual reasoning than one of interpersonal negotiation […] (S-V/A 2007: 199).

Because of the fact that Simon-Vandenbergen/Aijmer furthermore combined different genres in their analysis of the adverbs, including text-type classified information, they not only obtained a versatile picture of the adverbs’ various pragmatic functions but also of their usage in a social context: they had a closer look “at who uses them to whom, when and for what purpose” (S-V/A 2007: 195). The main distinction was drawn between speech and writing, whereby the former comprises, for instance, various types of monologue (broadcast news, spontaneous commentaries etc.) and dialogues such as direct and telephone con-

162 versations or parliamentary debates. The latter involves, inter alia, student essays, social letters, novels, academic writing and media news reporting. To take up again the examples from above: surely, for instance, never occurs in scientific writing; its occurrence in writing is generally very rare. It was found in novels containing dialogue. So it is an adverb which is only orally used (SV/A 2007: 235). Clearly and obviously were shown to often occur in argumentative contexts, whereby they differ according to their ‘addressee-orientation’: “obviously is more addressee-oriented, more solidarity-building and also more dismissive. Clearly suggests more solid evidence” (S-V/A 2007: 226). To sum up, Simon-Vandenbergen/Aijmer (2007: 247) have shown “that items which seem to be near-synonyms on the basis of dictionary definitions have yet developed different functions”, which is the reason why I would speak about ‘pragmatic broadening’ in this connection. Through their bottom-up approach, their consideration of different text types and their ‘double-checking’ of the adverbs’ functions by taking into account translations the authors are able to give a fine-grained picture of the different functions and uses of adverbs of modal certainty. Their analysis has furthermore revealed that in the case of English certainty adverbs modality is the encompassing notion which includes evidentiality. It has further appeared that types of evidence are not strictly separated by English adverbs. Perception, reasoning, and hearsay tend to be expressed by the same forms, and the exact nature of the evidence is typically to be derived from the context (S-V/A 2007: 322).

The authors exemplarily studied the English modal adverbs. There is only one detail that cannot methodologically be agreed with: when taking into account the different genres they explain: As the text categories in the ICE-GB vary in size, the figures have been normalized to one million words. […] For example, certainly occurs 63 times in direct conversations, which comprise 180.000 words. This gives a frequency of 350 occurrences in one million words (S-V/A 2007: 200).

Perhaps it sounds a little bit exaggerated, but: would it be sufficient if – provided that I work quantitatively – I look at five instances of each linguistic item that is the subject under discussion in the present study and then ‘normalise the figure’?

163

3.10 Evidentiality and epistemicity in English verbs of cognitive attitude “I reckon I know how Leonardo da Vinci must have felt…”. Epistemicity, evidentiality and English verbs of cognitive attitude (2007) represents another profound study that should be considered here. Cappelli’s exemplary study of the epistemic and/or evidential nature of English verbs of cognitive attitude, such as know, believe, think, doubt or suppose, is of high importance for the present study because Cappelli first shows that the class of verbs of cognitive attitude occupy “a rich semantic space in which epistemicity and evidentiality interact in intricate ways” (Cappelli 2007: 301) and secondly that, [d]espite the possible areas of overlap, […] each and every one of the members of the class has its own place within the system, even though contextual constraints can force the single verbs into the semantic space of the others (Cappelli 2007: 301).

The underlying hypothesis of Cappelli’s study is that a speaker who communicates that [p] – without adding any further epistemic qualification – communicates that he is certain with regard to the state of affairs he transmits. “The hearer is therefore entitled to think that the speaker knows the state of affairs presented for a fact and that he […] has sufficient evidence in order to qualify it as true” (Cappelli 2007: 301). This premise represented the starting point for determining why then a speaker explicitly indicates that he knows that [p]. Cappelli found out that the speaker does so if he wants to communicate that “he […] has reliable evidence to say that p is in fact the case” (Cappelli 2007: 302; cf. chapter 5.2.4 of the present study) or if he expects “that there might be reasons to refute his […] conclusion” (Cappelli 2007: 304). Hence, Cappelli has shown that the ‘most epistemic’ verb know may be used evidentially and that contextually provided information should be involved in the analysis of every single token of a verb of cognitive attitude. The other verbs of cognitive attitude are also analysed concerning their semantic potential and pragmatic use. She consequently does not only describe the semantic features of the verbs but also the pragmatic ones. In Cappelli’s study semantics and pragmatics are seen “as two facets of one and the same coin” because “word meaning is taken to be a dynamic construal emerging from contextual and cotextual pressures on conceptual material” (2007: 11). By describing the semantics of the verbs and their meaning forced by ‘pragmatic pressures’ she does not only shed light on the interaction of evidentiality and epistemicity in the verbs but also on the reasons that force a speaker to choose a certain verb in a certain communicative situation (cf. Cappelli 2007: 9). As she focuses on the use of the verbs in communication, Cappelli furthermore stresses the intersubjective – and one could say ‘interac-

164 tive’ – role of verbs of cognitive attitude: the intersubjective role relies on the fact that the verbs qualify the information provided by the speaker in the communicative setting also in consideration of what he […] expects his […] interlocutors to know, or how much he […] expects them to rely on him […] as a source of information (Cappelli 2007: 302).

The interactive role of the verbs refers to the fact that the speaker, by using a verb of cognitive attitude, signals the interlocutor that it is up to him to go through a verification process (cf. Cappelli 2007: 304), since by marking a proposition with a verb of cognitive attitude the speaker signals that the hearer should not (totally) rely on his knowledge. The examples are taken from the BNC (Cappelli 2007: 10), whereby reportive, non-qualificational and descriptive qualificational uses of the verbs are excluded from the analysis. The research, instead, focuses exclusively on the analysis of the verbs that are used to express the speaker’s own cognitive attitude. So it focuses on the analysis of the verbs in their qualificational non-descriptive use if occurring in the syntactic pattern I + verbpresent tense (cf. Cappelli 2007: 112). In the following I will (only) review how Cappelli contrasts know – roughly presented above – with the other “‘prototypical’ members” of the class (Cappelli 2007: 53), namely think and believe. The verb think can occur in a variety of contexts (cf. Cappelli 2007: 178). As far as evidentiality is concerned, the verb is said to be ‘neutral’; as far as certainty expressed by the speaker is concerned, the verb is considered ‘vague’ (cf. Cappelli 2007: 184). The meaning of think is thus regarded as highly context dependent: Where evidence is provided by the context, think is construed as the output of a computational process, whereas where the context does not make explicit reference to the reasons for the evaluator’s attitude, think only indicates that what follows is the subjective judgement of the speaker, his […] opinion (Cappelli 2007: 184).

So think and know can be opposed to each other as think is considered a pure epistemic verb, being no reference to evidence lexicalised. Know, in contrast, conveys that the speaker is certain that [p] “by virtue of the evidence he […] has to support his judgement” (Cappelli 2007: 199). The two verbs can be considered antonyms according to the speaker’s certainty: know conveys certainty, while think conveys uncertainty (cf. Cappelli 2007: 199). However, where evidence is contextually provided, even think may be said to convey an evidential reading (cf. the use of pensar in chapters 5.2.2 and 5.2.3). When a state of affairs – or, in Boye’s (2010b) terms, a proposition – is qualified with believe, the speaker conveys, on the one hand, that he is not cer-

165 tain that [p] is the case and, on the other hand, that he considers [p] likely to be the case, that is, he assigns to [p] “a positive place on the likelihood scale” (Cappelli 2007: 169). “The speaker is highly committed to the likelihood that a state of affairs is the case on the grounds of some sort of affective evidence that he […] has for it” (Cappelli 2007: 170), whereby the strength of the commitment may vary, which, in turn, depends on the context (cf. Cappelli 2007: 168). So believe involves both the epistemic domain and the evidential one. Because of its affective evidential dimension it is often strategically used in argumentative discourse (cf. Cappelli 2007: 178). Thus, both know and believe point towards the evidential domain, but they point to different domains within the evidential domain: “believe lexicalizes reference to some sort of ‘affective evidence’” and “know lexicalizes reference to verifiable, objective evidence” (Cappelli 2007: 199). Having opposed know with think and believe, Cappelli concludes that believe and know belong to those verbs that “lexicalize a higher level of subject’s commitment to the evaluation, probably deriving from the evidential dimension encoded” (Cappelli 2007: 184). Think, in contrast, can be used to express “various degrees of uncertainty according to the contextual elements” (Cappelli 2007: 184). The verbs’ characteristics are always summarised in tables that are subdivided into the features ‘evidentiality’, ‘epistemicity’ (and sometimes also ‘cognition’), whereby the feature of epistemicity is again divided into ‘likelihood’ and ‘certainty’ (cf. the chapters on the verbs of the present study). Cappelli furthermore investigates to what extent verbs of cognitive attitude are ‘affected’ by adverbial modifications and shows that the adverbs are affected by the verbs (and not the other way round) if co-occurring because they “take in their scope a ‘portion’ of the meaning of the verb” (Cappelli 2007: 108). The consideration of the adverbs that co-occur with the verbs is necessary in order to identify the semantic potential of each verb: [The] study of the occurrences of these verbs and in particular of the co-occurrences with certain adverbs, shows clearly that there are indeed differences between each and every one of them, which, however, sometimes can be overwritten by contextual effects […] (Cappelli 2007: 119).

To pose an example, the fact that believe co-occurs with, for instance, certainly, strongly or definitely (cf. Cappelli (2007: 175) underlines the fact that believe may be used by the speaker to convey that he is fairly committed, assigning [p] a positive degree on the likelihood-scale. The fact that believe can also co-occur with adverbs like scarcely or perhaps emphasises that it is also used if the

166 speaker is less committed. So the different adverbs “confirm the scalarity of the semantic dimensions lexicalized by believe” (Cappelli 2007: 175). In summary, on the semantic level, Cappelli illustrates the complex interaction of epistemicity and evidentiality and how this interaction is lexicalised in every single verb (cf. the tables in her study). So Cappelli shows – with respect to verbs of cognitive attitude – that epistemic modality and evidentiality are “strictly related and interwoven dimensions both at the conceptual and linguistic level” (Cappelli 2007: 83). The different functions (related to the interwoven dimensions) that verbs of cognitive attitude may fulfil are seen as “a consequence of their meaning potential” (Cappelli 2007: 102), whereby the meaning is construed out of this semantic potential and contextual information.92 Every member of the class of verbs of cognitive attitude is thus described as a complex dynamical micro-system which is highly affected by external pressures or contextual variables (Cappelli 2007: 112, 113, 119). The verbs are considered highly complex micro-systems at three distinct levels: at the conceptual level because of the interaction of the two linguistic domains; at the level of lexicalisation because the complexity at the conceptual level may be inherent in the verbs; as well as at the level of the single lexical items because each verb “can be envisaged as a complex and dynamical sub-system, which can reach a relatively stable state only if particular constraints act as organizing principles” (Cappelli 2007: 118). Their so-called loose internal organisation contributes to their complexity (cf. Cappelli 2007: 137). Furthermore, “since there can be no plausible semantic theory of lexical meaning without a hypotheses regarding the functioning of communication” (Cappelli 2007: 27), the study also deals with pragmatic and discursive issues so that the study contributes “to the understanding of the role and effects on the communicative exchange of the explicit encoding of the subject’s attitudes” (Cappelli 2007: 11). Even though the integration of a relevance-theoretic framework is not regarded as necessary (cf. chapter 4.5.2 of the present study), Cappelli’s study will be relied on in chapter 5 and in its subchapters where the use of some Spanish verbs of cognitive attitude is analysed and sometimes compared to the use of their English equivalents.

92

The present study distinguishes between meaning aspects that are encoded by a certain linguistic item and meaning aspects that are contributed by contextual information (cf., for instance chapter 4.3).

167

3.11 Evidentiality and deixis in English and German verbs of perception As the present study will not be concerned with the analysis of perception verbs, such as see, look, hear or feel, Whitt’s (2010) investigation of perception verbs in English and German will only briefly be touched upon. Whitt regards perception verbs as lexical means which a speaker may use to indicate the source for his proposition: […] perception verbs – verbs denoting sight, sound, touch, taste, and smell – are one of the primary lexical means speakers of English and German have at their disposal to convey the evidence for what they say (Whitt 2010: 1).

His study of this verb class is guided by the following questions: What kinds of evidential meaning do we find indicated by the perception verbs? Which perception verbs favor certain types of evidential readings over others? What types of complementation patterns do we find conducive to evidential readings? And do some of these complements favor certain types of evidential meaning? What kinds of evidential meaning or complementation occur more frequently than others? What about the role of polysemy, i.e. can the polysemy found among perception verbs such as see or sehen give rise to polysemic evidential meanings signified by these verbs? What are the frequencies of the various evidential constructions? Are there any recent historical developments regarding complementation patterns and evidential meaning? What are the salient similarities and disparities between English and German (Whitt 2010: 2)?

These are definitely many issues that are addressed in Whitt’s study. As the author considers it necessary to include a historical perspective in order to see “any recent trends or developments [that] are occurring, such as the evolution of new complementation patterns and/or evidential uses” (Whitt 2010: 48), the language data were retrieved from different corpora and the Digitales Wörterbuch der deutschen Sprache: the Helsinki Corpus that contains texts from Old, Middle and Early Modern English as well as different English dialects; the Representative Corpus of Historical English Registers that partly overlaps with the Helsinki Corpus as it contains texts from 1650 to 1990; the Bonner Corpus containing texts of Early New High German as well as the Goethe Corpus and the Kant Corpus, which provide texts that are representative for the German of the eighteenth and nineteenth century (cf. Whitt 2010: 49-51). Usually, every study that, roughly speaking, deals with evidentiality defines or – at least – clarifies its understanding of evidentiality. This is what chapter 1.1 – according to the title ‘Defining Evidentiality’ – is concerned with. Without really explaining, illustrating or aiming to (convincingly) argue that evidentiality is deictic, it is conveyed as a fact but ‘covered’ as an argumentation. Between

168 the review of Jakobson’s essay on “Shifters, Verbal Categories, and the Russian Verb” (1957) (cf. Whitt 2010: 6) and the sentence “Now that we’ve established that evidentiality is deictic […]” (Whitt 2010: 8) there is no line of argumentation for the deictic nature of evidentiality. Nevertheless it is ‘concluded’ that “evidentiality – whether expressed grammatically or lexically – is a deictic process whereby speakers encode the evidence for the statements (propositions) they make” (Whitt 2010: 11). The use of a perception verb is only then evidential if it contains “a deictic meaning in addition to the regular perceptive meaning” (Whitt 2010: 26). After offering the following examples (Whitt 2010: 26) (a) I see the house.

(b) Ich sehe das Haus.

(c) I see the house burning.

(d) Ich sehe das Haus brennen,

Whitt explains that the verb in (a) and (b) conveys that the speaker expresses the experience of seeing the house. In (c) and (d), in contrast, the speaker conveys, on the one hand, that the house is burning ([p]) and, on the other, that he sees the house burning. So, according to Whitt, the speaker’s indication that he has visual evidence for [p] represents a second proposition. “This second proposition […] contains the additional deictic component because it points to the speaker’s evidence and does not solely indicate an act of perception” (Whitt 2010: 26). At this point I wonder whether it could not also be argued, concerning (a) and (b), that ‘I see the house’ implies ‘I see the house being’ which, in turn, is – only linguistically – the same as ‘I see the house burning’, whereby here again the indication of visual evidence would represent the second proposition. Whitt interprets the burning of the house as the external factor, the additional component, the speaker somehow ‘points to’. But does the speaker not always point to an additional component, to an external factor, when he reports on seeing something – whether it is burning or not? It does not seem easy to follow Whitt’s line of argumentation why (c) and (d) contain a deictic use of see/sehen and (a) and (b) do not. The following situation and the corresponding dialogue may function as a counterproof: A and B are having a telephone conversation. A tells B: “C’s house is burning”. B: “How do you know that?” A: “I see the house”. Thus, to indicate that A has evidence for [p] (C’s house is burning) the answer “I see the house” instead of “I see the house burning” is sufficient and serves an evidential function. A sees the house – whatever happens to it. This seeing of the house serves itself an evidential function. Anyway, A should have gone to help C instead of phoning B. The chapters 2, 3 and 4 deal with verbs expressing visual perception, auditory perception and tactile perception, respectively. So the English-German verb pairs see – sehen, look – aussehen, hear – hören, feel – fühlen and the group sound – (sich) anhören/klingen are analysed. The olfactory and gustatory per-

169 ception verbs such as taste – schmecken and smell – riechen are only briefly dealt with in chapter 5. Regarding the question of polysemy, Whitt explains that the perception verbs’ polysemy “can indeed transfer over into the evidential domain, in that a related set of evidential meanings can be expressed by a single perception verb” (Whitt 2010: 25). As perception verbs which are used evidentially can occur in various complementation patterns (cf. Whitt 2010: 27), Whitt shows which verbs in which types of complementation pattern usually convey which verb meaning (cf. Whitt 2010: 219). The most common complementation pattern in general is perception verb + finite complement clause, whereby that, the complementiser, is not always present (cf. Whitt 2010: 27). See and sehen even have a variety of evidential meanings if occurring in the construction perception verb + finite complement clause, namely “direct perception, general observation, inference, knowledge, and understanding” (Whitt 2010: 220). The study has furthermore revealed that English and German are remarkably similar concerning their evidential use of perception verbs (cf. Whitt 2010: 219) and that “no new evidential meanings appear to have arisen within the last 500 or so years” (Whitt 2010: 222), that is, the verbs have not developed any ‘new’ evidential meanings since the Early Modern period. Even though it does not seem easy to follow the ‘deictic’ line of argumentation as it is employed in this study, Whitt’s analysis is especially interesting regarding the polysemy of the verbs and their diachronic development, or rather, their obvious diachronic non-development in the last 500 years.

3.12 Evidential expressions and their scope In his paper “Evidence for what? Evidentiality and scope” Boye (2010b) argues that evidence is only ‘for propositions’ so that the paper is worth being briefly but deeply dealt with here. The paper is thus concerned, above all, with the correct handling of certain notions, namely with the notions ‘proposition’, ‘state of affairs’ and ‘speech act’. It argues that evidential meanings have scope over neither states of affairs nor speech acts. By contrast, evidential meanings are said to “share scope properties in the sense that they are all conceptually dependent on a ‘proposition’” (Boye 2010b: 290). Hence, the definitions of the terms ‘evidentiality’, ‘scope’, ‘proposition’, ‘speech act’ and ‘state of affairs’ are crucial to follow his line of argumentation that evidence is for propositions. Evidentiality is defined as a cross-linguistic descriptive category, notionally defined in terms of ‘evidence’, which comprises grammatical as well as lexical means (Boye 2010b: 292, 295).

170 He uses the term ‘scope’ “in the specific sense of a meaning upon which evidential meaning is conceptually dependent” (Boye 2010b: 292), whereby he distinguishes between explicit and implicit scope. He offers the following two examples and explains that unfortunately has in both cases different explicit scopes but the same implicit scope: (a) There were three other guys on the train – they were thieves, unfortunately. (b) There were three other guys on the train – thieves, unfortunately.

While in (a) unfortunately has explicit scope over the whole clause they were thieves, the adverb in (b) “takes only the meaning of thieves as its explicit scope” (Boye 2010b: 292). However, the implicit scopes are highly similar as thieves alone implies that they were thieves so that the adverb’s implicit scopes are similar. Speech acts are defined as the linguistic meaning units that are “potentially semantic, conventional meanings” (Boye 2010b: 293). In order to define the notions ‘state of affairs’ and ‘proposition’ he illustratively opposes them to each other because “[a]mong the scholars that associate evidentiality and evidential meaning with propositions, the majority probably do not distinguish propositions from states of affairs” (Boye 2010b: 299). Throughout the present study ‘state of affairs’ and ‘proposition’ are used synonymously but I may also speak of ‘situation’ or ‘event’, even though I am aware of the notional distinction. Clearly, when speaking about a state of affairs it is referred to it in the sense of being the information that is represented in the form of a proposition. To my knowledge, the English language does not provide a synonym for ‘proposition’. That is why one sometimes has to – more or less successfully – sidestep. Capelli’s study is another example where ‘state of affairs’ and ‘proposition’ are used synonymously. Cappelli (2007: 121-123) speaks about evaluating a ‘state of affairs’, but she cleverly explains: A state of affairs is traditionally defined as an abstract construct, a set of objects related to one another, which, in order to be linguistically expressed must take a propositional form, which is represented as p (Cappelli 2007: 122-123).

Thus, she relates a ‘state of affairs’ with a ‘proposition’ explaining that the proposition (only) is the linguistic form by which a state of affairs can be represented. Therefore, the following part of a sentence, found in Cappelli’s study, is considered sparsely consequent: “If we consider the speaker to be the evaluator of a state of affairs, or of the propositional content p of a sentence […]” (Cappelli 2007: 129). Nevertheless, the relationship between the two notions is well illustrated so that a synonymous treatment of both terms may be considered legitimate and should be tolerated because of the lack of synonyms. Anyway, Boye explains that

171 states of affairs are pure conceptual representations whereas propositions constitute information about the world in the sense that they are conceptual representations construed as referring – i.e. as having as external referent […] (Boye 2010b: 293). Since propositions, but not states of affairs, constitute information about the external world in the sense that they have a truth value or refer, only propositions can be evaluated epistemically [or marked evidentially] (Boye 2010b: 293).

In the introductory part of his paper, Boye (2010b: 290-291) differentiates between three types of linguists: one group, which he himself belongs to, takes evidentiality to apply to propositions. The second group comprises linguists such as Anderson (1986), Dendale/Tasmowski (2001) or Cornillie (2009), who say that evidential meanings depend on speech acts or related notions like ‘claims’, ‘statements’ or ‘assertions’. The third group – and here Nuyts (2006) is representatively cited – takes evidential meanings to apply to ‘states of affairs’. Before providing arguments in support of the view that evidential expressions have scope over propositions, Boye argues against the assumption that the scope of evidential meanings is a speech act or a state of affairs. One argument against the view that evidential expressions have scope over speech acts is that “evidential expressions are incompatible with genuine imperatives” (Boye 2010b: 296; cf. Aikhenvald 2004: 250). The assumption that evidential meanings have scope over assertions can also be refuted because Aikhenvald (2004) observes that evidentials are also found in questions (Boye 2010b: 296). So the conjecture that evidential meanings have speech acts as their scope has to be rejected: Speech acts […] are not something for which there can be evidence. We can say that we base our assertion of something on some evidence, but this does not mean that the evidence at hand applies to the assertion we make. Clearly, the evidence applies to what we assert, not our assertion of it (Boye 2010b: 296-297; emphasis in the original).

If allegedly in the following example took the speech act as its semantic scope, “[t]his would imply that somebody told her that she made an assertion” (Boye 2010b: 297): Anja allegedly ran away yesterday (cf. Boye 2010b: 297). So the adverb cannot be said to convey that the speaker has reportive evidence for the assertion. Instead, allegedly conveys that the speaker has evidence for what he asserts. The hypothesis that evidential meanings have states of affairs as their scope should be equally rejected because [s]tates of affairs are not something for which there can be evidence. They are not about the world in the sense of having a truth value or in the sense of referring.

172 Rather, in the traditional denotational conception of them, they are in the world […] (Boye 2010b: 295; emphasis in the original).

As it is one aim of Boye’s paper to demonstrate that evidential scope properties can help to draw the line between what is evidential and what is not, Boye lists five arguments in favour of the hypothesis that evidential meanings can exclusively have scope over propositions and that certain scope criteria demonstrate which linguistic means belong to the category of evidentiality and which do not (cf. Boye 2010b: 291): 1. There can only be evidence for propositions (as opposed to states of affairs and speech acts), whereby “there may not even be a whole clause in the semantic scope of an evidential expression” (Boye 2010b: 301). It is also possible for an adverb, for instance, only to have scope over a noun phrase as in the example natives of Spain: When passing in front of the police barracks the coach was hailed by two well-dressed men, apparently natives of Spain, who asked the driver…. It is furthermore possible that an adverb such as apparently has no explicit scope. This is the case when it alone represents an answer: A – Is Anja still writing? B – Apparently! The implicit scope would be the affirmative version of the question. 2. Uncontroversial examples of evidential expressions are compatible with propositional clauses such as declaratives (e.g. Anja is evidently still writing) and English finite complement clauses (He saw that Anja is evidently still writing). 3. Evidential expressions disambiguate ambiguous clauses as propositional. For instance, in written Danish, one and the same verbal form may be read either as an imperative expressing a command or as a declarative expressing an assertion. In the first case a state of affairs would be involved, in the second a proposition. Since only “propositions are what evidential meaning applies to” (Boye 2010b: 299), the presence of an evidential adverb disambiguates the two possible readings with the result that the clause must be a declarative expressing an assertion. 4. Propositional clauses disambiguate ambiguous expressions as evidential. The expression I hear, for example, has the meaning of ‘act of auditory perception’ when used in a construction “together with a state-of-affairsdesignating accusative-with-infinitive without infinitival marker” (Boye 2010b: 299) as in I heard her type. It has an evidential meaning, in contrast, “in a construction with a proposition-designating finite complement clause” (Boye 2010b: 300): I heard that she was typing. 5. The last argument is concerned with “diagrammatic iconicity and with the link between deictic tense and propositions” (Boye 2010b: 300). There are

173 languages in which evidential markers and tense markers occupy different positions in a clause. If so, the evidential marker usually occurs outside the tense marker regarding the predicational core. So this ordering of evidential markers with respect to tense markers, Boye argues, can be seen “to reflect the fact that the scope of evidential meaning is a proposition” (Boye 2010b: 300). Boye conclusively proposes a definition of evidentiality which includes scope criteria: evidentiality covers meanings that represent the evidence for a proposition. Evidential meanings can have the meaning of a proposition-designating clause as their explicit scope and cannot have the meaning of a state-ofaffairs-designating clause as their explicit scope (Boye 2010b: 304). In sum, Boye’s paper illustrates how to handle which notion.

3.13 Evidentiality, epistemic modality and grammaticalisation in Italian We have seen in chapter 2.2 how Pietrandrea (2005) defines epistemic modality. To have a point of departure for the discussion of her study about this functional category in the Italian language system, it is useful again to briefly summarise her definition. Modality is seen as highly subjective, whereby subjectivity is to be understood in all of its three senses: subjectivity in the sense of ‘performativity’, ‘meta-propositionality’ and ‘genuine epistemicity’ (cf. Pietrandrea 2005: 36). That leads her to include all these aspects in her definition of modality so that Pietrandrea has worked out the following definition: modality is the performative category expressing the speaker’s genuine opinion towards the modalised proposition (Pietrandrea 2005: 39). Her definition does not involve any words about the nature of the linguistic means that (may) express epistemic modality. Pietrandrea, by all means, rejects Lazard’s strict definition of a grammatical category (2001: 360). According to him, a grammatical category is only then grammatical if there are specific grammatical forms whose semantic-pragmatic content is that of the category’s meaning, i.e. ‘source of information’ for evidentiality or ‘speaker’s attitude’ for epistemic modality (cf. Pietrandrea 2005: 53, cf. Lazard 2001: 360). Pietrandrea explains that accepting this strict definition or status of a grammatical category “would lead to the paradoxical conclusion that a grammatical category of epistemic modality is almost non-existent” as languages that express epistemic modality via specific forms are rare rather than usual (Pietrandrea 2005: 53). Pietrandrea instead shows that although in Italian there are no grammatical epistemic forms strictu sensu […], if one is willing to take into account forms with a different central meaning and uses grammaticality scales […], it is possible to recognize, among the different forms ex-

174 pressing epistemic modality, some that have more grammaticality than others […] (Pietrandrea 2005: 54).

To these forms that have a different central meaning and ‘a second modal’ one belong the modal verbs dovere ‘must’ and potere ‘can’; the epistemic use of the synthetic future form; epistemic adjectives such as possibile ‘possible’ and probabile ‘probable’; the corresponding epistemic adjectival constructions like essere probabile ‘to be probable’; epistemic sentence adverbs like sicuramente ‘surely’, indubbiamente ‘undoubtedly’, certamente ‘certainly’ or forse ‘perhaps’; epistemic verbs like credere ‘believe’, suppore ‘suppose’, dedurre ‘deduce’, pensare ‘think’ or sembrare ‘seem’; epistemic syntagmatic constructions such as mi sa ‘(lit.) it is known to me’ or mi pare ‘(lit.) it appears to me’; and also phonetic lengthening which marks hesitation or other particular prosodic profiles (cf. Pietrandrea 2005: 54-61). She furthermore lists ‘epistemic utterances’. It does not seem really intersubjectively comprehensible what this means: Pietrandrea provides one example for an ‘epistemic utterance’: the Italian original has the English translation ‘I feel very strongly that hunting is a barbarity that should be abolished’ (Pietrandrea 2005: 61). It is possible to detect the epistemic component, namely the phrase ‘I feel very strongly that…’, but it is not easy to comprehend why this is called to be an epistemic utterance and an utterance containing, for instance, ‘It is probable that…’ is listed below ‘epistemic adjectival construction’. Is not an utterance that contains such a construction also an ‘epistemic utterance’ – regardless of whether it contains an adjectival construction as well? Or is everything which does not fit into one of the categories listed above regarded as an ‘epistemic utterance’? Unfortunately, there is no further explanation as to what the notion of epistemic utterance is designated for. The expression of epistemic modality via these forms listed above is said to be “very little dispersed” in spontaneous speech and considered strongly context-sensitive since the use of epistemic markers is not obligatory in Italian (Pietrandrea 2005: 3). The fact that epistemic markers usually occur in argumentative contexts was shown by the heterogeneous data Pietrandrea analysed: data of spoken Italian, from press, personal e-mails, scientific and electronic publications, extemporary utterances and intuition data (cf. Pietrandrea 2005: 4). Some of these forms are considered more grammaticalised than others. Here, the notion of ‘grammaticality scales’ comes in. Applying Lehmann’s six criteria of grammaticality93 and leaving aside the question of grammaticality of

93

In her study Pietrandrea refers to Lehmann’s six criteria of grammaticality from 1985 (in: “Grammaticalization: Synchronic Variation and Diachronic Change”. In: Lingua e

175 the prosodic forms – as it would require another investigational background – Pietrandrea (2005: 64) identifies the following “undoubtedly non-grammaticalized” forms: epistemic utterances (whatever this exactly is); epistemic syntagmatic constructions; epistemic adjectival constructions; epistemic verbs, sentence adverbs and adjective modifiers of predicative nouns. But regarding the epistemic syntagmatic constructions she distinguishes between grammaticalised forms and non-grammaticalised forms. The latter is, for example, represented by secondo me ‘according to me’ or mi pare ‘(lit.) it appears to me’. Relying on the criterion of integrity, grammaticalised forms, i.e. forms that “have lost their etymological meaning and developed a generic epistemic meaning” (Pietrandrea 2005: 64), are capace che ‘(lit.) it is capable that’ or si vede che ‘(lit.) one sees that’. Secondo me and mi pare, in contrast, have not lost their etymological meaning. According to Pietrandrea (2005: 66), modal verbs and the epistemic future clearly belong to the grammaticalised forms. Pietrandrea’s grammaticality scale consequently looks as follows:

Stile 20 (3), 303-319). They are slightly different to those proposed by Lehmann in 2002, where he explains: The weight of a sign, viewed paradigmatically, is its integrity, its substantial size, both on the semantic and phonological sides. Viewed syntagmatically, it is its structural scope, that is, the extent of the construction which it enters or helps to form. The cohesion of a sign with other signs in a paradigm will be called its paradigmaticity, that is, the degree to which it enters a paradigm, is integrated into it and dependent on it. The cohesion of a sign with other signs in a syntagm will be called its bondedness; this is the degree to which it depends on, or attaches to, such other signs. The paradigmatic variability of a sign is the possibility of using other signs in its stead or of omitting it altogether. The syntagmatic variability of a sign is the possibility of shifting it around in its construction (Lehmann 2002: 110). So Lehmann’s six criteria or parameters of grammaticalisation can be arranged according to three aspects of grammaticalisation: decrease in weight and variability; increase in cohesion (cf. Lehmann 2002: 109, 110), whereby the three aspects are divided into the criteria of selection and combination of linguistic signs, that is, into paradigmatic and syntagmatic criteria (cf. Lehmann 2002: 110):

WEIGHT COHESION VARIABILITY

PARADIGMATIC Integrity Paradigmaticity Paradigmatic Variability

SYNTAGMATIC Structural Scope Bondedness Syntagmatic Variability

176 Grammaticality

+ grammatical

– grammatical

Epistemic Forms

utterances verbs adjectives sentence adverbs adjectival constructions syntagmatic constructions

Criteria of Grammaticality

si vede che capace che

modal verbs epistemic future

(integrity)

(integrity) (scope) (fusion) (integration) (mobility)

Fig. 8: Pietrandrea’s grammaticality scale (cf. Pietrandrea 2005: 67; slightly adapted)

The conditional forms of the modal verbs dovere and potere are also said to belong to the more grammaticalised forms (Pietrandrea 2005: 67) that express epistemic modality. So she explains that a similar proposition like Anja è a casa ‘Anja is at home’ may be epistemically modalised by the following five forms (cf. Pietrandrea 2005: 69):

Anja

deve

must

dovrebbe

should

può

essere a casa.

‘Anja may

potrebbe

could

sarà

will

be at home.’

It is a little bit hard to follow why Pietrandrea analyses ‘contextless’ sentences and it is consequently hard to track why the forms above are analysed as being modal and not evidential, or why it is not argued – as in the introductory part – that their use is context-sensitive.94 Instead, dovere is said to convey a strong judgement, that is, it expresses epistemic necessity, while potere is said to convey a weak judgement and consequently expresses epistemic possibility

94

In a later chapter (4) of Pietrandrea’s study, however, the evidential nature of the modals is dealt with (cf. Pietrandrea 2005: 80-99).

177 (Pietrandrea 2005: 70). So her next step is to investigate whether the four modal verb forms can be disposed vertically along the same certainty scale, in other words, whether, for example, [dovrebbe] expresses a weaker judgement compared to [deve] or a stronger judgement compared to [potrebbe] and [può] (Pietrandrea 2005: 72-73).

To investigate this, it is necessary to find out how the conditional morphemes influence the semantics of the modals: dovrebbe is shown to weaken the degree of certainty which is conveyed by the indicative form (cf. Pietrandrea 2005: 74). By opposing deve and dovrebbe, Pietrandrea furthermore concludes that the former is used to express an inference based on objective knowledge95, while the latter is used to express an inference based on uncertain knowledge. Consequently, the indicative form conveys unconditioned necessity and the conditional form a conditioned one (cf. Pietrandrea 2005: 74-76). So the conditional morpheme indicates “that the truth of the modal that it modifies holds in a world where certain conditions are met. That increases the degree of uncertainty of the modal only indirectly” (Pietrandrea 2005: 76). The findings of dovrebbe cannot analogously be applied to potrebbe: the conditional form of potere is analysed as being more epistemic than the indicative form which is considered more deontic (cf. Pietrandrea 2005: 80). Hence, and as will be shown later, the four forms cannot simply be arranged along a certainty scale; a further dimension – the deontic-epistemic axis – has to be included. Chapter 4, including its subchapters, deals with the evidential nature of the modals. Here, she even emphasises that “all the Italian modal forms have a central evidential meaning [!]” (Pietrandrea 2005: 85) and for deve, in particular, she explains: 95

Pietrandrea does not clearly define what she means by ‘objective knowledge’. But the example she poses could lead to the conclusion that ‘objective’ could here be paraphrased as ‘logical’. She introduces the example where she opposes dovrebbe and deve via the description of the following situation: Let us imagine that we are in the middle of a crowd waiting for Clint Eastwood to come out from his hotel and that Eastwood’s exit follows the exit of other stars. If we have objective knowledge allowing us to infer the certainty of Eastwood’s exit (for example, all the other stars have already come out and Eastwood is the last one expected, or according to a strict schedule, Eastwood must come out right now) we can only legitimately say: [Ora deve uscire Clint Eastwood and not Ora dovrebbe uscire Clint Eastwood] (Pietrandrea 2005: 74).

178 […] the reason why [deve] expresses a strong degree judgment is to be sought in the fact that this form indicates that the speaker has objective knowledge, from which (s)he can infer the certainty of the conclusion introduced by [deve] (Pietrandrea 2005: 80).

So deve may not only be used to modalise but also to ‘evidentialise’ a proposition, that is, it shows a variety of meanings, whereby deve is in fact said to be used “much more as an evidential marker, than as a genuine epistemic marker” (Pietrandrea 2005: 85). Dovrebbe may also convey not only a ‘conditioned inferential’ reading but also a reportive one (cf. Pietrandrea 2005: 87). Even though most uses of può are epistemic, and not evidential, in chapter 4.2 this form is also conceded to be used with a clear evidential meaning (cf. Pietrandrea 2005: 90). Potrebbe finally “both the evidential meaning of one of the possible conclusions of an inferential process and a reportive evidential meaning” are applied (Pietrandrea 2005: 99). Therefore, the modal verbs should not be represented as ‘so modal’ at the beginning of her study. It is said that the future – unlike the modals – “seems not to condense any inferential process, but appears to function as the form expressing the speaker’s genuine opinion about the propositional content” (Pietrandrea 2005: 93). It is indeed denied that the future conveys an inferential reading, but is said to express the speaker’s subjective opinion, whereby it is regarded as neutral as to the degree of certainty (cf. Pietrandrea 2005: 94, 98). I wonder whether the Italian future can really be said to be neutral regarding the degree of certainty that is conveyed. It is supposed that by speaking of the ‘future’s neutrality’ Pietrandrea refers to its semantics, that is, the future form itself does not convey a higher or lower degree of certainty in comparison to dovere and potere, for instance. The degree of certainty is generally estimated higher if an inference is marked by dovere in comparison to an inference marked by potere. Such an assertion cannot be made with regard to the future. But – with regard to the Spanish future – contextual information may provide hints regarding the speaker’s degree of certainty that the future in a particular context conveys (cf. chapter 7.2). Furthermore, it should not be stated that the future has “no evidential nuances” (Pietrandrea 2005: 101) because similar uses in similar contexts could certainly also be found in Italian: (50) En el día que cumple 74 años, Berlusconi ha admitido que la “dialéctica interna de la mayoría ha superado todos los límites”, ha reclamado un escudo judicial para los altos cargos del Estado y ha ofrecido un obligado armisticio, que será sin duda fugaz, a los 35 diputados rebeldes finianos, purgados del partido el pasado 29 de julio, sabedor de que su voto es crucial para ganar tiempo hasta que los sondeos le sean más favorables. (El País 01/10/2010)

179 Example (50) clearly indicates that sin duda ‘no doubt’ represents contextual information that shows that in this text passage the degree of certainty conveyed by será cannot be regarded as ‘neutral’. If [p] will undoubtedly be the case, the degree of certainty is clearly higher than only neutral. (51) No bstante, el piloto de Yamaha lo tiene claro. “El objetivo vuelve a ser el podio, para así seguir sumando el máximo de puntos en la carrera por el campeonato del mundo. Será importante ir rápidos desde el viernes y así poder iniciar de la mejor manera posible las tres carreras consecutivas que ahora tenemos por delante”. A Pedrosa, por su parte, tampoco se le da mal el trazado nipón. (El País 01/10/2010) (52) [...] Ignacio Fernández Toxo, se ha mostrado “absolutamente convencido” del éxito de la huelga y aunque “no probablemente al día siguiente”, sus resultados se verán “en un espacio corto de tiempo”. “El éxito de la huelga será el cambio de las políticas que están haciendo tanto daño”, ha indicado. (El País 30/09/2010)

Text passages (51) and (52), in contrast, represent examples where indeed no contextual information is provided that would allow any conclusion to be drawn about the degree of certainty conveyed by the forms. (53) “[...] Lo que más me preocupa es responder y los rivales”. Entre ellos, el belga Philippe Gilbert, que “seguramente atacará y será uno de los peligrosos”, apunta el español. Con escaso rodaje pero “con la misma ilusión que en su primer Mundial”, el cántabro cree que el trazado de Melbourne se adapta bien a sus características. (El País 01/10/2010)

Será and atacará in example (53) are clearly used evidentially: they indicate that [p] is an inference, which is underlined by seguramente – a modal adverb that is also often used to mark a proposition evidentially (cf. chapter 5.1.6). The speaker is fairly certain that [p] will be/must be the case. Pietrandrea (2005: 98) arranges the four modal verb forms and the future along a certainty scale (first axis) and distinguishes them along the deonticepistemic axis (second axis). “The third axis is represented by the opposition between evidential and genuine epistemic meaning” (Pietrandrea 2005: 98) because from the fourth chapter onwards of Pietrandrea’s study it is clear that the opposition between the evidential and the genuine epistemic meaning of the forms should be considered as well. The following scheme represents the internal configuration of the epistemic domain in Italian according to Pietrandrea (2005: 98):

180 + deontic

+ epistemic epistemic future

+ certain

deve dovrebbe può

Ø neutral

potrebbe + evidential

+ epistemic

- certain

Fig. 9: The epistemic domain in Italian (Pietrandrea 2005: 98; scheme is slightly adapted)

Obviously, her epistemic domain includes the evidential one. So Pietrandrea seems to assume a relationship of inclusion concerning the linguistic categories of evidentiality and epistemic modality. Pietrandrea’s study of epistemic modality in Italian reveals, above all, interesting insights about the conditional forms of the modal verbs. It should also be analysed whether Spanish debe and debería behave similarly compared to deve and dovrebbe (cf. chapter 6.1). For dovrebbe it is said that it marks the fact that the speaker has an uncertain (conditioned) control over the knowledge that leads him to conclude the truth of the modalized [or ‘evidentialised’] proposition. This proposition, due to the uncertainty of the premises, is put forward as the tentative conclusion of a deductive process (Pietrandrea 2005: 126).

The two following examples may indeed be interpreted as tentative conclusions. Example (54) additionally contains en mi opinión ‘in my opinion’, which underlines that the speaker does not have unconditioned control over the knowledge: (54) En mi opinión la sensibilación debería ir enfocada al hombre. Las campañas de prevención se centran en la mujer para que denuncie al agresor, pero nunca en el agresor diciéndole que lo que hace es maltratar y que está mal. (El País 04/08/2010) (55) Jed Mercurio, médico residente tras su paso por la Royal Air Force y ahora escritor, apuesta por la obsesión por el sexo para explicar su perfil de mujeriego: “El término de adicción es controvertido. Debería hablarse de obsesión, frustración, ya que la adicción al sexo no tiene repercusiones físicas. Puede provocar depresión, pero no dolores de cabeza”. (El País 18/08/2010)

The use of debería in example (55) could also be interpreted as being pragmatically motivated: it seems to be accompanied by a moment of courtesy. The speaker explains why he rejects the use of the word adicción ‘addiction’ and

181 courteously proposes alternative terms. Nevertheless, he also justifies why he argues for the use of alternative terms like obsesión or frustración: ya que ‘because’ introduces the justification for ‘debería [p]’. As already mentioned, Pietrandrea concedes dovrebbe – besides its ‘conditioned inferential’ meaning – a reportive reading (cf. Pietrandrea 2005: 87). The Spanish equivalent (here with the past tense) may also be used to convey a reportive reading: (56) El punto en que Obama se bañó está, sin embargo, a 398 kilómetros de la costa frente a la que se hundió la plataforma de BP el pasado mes de abril. Según sus propias estimaciones, BP debería haber sellado el pozo definitivamente, a través de un conducto paralelo al de la fuga, este mismo fin de semana. (El País 19/08/2010)

However, the use of debería or dovrebbe to mark reportive information is less due to the semantic potential of the modal verbs but more to the conditional morpheme as such, as Pietrandrea also remarks: […] mood can also mark evidentiality. This is so in the case of the Italian or French ‘reportive’ conditional and the German subjunctive, which are employed to indicate that the speaker brings what has been said by someone else as evidence for the truth of his/her proposition (Pietrandrea 2005: 18).

It should be challenged whether the conditional form is always used by the speaker to bring evidence for the truth of his proposition. A speaker may also choose this form of quotation when he himself does not believe in the truth of [p]. In Spanish, this form of indirect quotation is frequently found in newspaper articles and very often accompanied by an introductory phrase beginning with según ‘according to’: (57) Según los analistas, sería difícil llevar a cabo algunas partes de la ley de la reforma tales como la aplicación de la cobertura a aquellos que no cuentan con seguro médico. (El Mundo 15/12/2010)

In chapter 7.1 this form of quotation is dealt with in more detail. In any case, Pietrandrea’s study represents one example of how to treat modal/evidential forms in a Romance language, and her findings on the use of the conditional form, especially on dovrebbe, will be relied on as theoretical background for the present study (cf. chapter 6.1).96

96

For reasons of space, I only focussed on Pietrandrea’s analysis of the use of the modal verbs and the epistemic future. Furthermore, her remarks on the use of the modals and the future will be revisited in the context of the present study.

182

3.14 Evidentiality, modality, epistemicity and grammaticalisation in German A further study of modality, which clearly should be listed here among the recent studies, is the collected volume on modality in German edited by Abraham/Leiss (2009). As the papers by Leiss (2009) and Maché (2009) are dealt with or referred to in other chapters of the present study, only the papers by Diewald (2009) and Mortelmans (2009) will briefly be summarised. Diewald’s paper is not explicitly concerned with the epistemic and/or evidential nature of the German modals but with their grammaticalisation process they have undergone. Diewald (2009 [2006]97) shows that the stages of development of new grammatical functions of a particular (former lexical) item are closely tied to particular contexts or rather constructions. Different context types such as critical context and isolating context are described to be involved in grammaticalisation processes. These contexts, in turn, are seen – within the framework of Construction Grammar – as specific types of idiomatic constructions (cf. http://www.constructions-online.de/articles/specvol1/686; cf. also Diewald 2009: 101). Diewald develops a three-stage model of context types in grammaticalisation. Her model “distinguishes three chronologically ordered stages in the diachronic rise of grammatical functions, each of them associated with a particular type of context” (http://www.constructions-online.de/articles/specvol1/686): Table 9: The diachronic rise of grammatical functions (http://www.constructionsonline.de/articles/specvol1/686; cf. also Diewald 2009: 109) I II III

STAGE Preconditions of grammaticalization Triggering of grammaticalization Reorganization and differentiation

CONTEXT TYPE Untypical context Critical context Isolating context

MEANING/FUNCTION Conversational implicature Multiple opacity Polysemous/heterosemous

The first stage is characterised by the development of the preconditions for a further grammaticalisation process because certain lexical items begin to be used in untypical contexts. The critical context corresponds to a highly marked construction. This new construction is semantically opaque as it can be interpreted differently. One of the meanings is the new grammatical one. The third stage is

97

Her paper “Die Interdependenzen von Kontexttypen bei Grammatikalisierungsprozessen illustriert am Beispiel der deutschen Modalverben” was published in 2009 but it relies on an earlier version from 2006. See the web page: http://www.constructions-online.de/articles/specvol1/686.

183 characterised by the restructuring of the semiotic correlation between form and meaning (cf. Diewald 2009: 107): Stage three shows the consolidation of the grammaticalization process, i.e. the re-organization and differentiation of the grammatical formatives and the paradigm that is the target category of the ongoing grammaticalization process. In this phase, the new grammatical meaning is isolated as a separate meaning from the older, more lexical, meaning (http://www.constructions-online.de/articles/specvol1/686; cf. also Diewald 2009: 107-108).

Diewald then applies the different stages of the grammaticalisation process to the development of the six German modals dürfen ‘to be allowed to’, können ‘can/to be able to’, mögen ‘to like/may’, müssen ‘must/to have to’, sollen ‘shall/to be to’ and wollen ‘to want’ (cf. Diewald 2009: 110). It is shown that the critical context developed as the second stage in the Middle High German period, in which both a lexical (non-deictic) reading as well as a grammaticalised (deictic) reading of the modal verbs was possible (cf. Diewald 2009: 117). The isolating contexts developed later, namely in the Early New High German period (1500-1650) (http://www.constructions-online.de/articles/specvol1/686). This implies that if one were only looking back 500 years, both readings of the German modals – the deictic and the non-deictic one – would have already emerged. Whitt (2010), studying the diachronic development of perception verbs, albeit not exhaustively, only looked back 500 years (Whitt 2010: 222). Perception verbs in untypical contexts and later critical contexts could maybe have been found if one had looked back more than 500 years. Diewald (2009: 119) does not describe the third stage as the end of the grammaticalisation process because the stage of reorganisation and differentiation does not yet consider features such as paradigmatisation and information about when the new (grammaticalised) element should be obligatorily used. So she proposes to add a fourth stage to the model of context types: Table 10: Stage IV of the diachronic rise of grammatical functions (cf. Diewald 2009: 119) STAGE IV

Paradigmatic integration

CONTEXT TYPE Paradigmatic context

MEANING/FUNCTION Paradigmatic oppositions within existing or newly reorganising paradigms

In summary, Diewald has shown how to arrange grammaticalisation theory and Construction Grammar by developing a complex model to capture the grammaticalisation process of the German modals, whereby this model could be possibly applied to other items affected by linguistic change. Her study has furthermore revealed that, if working diachronically, it is possibly not enough to look back (only) 500 years as the non-deictic, besides the deictic, reading of the

184 German modals, for instance, was already developed in the Middle High German period. In her study “Erscheinungsformen der indirekten Rede im Niederländischen und Deutschen: zou-, soll(te)- und der Konjunktiv I” Mortelmans (2009) is concerned with the question of how German and Dutch indirectly indicate reportive information. She thereby concentrates on the analysis of the German modal soll(te) and the Dutch ‘almost-equivalent’ zou/zouden. Mortelmans furthermore investigates the differences between the use of soll(te) and the German subjunctive. Mortelmans’ study is guided by the thesis that evidential soll- refers more strongly to the current (reporting) speaker than zou- and the German subjunctive, which are said to refer more strongly to the perspective of the quoted speaker. So soll-, on the one hand, and zou-, on the other, express different ‘speaker-alignments’98 (cf. Mortelmans 2009: 171). Zou- is termed a ‘non-present form’ of the Dutch modal zullen, which combines two functions (cf. Mortelmans 2009: 171): as it is non-present it refers to an alternative reference point that is distinct (or ‘away’) from the current speaker, that is, it can also occur in past contexts (cf. Mortelmans 2009: 179). This function represents the evidential-quotative reading (cf. Mortelmans 2009: 180): (a) Stef Kamil Carlens: kwam niet opdagen, zou ziek zijn. ‘Stef Kamil Carlens zeigte zich [sic] nicht, er sei krank, hieß es.’ [‘Stef Kamil Carlens did not show up. He was sick, they said.’]

The other function of zou- is to indicate reference to non-realised conditions, that is, to non-factuality. Thus, it may also indicate a counter-factual reading as in the following example (cf. Mortelmans 2009: 172): (b) Het zou een geode film zijn geweest (als de verfilming zou zijn doorgegaan). ‘Es wäre ein guter Film gewesen (wenn die Verfilmung erfolgt wäre).’ [‘It would have been a good film if it were made into a film.’]

In my opinion, the speaker does not only refer to an if-situation, that is, to a nonrealised condition, but also conveys a supposition because the speaker obviously supposes that the film would have been a good one (cf. also Pelyvás 1996). If more context – or context at all – were provided it might possibly also be interpreted as conveying an inferential reading. There must have clearly been certain circumstances, certain experiences that ‘forced’ the speaker to utter [p]. To allege an example: the speaker has already read more books by the same author 98

In German, Mortelmans (2009: 171) speaks about “Sprecherorientierung”.

185 that were made into films and these were always good ones. These experiences – or these experiences stored as the speaker’s knowledge – could lead the speaker to utter inference (b). Zou- is said to combine two functions, while soll- fulfils, according to the author, only one function (cf. Mortelmans 2009: 171-172). This is in slight contrast to Leiss’ findings (2008, 2009), who argues that the German modals are double-deictic or complex-deictic because of the fact that they combine the features of speaker’s attitude and information source (cf. chapter 2.1.2.1). However, Mortelmans’ scheme (2009: 186; see below), in turn, lets me suggest that she also assumes soll- to convey evidentiality as well as epistemic modality. Comparing the different fields of use of the quotative subjunctive and evidential soll- in German, Mortelmans explains that the current speaker may coincide with the reporting speaker. The current speaker consequently overlaps with the information source. In those cases soll- cannot be used but the German subjunctive (Konjunktiv I): (c) Ich sagte, dass er wahrscheinlich recht habe. (d) *Ich sagte, dass er wahrscheinlich recht haben soll (Mortelmans 2009: 177).

In its evidential-quotative reading, zou- is more similar to the use of the German subjunctive than to the use of soll(te) because zou- is more aligned to the perspective of the quoted speaker (cf. Mortelmans 2009: 185). Nevertheless, zou- cannot be (totally) equated with the use of the Konjunktiv I: zou- is only to be found in matrix clauses and it may also refer to a ‘judging’ speaker, which may influence the evidential-quotative reading (cf. Mortelmans 2009: 186). That is why zou- is described as an intermediate stage between German soll- and the German subjunctive: soll-

reporting speaker epistemic / evidential

zou-

Konjunktiv I

quoted speaker quotative

Fig. 10: The intermediate stage of zou- between German soll- and the German subjunctive (cf. Mortelmans 2009: 186; slightly adapted)

186 Mortelmans’ paper is an interesting contribution to the research field of quotation, especially indirect quotation because it is concerned with the comparison of the different functions of the German subjunctive, the German modal sollen and Dutch zou and because it deals with the question of how these forms are used to indicate reported information. Cross-linguistically, zou- and the German forms are compared; intralinguistically, the quotative use of the German modal and the Konjunktiv I are compared.99

3.15 Evidentiality, modality and subjectivity In 2005, Klinge and Müller published the collected volume Modality. Studies in Form and Function, which contains various papers that are of interest for the present study. In the following, the papers by Nuyts (2005) and Boye (2005) will be reviewed because both linguists argue for “conceptual clarification of the notions involved under the general heading of modality” (Klinge/Müller 2005: 2). The paper by Heltoft (2005) will also be reviewed because it deals with the relation between modality and subjectivity. In “The modal confusion: on terminology and the concepts behind it” Nuyts (2005) initially explains that he takes “a no doubt highly controversial position” (Nuyts 2005: 5): […] the current notion of ‘modality’ is not a very fortunate one, for basic conceptual reasons. First of all, it should not be considered a notion at the same level of analysis as tense (or time marking more generally) and aspect: it rather constitutes a ‘higher order category’ (Nuyts 2005: 5).

I always thought that ‘modality’ in any case represents a somehow superordinate category for epistemic modality, deontic modality etc. and it seems difficult to 99

Helin’s study (2004) – just as the paper by Mortelmans (2009) – is also concerned with the quotative use of the German subjunctive (Konjunktiv I). Her study …So der Wetterbericht. Evidentialität und Redewiedergabe in deutschen und finnischen Medientexten und Übersetzungen has a contrastive character since it is also concerned with the Finnish linguistic devices that are used to translate the German subjunctive. As the title of her paper shows, the author concentrates on the analysis of modern media texts, that is, of a corpus consisting of newspaper articles and magazines, and on their translations (cf. Helin 2004: 13-14; 67-71). She shows that in Finnish the indicative pluperfect tense represents a common stylistic device to translate the German subjunctive if used to convey a hearsay reading. This is not surprising as the Finnish pluperfect has always been related with the notion of hearsay (Helin 2004: 96). Nevertheless, this Finnish tense has never been described in terms of evidentiality (cf. Helin 2004: 15, 237) so that Helin’s contrastive study of German and Finnish media texts is important insofar as it shows that the Finnish language also provides linguistic means that could or should be analysed in terms of evidentiality.

187 see the point where the ‘modal confusion’ is or where it suddenly comes from. Will there at some point be an ‘evidential confusion’ because of all the evidential subcategories? The fact that Nuyts proposes to abandon the notion of modality as such – even though I do not really see the necessity – and “to replace it by the label ‘attitude’” (Nuyts 2005: 28) can be explained by the fact that all kinds of modality express a speaker-oriented or subject-oriented attitude (cf. Nuyts 2005: 27).100 But what is not that easy to explain is that he lists among the – now called – ‘attitudinal’ – instead of ‘modal’ – categories evidentiality, even though he previously clearly distinguished between epistemic modality and evidentiality (cf. Nuyts 2001b). In this paper, the headline ‘Epistemic modality’ is complemented by (an unemotional) “plus or minus evidentiality” (Nuyts 2005: 10). Nevertheless, he then distinguishes between the two categories by definition. Epistemic modality is said to concern an indication of the estimation, […] typically but not necessarily by the speaker, of the chances that the state of affairs expressed in the clause applies in the world or not, or, in other words, of the degree of probability of the state of affairs […] (Nuyts 2005: 10). […] epistemic modality involves an explication of the degree of ‘existential’ commitment of the speaker to the state of affairs, i.e. the extent to which (s)he believes the state of affairs has been or will be realised in the ‘real world’ (Nuyts 2005: 23).

Evidentiality, in contrast, is said to indicate the reasons “to assume or accept the existence of the state of affairs expressed by the clause” (Nuyts 2005: 11). Thus, he argues for a clear-cut distinction between epistemic modality and evidentiality, although evidence represents the basis for epistemic judgements and evidentials are often concerned with certain epistemic judgements (cf. Nuyts 2005: 12). But he then explains that, in his view, “the question whether evidentiality should be included among the modal notions” is still left open (Nuyts 2005: 12). Although he does not explain further why this question is still to be answered – he distinguished the categories by definition – because he argues for the abandonment of the notion of modality anyway, he subsumes both epistemic modality and evidentiality below the notion of ‘attitudinal category’, which sounds clearly more modal than evidential. Four categories are considered ‘attitudinal’ and consequently grouped together: deontic modality, epistemic modality, evidentiality and boulomaic attitude. They are grouped together because “they are all about types of commitments to state of affairs” (Nuyts 2005: 27). But the fact that epistemic modality is concerned with the speaker’s commitment, while evi100 Nuyts (2005: 25, 26) is concerned with the following modal – or, in his terms, ‘attitudinal’ – categories: evidentiality (!), epistemic modality, deontic modality and boulomaic attitude.

188 dentiality may be concerned with the speaker’s commitment to a particular state of affairs should be emphasised. Nuyts himself states that there are certain types of evidentials that do more than only naming the source of information, namely indicating the speaker’s commitment (cf. Nuyts 2005: 23). It is certainly correct to say that evidentiality and epistemic modality may (!) have more in common than, for instance, epistemic modality and dynamic modality, but why then group the former categories together below the heading of ‘attitudinal categories’? As there are only certain types of evidentials or evidential expressions that may simultaneously indicate the speaker’s attitude (the German modals sollen and wollen, for instance) they are not always ‘attitudinal’.101 Grouping them together is certainly right (cf. chapter 2.2). Grouping them together below a notion that sounds ‘too modal’ and at the same time arguing for the abandonment of the category of modality as such does not seem very consequent. Indeed, there is ‘modal confusion’. Boye (2005), as well as Nuyts, argues for a conceptual clarification, but without having the wish to abandon the term ‘modality’ as such. According to Boye (2005: 49), “[t]here is no consensus as regards the application of the term modality” and “as regards the delimitation of modality from related areas” such as evidentiality. The main aim of his paper is to show that the concept of forcedynamic potential helps to define the domain of modality (cf. Boye 2005: 61). First of all, Boye proposes to apply the term ‘modality’ only to “the range of meanings that are often paraphrased with ‘necessity’ and ‘possibility’” (Boye 2005: 49), that is, to meanings which correlate with both epistemic and nonepistemic meanings (cf. Boye 2005: 55). The Danish modal verb kunne is used to illustrate that the meaning of possibility can correspond to epistemic as well as non-epistemic meanings because kunne may convey physical possibility, social possibility (i.e. permission) and epistemic possibility (cf. Boye 2005: 56). Having made this principally clear-cut distinction between what is modal and what is not – an item’s meaning is modal if it expresses necessity or possibility, be it epistemic or not – the author sees himself nevertheless confronted with the questions “What is it that unifies the modal intensities? What is the basic function of this meaning domain susceptible to both epistemic and non-epistemic interpretations?” (Boye 2005: 57). This meaning domain, he concludes, should be conceptually defined, viz. by a “concept that may function as a common conceptual denominator of the meanings encompassed by the field” (Boye 2005: 57). The concept he refers to is the concept of force-dynamic potential which

101 In this regard, Plungian (2001: 354) also says: “not all evidential markers are modal in that they do not all necessarily imply an epistemic judgment”.

189 represents a unification of the concept of force dynamics and the concept of potential. The concept of force dynamics captures the different modal intensities in terms of ‘force’: necessity, for instance, would represent a strong force, while possibility would consequently be identified with a weaker force (cf. Boye 2005: 58). So the different modality types could be described as different types of force dynamics. But as this concept is not a good conceptual denominator to draw the line between modality and other linguistic areas, the concept of force dynamics cannot be used as the sole basis to define the domain of modality (cf. Boye 2005: 59). According to Boye, the concept of potential represents a good complement to the concept of force dynamics. This concept can also function well to capture the different modal intensities: At the one end of the modal intensity scale, possibility may be described as the mere existence of a potential […]. At the other end, necessity may be described as a strong potential […]. Second, the concept makes possible an abstraction from the different types of modality. These may be described as different types of potential (Boye 2005: 59-60).

Boye represents the concept of force-dynamic potential by a scheme which is, on the one hand, similar to Diewald’s (1999) scheme for the deictic use of modal verbs but, on the other hand, more complex/complicated than the scheme proposed by Diewald. The schemes are somehow similar because both represent a relational structure but they differ in one important point: Boye’s scheme is meant to capture all (!) modal meanings. Diewald’s scheme, instead, can only be applied to the deictic (grammaticalised) use of the modal verbs. Like Diewald, Boye operates with a ‘source’, a ‘goal’ and a ‘force’ – whereas Diewald (1999) uses the term ‘path’. But Boye adds a fourth component, namely the ‘agonist’. The ‘agonist’ is what is affected by the modal intensity because “[a] force necessarily affects something” (Boye 2005: 58). So while in Diewald’s terminology the goal is represented by the modalised proposition, that is, the proposition itself is affected by the deictic-modal value, in Boye’s scheme even the goal is represented by an entity that is modified: the agonist. Boye divides his scheme of the concept of force-dynamic potential in three subsituations: (1) represents the initial force-dynamic subsituation, (2) represents the intermediate force-dynamic subsituation, namely the force-dynamic potential, and (3) represents the result, i.e. the final force-dynamic subsituation (cf. Boye 2005: 61). In the second subsituation the concept of potential combines with the concept of forcedynamics because “a force drives an agonist towards a goal that has not yet been reached” (Boye 2005: 62):

190

(1)

Source

(2) FORCE

(3)

Agonist

FORCE

Agonist

FORCE

Agonist

Goal

Goal

Fig. 11: The force-dynamic situation (cf. Boye 2005: 61; slightly adapted)

Boye first explains his scheme of a force-dynamic situation in a very vivid – but ‘non-modal’ – way, that is, without using the term ‘modal’ once: In the initial subsituation, a force-dynamic source, for instance a car owner, produces a force that affects an agonist, for instance a car. In the intermediate subsituation, the force drives the agonist towards a goal, for instance a garage. In the final subsituation, the agonist has reached the goal towards which it was being driven. Thus, this final subsituation is identical with the result of the force affection: the car is in the garage (Boye 2005: 62).

Boye then illustrates how the same concept with its subsituations may be applied to an utterance that contains deontic must: Anja must write because I say so (cf. Boye 2005: 62). The force is represented by the modal. Anja is the agonist so that Anja is affected by the force. The goal is represented by the verb write because the speaker wants Anja to write. This is his goal that, in turn, has its origin in the source which “is expressed by the non-obligatory clausal constituent because I say so” (Boye 2005: 62). The source could therefore be said to be equivalent to the speaker’s origo (cf. Diewald’s (1999) relational structure). The force-dynamic result is “the possible actualisation of the predicational content [Anja write]” (Boye 2005: 63). Hence, as indicated above, Boye also includes the agonist-component to capture uses of modals that are non-deictic, that is, to capture uses of modals that affect an entity being part of [p]. Deictic uses, in contrast, affect the proposition as a whole (cf. Diewald 1999). Diewald clearly

191 needs a source, or origo, to represent the relational structure because the modaldeictic value always originates in a source that is not part of the proposition. Boye includes a source in his scheme adding that the source is expressed by a non-obligatory clausal constituent (Boye 2005: 62). In my view, the clausal constituent because I say so is here necessary unequivocally to identify the use of must as deontic.102 But what is done with the source if I want to capture the meaning of can expressing physical possibility as in Anja can typewrite? What then is the force? Or the goal? From a personal (e-mail) conversation I had with Boye I know103 that the source must not necessarily be explicit; it may also be implicit: There is no explicit source, and therefore we can only speculate about it. The question is: What could elicit the possibility expressed by can? On a dynamic reading of can, it must be what brought about your ability in the first place. For instance, if you broke all your fingers and then recovered and found out you could typewrite again, your mother could inform your father with your example sentence, adding an explicit description of the goal as in …because she recovered. Alternative dynamic sources might be the fact that you have fingers, the fact that you practiced and learned typewriting, or the fact that the good fairy made one of your wishes come true (Boye 14.03.2011).

While the source is not always explicit, the force does not always have to be active: Boye (14.03.2011) explains that there is no active force because of the fact that can is a possibility modal. So it functions as a “remover of a barrier” (Sweetser 1990: 53): In this respect, the meaning of can can be compared to a situation where a water tap is open, but where there is no pressure (i.e. no active force) on the water. By contrast, the meaning of must would be similar to a situation where a water tap is open, and there is a strong pressure on the water (Boye 14.03.2011).

Thanks to this ‘liquid illustration’ the difference between an active and a passive modal force should now be clear. Although neither having an explicit source nor an active force, the goal is a ‘typical’ goal: it is ‘Anja typewrites’. It is possible

102 If the use of must were epistemic, the components that represent a force-dynamic situation would be slightly modified: the source then is the ‘epistemic source’, the force is the ‘epistemic force’ and what is affected by this force would not be an agonist but the ‘predicational content’, whereby the goal then is a certain ‘existential relation’, that is, the epistemically modified content (cf. Boye’s scheme 2005: 66). This scheme is thus highly similar to the one proposed by Diewald (1999). 103 Here, I know is used because the speaker/writer (me) has evidence for her knowing (cf. chapter 5.2.4). I have evidence for the information I transmit, namely Boye’s own words.

192 for Anja to typewrite and if she realises the possibility, she actually typewrites (Boye 14.03.2011). So after having analysed the utterance ‘Anja can typewrite’ against the background of possible contexts and under consideration of the fact that the source may be implicit and the force passive, Boye’s concept of force-dynamic potential may indeed be said to be applicable to all possible uses of modals. Boye then uses modality – understood as a concept of force-dynamic potential – as a point of departure to delimit evidentiality and modality from each other as well as to relate the two domains to each other. It is proposed to capture their relation by viewing the categories “as constituting separate parts of an epistemic scale” (Boye 2005: 73). He explains: By an epistemic scale I mean a continuum consisting of meanings which specify the degree of certainty about the truth (or falsity) of a predicational content. […] the different degrees of epistemic force-dynamic strength or intensity encoded by epistemic modal linguistic items directly reflect different degrees of certainty about the truth of a predicational content. The meaning domain of evidentiality […] may be conceived of as being part of an epistemic scale as well, although not inherently [!]: the different types of epistemic sources encoded by evidential linguistic items only imply different degrees of certainty about the truth of a predicational content (Boye 2005: 73).

This means that evidential meaning is said to involve an epistemic source which is so strong or strong enough that it can serve as evidence for a certain predicational content, whereas epistemic meaning is never so strong as to be able to serve directly as evidence (cf. Boye 2005: 75-76). Generally, evidential meaning is said to involve a stronger epistemic source than epistemic modal meaning (cf. Boye 2005: 76), whereby the two categories are obviously related and one domain is explained by mentioning a feature or the lack of a feature of the other domain. Boye sees one important advantage of representing evidential meaning as epistemically stronger than epistemic modal meaning: it captures the interdependence of both meanings as “strong epistemic modal meaning is closely related to weak evidential meaning” (Boye 2005: 76). While Boye’s line of argumentation seems intersubjectively comprehensible and simply ‘logic’, the following point should be stressed: in order to differentiate modal from evidential meaning, that which is linguistically expressed should always be investigated, whether certain information is contextually provided and whether such information is thus added to the modal/evidential meaning. On the basis of his distinction between modal and evidential meaning, Boye develops the following epistemic scale (or rather, continuum) containing the epistemic variant of modal meaning and evidential meaning as constituents of the scale, whereby the non-epistemic variants of the meaning domain of modal-

193 ity are described as “irrelevant to the scale and therefore placed outside it” (Boye 2005: 74) because they have nothing to do with either the one meaning or the other:

Non-factive, hypothetical meaning

MODAL MEANING · epistemic variant

EVIDENTIAL MEANING

Factive meaning

· deontic variant

· dynamic variant

Fig. 12: Boye’s epistemic scale (cf. 2005: 74)

I am only ‘semi-convinced’ of schematically including evidentiality within the epistemic scale, even though it is said that evidentiality is not an inherent part of that scale or continuum (cf. Boye 2005: 73). But it indeed is true – and that is also revealed by Boye’s epistemic scale – that epistemic modality and evidentiality have more in common than the deontic and dynamic variants of modal meaning with evidentiality, or even with the epistemic meaning as Nuyts (2005) also points out. Heltoft (2005), being concerned with “Modality and subjectivity”, divides the domain of modality in two dimensions, whereby each dimension reflects an opposition pair: necessity and possibility constitute one dimension, subjectivity and objectivity the other one. He – like Boye (2005) – focuses on the development of a conceptual definition of modality, i.e. on the concept of a modal factor (cf. Heltoft 2005: 81). Heltoft works with examples from international literature, whereby most of them are taken from Scandinavian languages, above all from Danish (cf. Heltoft 2005: 82). Heltoft’s aim is to incorporate the dimension of subjectivity in the concept of a modal factor. It is argued that in languages in which a distinction between epistemic and non-epistemic modals is lexicalised a third dimension should be included in order to handle modal phenomena and this third dimension is subjectivity (cf. Heltoft 2005: 85). Subjectivity is distinguished from objectivity via the location of the modal factor: “modal factor location can be either in the speaker or outside the speaker and in this sense, we shall speak of subjectivity

194 vs. objectivity” (Heltoft 2005: 85-86). Hence, if the modal factor is positioned in the speaker the epistemic qualification is termed subjective, while it is objective if the modal factor lies outside the speaker. This illustration – as with Boye’s (2005) scheme – again is similar to Diewald’s (1999) relational structure for the deictic use of modal verbs, whereby Diewald is not explicitly concerned with the differentiation between an objective use of the modals and a subjective use: according to Heltoft, the use of the modal is subjective, if the modal factor is located within the speaker. Diewald indicates that the use of the modal is deictic, if the deictic value originates in the speaker’s origo. According to Boye, the epistemic force originates in an epistemic source (cf. Boye 2005: 66). Diewald’s and Heltoft’s explanations can be seen as a proof for the (notional) relation between deixis and subjectivity (cf. also Cappelli 2007 and chapter 2.2 of the present study): what the one author explains in terms of subjectivity, the other explains in terms of deixis. So according to Heltoft, modality should be defined as comprising two semantic domains, namely necessity vs. possibility and the location of a modal factor (cf. Heltoft 2005: 86), whereby the latter obviously corresponds to the formerly made distinction between subjectivity and objectivity. Concerning the modal factor’s location, it is not only distinguished between inside and outside the speaker; the modal factor’s position is, more precisely, “bound up with some notion of a predicational hierarchy” (Heltoft 2005: 86). The author hereby relies on theories of the clause as a layered system (see, for instance, Hengeveld 1989). The above presented distinction between subjective and objective uses of the modals belongs, for instance, to the third, propositional, level. “The third level […] is the propositional level and the modal factor is either subjective in the sense of speaker-bound, or it is objective in the sense of located in the world described by the speaker” (Heltoft 2005: 87). Logically, there are also levels one and two, as well as level four. While the bottom and second levels are concerned with the conditions that only one single predication represents the level of analysis and that the modal factor is not ‘allowed’ to be located in the modal predication; it has to be located in a different predication (cf. Heltoft 2005: 87), the fourth level differentiates more precisely between the opposition subjectivity vs. objectivity: At the fourth level, again, the modal factor can be subjective in the sense of speakerbound, but at this level, objective modal factor position will mean location in a potential or actual dialogue partner different from the speaker (Heltoft 2005: 87).

So while the third level is characterised by the propositional position of the modal factor, the fourth level is characterised by the dialogic position of the modal factor:

195 Table 11: The positions of the modal factor (cf. Heltoft 2005: 87; schematised) LEVELS THREE AND FOUR 3) Propositional position

SUBJECTIVE Speaker-borne point of view

4) Dialogic position

View held by speaker

OBJECTIVE View held by some other agent described in the proposition View held by some other locutionary agent

In the Dutch Theory of Functional Grammar, there are three classes of modals. The classes correspond to the modals’ positions in the layered system: if the modals are inherent to the predication, the class is called ‘inherent’; if the modals function as predicational operators, the class is termed ‘objective’ and if the modals represent propositional operators, the class is called ‘subjective’ (cf. Heltoft 2005: 89). So epistemic modals belong to the propositional class, but there is no agreement as far as the deontic modals are concerned as they can belong to different classes (cf. Heltoft 2005: 98). Heltoft thus concludes that it is not possible to describe the semantics of modality with respect to only one semantic field. Instead, in defining modality’s semantic substance one should “combine from two systems, namely necessity and possibility and modal factor position” (Heltoft 2005: 98). In sum, Heltoft’s study confirms the assumption that epistemic modals that modify a proposition are subjectively used and that, consequently, subjectivity and epistemic modality are related categories.

3.16 Summarising and concluding thoughts Boye’s Epistemic Meaning: A cross-linguistic study (2006), whose study is of high importance for the development of a superordinate category that subsumes the categories of evidentiality, epistemic modality, polyphony, subjectivity and deixis (cf. chapter 2.2), would have been equally worth being listed and reviewed here together with other studies that also have a ‘more modal than evidential’ background – such as Pietrandrea’s Epistemic Modality. Functional properties and the Italian system (2005; cf. chapter 3.13) or the collected volume edited by Abraham/Leiss (2009; cf. chapter 3.14). The 33rd Journal of Pragmatics (2001), edited by Dendale/Tasmowski, clearly represents another important collection of articles dealing with evidentiality. As most of its papers are cited ‘here and there’ throughout the present study, the articles of this journal were not reviewed in detail here. Equally worthy of mention under the heading of recent studies are the studies by Squartini (2001, 2004) and Wachtmeister Bermúdez (2004, 2005). But as they are exhaustively dealt with in chapter 2.1.2.2 in connection with the ‘deictic question’ or ‘problem of inference’, they have not been reviewed again here.

196 What could finally be concluded from this review-chapter is that different studies with different approaches to one and the same subject or similar phenomena, having different methodological and theoretical background and thus using different terminology, often reveal the same facts. The Belgian Journal of Linguistics that comprises papers on Topics in Subjectification and Modalization (2006) can be exemplarily singled out to illustrate what all studies have shown if looked at from a comparing perspective: 1. The journal’s title relates subjectification and modalisation because they refer to “the expression of the speaker’s stance to the proposition (or part of it)” (Cornillie/Delbecque 2006: vii). It relates the strategies of subjectification and modalisation as both means of modality and subjectivity serve the expression of the speaker’s stance or rather the speaker’s perspective. 2. The journal contains Vandelanotte’s paper (2006) dealing with indirect speech or thought – which could also be analysed in terms of ‘internal polyphony’ – and subjectification. 3. The journal contains Byloo/Kastein/Nuyts’ paper (2006) on certainly and zeker, where they show that zeker may convey an inferential reading (cf. Byloo/Kastein/Nuyts 2006: 53), thus touching upon evidentiality. Other representative examples which show that it is possible to treat one and the same or very similar research question(s) against different methodological and theoretical backgrounds, which implies the operation with different notions and concepts, are the studies by Diewald (1999), Boye (2005) and Heltoft (2005). They are all concerned with the use of modals. Against the background of Dutch Theory of Functional Grammar, Heltoft (2005) distinguishes between an objective and a subjective use of the modal on the basis of the location of the modal factor. The use of the modal is subjective, if the modal factor is located within the speaker. The ‘deictic approach’ to the modals that is adopted by Diewald (1999) reveals that the use of the modal is deictic, if the modal value originates in the origo of the speaker. Boye (2005), in (slight) contrast, explains that it is the epistemic force that originates in an epistemic source. Hence, what the first author explains in terms of subjectivity, the second explains in terms of deixis and the third in terms of force and source against the background of the concept of force-dynamic potential. So the fact that evidentiality and all the other categories should be encompassed by a superordinate category is proven by the point that certain linguistic phenomena are dealt with in terms of one notion by one author and in terms of another notion by another author. To pose another example, while Giacalone Ramat/Topadze highlight the inferential use of the Italian synthetic future, Pietrandrea (2005, 2007) terms it epistemic.

197 In summary, all the studies reviewed in this chapter – whether dealing with evidentiality, epistemic modality, subjectivity, deixis, polyphony (in the sense of quotation) or with two or three categories – illustrate the relatedness of the categories and thus highlight the importance that all these categories should be subsumed below a superordinate category. That is why the categories of evidentiality, epistemic modality, subjectivity, deixis and polyphony are considered being encompassed by the notion of speaker’s perspectivisation. All these categories bring to the fore the speaker’s perspective of the narrated event – even in the case of polyphony, where the speaker shows that he is able to include further perspectives. The fact that evidentiality and all the other categories should have a superordinate category imposed on them is furthermore proven by the different papers and presentations by Haßler, who deals with “Modalidad, evidencialidad y deixis como componentes de la narratividad” (2009), “Epistemic modality and evidentiality and their determination on a deictic basis: the case of Romance languages” (2010a), “Polifonía y deixis en las lenguas románicas” (2010b) or “Modalité et polyphonie: le marquage de la perspective du locuteur” (2011a). The different papers by Dendale and Dendale et al. also serve as proof of the relatedness of the categories: “La polyphonie comme notion épistémique” (Dendale 1992), “Introduction: evidentiality and related notions” (Dendale/Tasmowski 2001), “Point de vue et évidentialité” (Dendale/Coltier 2003) or “La notion de prise en charge ou de responsabilité dans la théorie scandinave de la polyphonie linguistique” (Dendale/Coltier 2005).

198

4 Theoretical and methodological background of the present study Against the background of the studies on evidentiality that have been done so far and the notions of evidentiality, epistemic modality, subjectivity etc. that have been worked out in chapter 2, it is now possible to formulate the theoretical and methodological background of the present study. The point of departure is the fact that the indication of information sources in languages that do not possess real evidentials represents a pragmatic phenomenon and should be context-sensitively analysed against a pragmatic background. As the expression of sources and types of knowledge belongs among the basic communicative or pragmatic necessities of any language, evidentiality must be seen as an indispensable semantic and functional domain in any language. Thus, although the possibility to express evidential judgments is not part of the grammatical system in all languages, i.e., it is not expressed by default and with grammatical means, speakers can always indicate the “reason” or “evidence” for their statements if they consider it neaessary [sic] for some reason (Diewald/Smirnova 2010a: 3).

As a matter of fact, evidentiality can be expressed in Contemporary Spanish and even needs to be expressed. It depends on the speaker-hearer convention or constellation as to whether the source of information for the proposition is expressed or not (cf. Givón 1982). In this connection, Givón (1982: 24) speaks about a pragmatic aspect called “evidentiary justification” (or “evidentiary substantiation”). Generally, with regard to types of propositions, he proposes three types of propositions. The first is “taken for granted, via the force of diverse conventions” and is not open to challenge by the hearer so that no evidentiary justification by the speaker is required (Givón 1982: 24). The second type of proposition is asserted with relative confidence and is therefore open to challenge by the hearer, requiring evidentiary justification (Givón 1982: 24). The third type of proposition is equivalent to hypotheses, is open to challenge and ‘needs’ evidentiary substantiation (Givón 1982: 24). Reading detective novels one often comes across propositions belonging to the second (and third) type as suspects/witnesses are always asked to provide the information source for [p], that is to provide an evidentiary justification for their utterances. Such a justification may be derived from direct evidence plus observation, evidence plus inference, inference (where evidence remains unspecified), reasoned expectation from logic and other facts, and it may rely on different perceptual sources, e.g. on auditory, visual or other perceptions (Diewald/Smirnova 2010a: 2).

199 As already mentioned in the introductory part of the present study, the reading of detective novels was part of the pre-research phase. In the following – as it is part of the methodological background of the present study – the research phases will briefly be dealt with in more detail.

4.1 The research process The research process of the present study can be divided into three phases, more precisely, into two phases of pre-research and one phase of a systematic research, in which a corpus was compiled in order to investigate what is formulated as the aims of the study in the introductory part of the present study. In the first pre-research phase I was concerned with a first and unsystematic analysis of evidential expressions that were found in detective novels. In the second pre-research phase I had a closer look at the same or similar expressions in onlinenews articles. At this point it was obvious that certain expressions should be investigated further. This was done in the phase of a systematic research, in which the corpus data were retrieved from the search engine GlossaNet.

4.1.1 The pre-research phase – part I: detective novels As already indicated (cf. chapter 1), examples from detective novels were obtained during the pre-research phase when I paid special attention to evidential expressions that I came across after having read detective novels by Vázquez Montalbán. Spanish detective novels containing feigned oral speech make clear that the indication of the information source for the state of affairs always depends on the speaker-hearer convention because witnesses are especially asked to tell what they have seen or heard (about). The following examples contain expressions that function as evidentiary justifications. These expressions may be lexical in nature (like examples 59, 60 or 64) or grammatical (like examples 61 or 63): Example (58) contains a form of indirect quoted speech, whereby the person who is also the subject under discussion is quoted. Speaker A quotes what the person has said about himself. By quoting him, speaker A seems to support the idea that ‘he never published anything’ (Nunca publicó nada.) because ‘he said about himself that he was such a perfectionist’ (Decía de sí mismo que era un perfeccionista): (58) [A] – Me consta que escribía versos que nunca publicó. [B] – Nunca publicó nada. [A] – Nunca. Decía de sí mismo que era un perfeccionista. Yo creo que carecía de lenguaje. Eso le pasa a mucha gente. (Los mares del Sur: 39)

200 Example (59) contains the modal verb deber that is inferentially used. The postpostioned explanation beginning with porque justifies why the speaker was let to the conclusion he had drawn. Furthermore, this example contains indirect quoted speech that is part of the justification clause. Because of the fact(s) that the speaker was told by Queta (la Queta me ha contado que…), the speaker was let to his conclusion that ‘it must go well for them’ (Mal no les debe ir): (59) – [...] Tienen corredores. De perfumería, de aparatos de peluquería. Él lleva toda esa parte del negocio. Mal no les debe ir porque la Queta me ha contado que se han comprado un terreno cerca de Mollet, en una zona muy buena porque van a poner las fábricas por allí. (Tatuaje: 181-182)

The adverb probablemente ‘probably’ – in the majority of cases used as a pure modal adverb (cf. chapter 5.1.5) – here seems to be used inferentially: (60) [A] – Yo no le diré que Ramón mató a Julio. [B] – Pero lo hizo. Fue en el chalet, seguro. La propietaria llegó a ver manchas de sangre. [...] Y no pudo hacerlo solo. ¿Cómo iba a poder él solo con el hombre alto y rubio como la cerveza? Queta [C] le miraba ahora perpleja. [C] – Probablemente le ayudaba la familia Larios. El padre. Los dos hermanos. Le deben muchos favores. [...] (Tatuaje: 237)

Speaker C knows that the Larios family owed Ramón a favour. From this fact the speaker infers that Ramón was probably helped by this family. Nevertheless, it could also be argued that the speaker expresses an assumption but the uttered probability that the suspect was helped by them seems to be the result of an inferential process originated in the fact that the Larios family owed Ramón a favour. Example (61) contains the inferentially used synthetic future form será. Due to the fact that speaker B knows that ‘nobobdy comes to see him’ (A mí nadie viene a verme.), he concludes that speaker A is also not asking for him but for Francesc: (61) [A] – ¿Sabe por quién pregunto? [B] – Será por Francesc. A mí nadie viene a verme. (Los mares del Sur: 34)

The following example contains another occurrence of deber. Speaker A concludes that – preconditioned that speaker B really is a satanist – he must like the food ‘well done’ (muy hecho). The afterthought a la brasa o al horno, that is, ‘from the barbecue or oven’ may indicate that speaker A is referring to hell, where the devil lives with burning heat: (62) [A] – ¿Es usted satánico o macrobiótico?

201 [B] – ¿Por qué lo pregunta? [A] – Porque mira el pastel como si fuera a colarle un gol macrobiótico. Me hago un lío con el Yin y el Yang, pero se trata de un inocente pastel de queso griego fresco y pasas. [B] – No soy macrobiótico pero tengo mi propia ideología sobre los alimentos. [A] – Si es usted satánico deben gustarle sobre todo muy hechos, a la brasa o al horno. (El hombre de mi vida: 241)

Example (63) contains a form of the condicional compuesto. ‘It must have been difficult for him’ (Le habría sido difícil) is the utterance under discussion, whereby this use of the condicional is highly likely to appear as a (simple) assumption where the modal component could be regarded as more prominent than the evidential one (cf. the definition of assumption by Boye 2010: 66). But it could be interpreted as an inference because that ‘it must have been difficult for him’ may represent the reason for the fact why the person who is the subject under discussion did not contact speaker B for a long time. Thus, speaker B infers from the fact that they did not have contact for a while because ‘it must have been difficult for him’: (63) [A] – ¿Durante su larga ausencia no se puso nunca en contacto con usted? [B] – Le habría sido difícil. Me conmovió mucho lo que había ocurrido. Me costaba creer que todo había terminado. (Los mares del Sur: 93)

In example (64), the proposition era un lío de maricones ‘it was a tanger of homosexuals’ is introduced by the construction Primero pensamos que ‘First we thought that’, whereof pensamos que could be described as an evidential fragment because it “provides an […] evidential […] perspective on the finite indicative utterance it goes with” (Thompson 2002: 146; cf. chapter 3.3). And this evidential perspective on the finite indicative utterance era un lío de maricones is due to the explanation given by speaker B afterwards: Es muy raro que un tío rico desaparezca y reaparezca apuñalado un año después. Parecía un caso claro de enculamiento ‘It is really weird that a rich guy disappears and than reappears one year later. It seemed to be an obvious case of ***’, and that is why they (the police) thought they would deal with a case where homosexuals were involved: (64) [A] – [...] ¿Usted participó en la investigación sobre Stuart Pedrell? Dígame todo lo que sepa. [B] – Poco. Primero pensamos que era un lío de maricones. Es muy raro que un tío rico desaparezca y reaparezca apuñalado un año después. Parecía un caso claro de enculamiento. Pero por una parte el forense nos dijo que era virgen de culo [...]. (Los Mares del Sur: 21-22)

202 Then the police was told by the forensic scientist that their theory does not work out, which explains why Primero pensamos que is used in past tense. The phrase el forense nos dijo que ‘The forensic scientist told us that’ introduces indirect reported speech, that is, another voice. Example (65) represents another instance of indirect reported speech which represents the evidential subcategory of hearsay as no particular other voice is quoted. Nevertheless, it is said that [p]: (65) [A] – Del muerto ese no sé nada seguro. [...] Ayer hubo una redada [...] [B] – ¿Qué tiene que ver la redada con la información que te pedí? [A] – Es posible que algo. [B] – Dime. [A] – En concreto no sé nada. Pero se dice que el asunto del ahogado tiene algo que ver con todo lo que está ocurriendo. (Tatuaje: 40-41)

4.1.2 The pre-research phase – part II: online-news articles News articles represent a particular type of text: the more information sources are provided, the more reliable is the transmitted news considered. During the pre-research phase – when having had a closer look at the evidential use of adverbs, verbs of cognitive attitude, the conditional and the future form in newspaper articles – I came across a linguistic phenomenon which seems to be a peculiarity in journalistic discourse (cf. also Hennemann 2013). As journalists often report on state of affairs they themselves have not seen, the speaker-hearer convention, or in this case, the recipient-author convention backgrounding onlinenews articles seems to be: ‘whenever possible, indicate the source of information for the information you transmit in your news article’. Sometimes, however, the source of information that is indicated is no more than ‘a source’, that is, the source is literally nominated as una fuente ‘a source’. And if the source of information that is indicated is no more than ‘a source’, the indication of the evidential source could be interpreted as an example for the need or ‘pressure’ to provide a source, which represents a peculiarity of (online-)news articles: Una fuente de seguridad citada por la agencia afirmó que se trató de un “golpe” importante y que los miembros de Al Qaeda mantenían una reunión “en la que planeaban ataques contra intereses yemeníes y extranjeros, entre ellos objetivos económicos clave”. [...] La fuente citada por Saba, no identificada, dijo que planeaban ataques contra intereses extranjeros en venganza por los ataques de la semana pasada [...]. “Las fuerzas de seguridad siguen persiguiendo a los elementos terroristas allá donde se escondan y no habrá tolerancia con ellos para desbaratar sus planes contra la seguridad del país”, añadió la fuente. (El País 24/12/09)

203 El fondo de inversión Ibersuizas anunció ayer la compra del 83% del capital de Ice Cream Factory Comaker, la antigua factoría de Avidesa en Alzira. Una fuente de Ibersuizas que cita Europa Press evitó mencionar el montante de la inversión y comentó: [...] (El País 18/09/2010)

The lexeme fuente ‘source (of information)’ was searched for to gain examples which indicate that news articles represent a particular text type. This particular type of text may not only be characterised by the transmission of news and by providing the corresponding source(s) but also by covering the ‘real’ sources and consequently some facts. The lexeme fuente obviously already refers to an information source because of its lexical meaning. The phrases containing fuente represent a particular kind of quotation since ‘a source is quoted’. The source of information for a certain state of affairs is literally nominated as ‘a source’. This kind of indication of the information source – the nomination of the source – represents a special stylistic device in journalistic discourse because it is usually used to dilute the precise fuente (leaving aside whether the journalist had the possibility of providing the precise source or not). This clearly represents a peculiarity of the text type news article and highlights the fact that news articles are/pretend to be full of ‘evidentiary justifications’ (Givón 1982: 24) on behalf of the journalist. The phenomenon of quoting ‘a source’ is especially interesting from a pragmatic perspective, more precisely, from the perspective of Grice’s conversational maxims (1967/1989; cf. chapter 4.5.1). His maxim of relation ‘be relevant’ often seems to be violated if journalists only provide ‘a source’ as the source of information for a certain proposition. In the following examples the use of the lexeme fuente can be described as being anaphoric because it refers back to a source that was already mentioned, i.e. to a particular source. As fuente represents a synonym for a source that was already mentioned, this use is ‘legitimised’: La AMDH recomendó por ello la apertura de “una investigación neutral y objetiva para determinar las responsabilidades con la aplicación de la Justicia y la activación del principio de no impunidad para los responsables”. La misma fuente confirmó que […] (El Mundo 26/12/2010) [...] según han confirmado fuentes de la Conselleria de Cultura. Las mismas fuentes han explicado que la elección del San Pío V para depositar provisionalmente la obra -que se encuentra en la cámara de seguridad del espacio cultural- se debe a que reúne las condiciones de conservación “más adecuadas” [...] (El Mundo 29/12/2010) El corte del suministro a Domodédovo se debió a la rotura de varios cables que conectan el aeropuerto con dos centrales que le suministran la electricidad, según informó un portavoz de Domodédovo a la agencia Interfax. Según la fuente, la avería fue causada por el repentino aumento de las temperaturas, que ha venido

204 acompañada del deshielo, de una lluvia helada, aguanieve y de una densa niebla. (El Mundo 28/12/2010) [...] según informó a Efe un portavoz de los servicios de emergencia que acudió al lugar del suceso. Al parecer, la explosión tuvo lugar entre la 1.30 y las 2.00 horas de la noche y fue “bastante fuerte”, según la misma fuente, que indicó que causó cierta alarma entre los vecinos de los edificios próximos. (El Mundo 27/12/2010)

In contrast to the anaphoric use of fuente in the examples above, the use of the lexeme in the following example becomes even more interesting from the perspective of Grice’s conversational maxim ‘be relevant’: La Policía iraní arrestó a siete presuntos miembros de la red terrorista internacional Al Qaida que se hallaban en su territorio nacional, informó hoy miércoles la agencia oficial de noticias local Irna. Según el medio, que cita una fuente propia que no identifica, los supuestos islamistas radicales eran objeto de seguimiento desde hace cerca de un mes y fueron detenidos días atrás por orden judicial en la localidad de Seresht [...] (El Mundo 31/12/2010)

On the one hand, one could argue that a news agency that quotes an own/internal source without identifying it – altogether representing the evidential source for [p] – is not information of high relevance. The reader is told that the journalist quotes a press agency which, in turn, cites an internal source. One might argue that the reader does not really know more than before: he knows that there is always an information source for the transmitted information. On the other hand, it could be argued that this information is relevant insofar as the reader now knows that it is up to him whether to rely on this source or not. While in the previous example the journalist provides at least a little information about the source, in the following example it is not more mentioned than ‘a source’ as the source for [p]: [...] a la investigación del suceso como la de un asesinato y que se está a la espera de que se realice la autopsia al cuerpo. Sin embargo “la sugerencia de que haya vínculos terroristas o de seguridad nacional con este caso es poco probable”, aseguró una fuente. (El País 29/08/2010)

From a pragmatic perspective it is especially interesting why the journalist directly quotes the words of ‘a source’ without specifying it. It is disputable how the journalist is able to quote someone literally but why he does not mention more than ‘a source’. In the following example the fuente is at least specified insofar as it is accompanied by the adjective diplomática ‘diplomatic’, even though una fuente diplomática seems to be only slightly more relevant than una fuente:

205 La Organización de Estados Americanos (OEA) ha destituido a su representante especial en Haití, Ricardo Seitenfus, tras las críticas de éste sobre la gestión internacional en la isla, informó una fuente diplomática. (El Mundo 27/12/2010)

The subject of source indication is also debated by readers themselves (see below El lector se queja de que… ‘The reader complains that…’). The following quotes represent parts of a discussion about the questions in which contexts and why it is allowed to quote anonymous or non-identified sources or which circumstances force the journalists to quote an anonymous source (emphases are mine): El control político de la información obliga a veces a no citar fuentes Pero al lado de crónicas y trabajos de investigación en los que el anonimato de la fuente está plenamente justificado, podemos encontrar también en las páginas de EL PAÍS, con más frecuencia de lo deseable, crónicas y reportajes basados en fuentes anónimas cuya ocultación no está en absoluto justificada. (El País 29/09/2010) El lector se queja de que los autores, Fernando Garea y Julio M. Lázaro, incurran en “conjeturas” y no aporten “ni una sola fuente consultada”. Fernando Garea considera que “no se debe abusar de las fuentes anónimas”, pero en este caso cree que estaban justificadas. “Los acontecimientos han demostrado que eran absolutamente fiables, porque distintos portavoces del PSOE [...]” (El País 29/09/2010) Julio M. Lázaro añade que se consultaron también fuentes jurídicas, pero “en asuntos internos de los tribunales no queda más remedio que utilizar fuentes innominadas”. Para muchos lectores, la fuente es un dato fundamental a la hora de valorar la solvencia de una información. El Libro de estilo establece que se respetará el anonimato cuando la fuente lo exija, pero se hará constar la razón. Los asuntos que requieren fuentes anónimas son muy variados, pero las causas no tanto: la mayoría de las veces se hace para proteger a los informantes de posibles represalias. (El País 29/09/2010) Todas las fuentes que se citaban eran genéricas y además la información no iba firmada, con lo que tampoco tenía un garante identificado. No sería la primera vez que se deja de firmar una noticia para proteger a una fuente, pero no era este el caso. “No se firmó porque era una información en la que habían trabajado hasta ocho periodistas”, explica Javier Casqueiro, redactor jefe de España. (El País 29/09/2010) “[...] Pero los datos que facilitaba Presidencia eran incompletos y no concordaban además con los que ya habíamos obtenido, por lo que optamos por publicar estos últimos, aunque sin identificar la fuente para evitar represalias a quienes nos los habían facilitado”. (El País 29/09/2010)

These quotes highlight that the quotation of fuentes represents a common stylistic device in journalistic discourse, that there are usually reasons for the quotation of fuentes only and that readers nevertheless demand reliable sources, to which anonymous sources do not belong: “[p]ara muchos lectores, la fuente es

206 un dato fundamental a la hora de valorar la solvencia de una información.” ‘For many readers, the source is fundamental evidence to evaluate the credibility of information’ (see above). The reader seems to be tolerant if (and only if) he can comprehend why the anonymity of a certain source is preserved by the journalist. It is easier for the reader to tolerate the source’s anonymity if it is explicitly mentioned that it was a wish expressed by the source itself, as in the following examples: Banco Santander, Bank of America e Itaú Unibanco Holding figuran entre los bancos que lideran la operación y que están trabajando en diseñar la oferta pública de venta (OPV) definitiva, según una fuente citada por Bloomberg que pide el anonimato. Un portavoz de la petrolera rehusó comentar el asunto. (El País 02/08/2010) “[...] una vez que el toque de queda entra en vigor, individuos de uniforme que circulan en vehículos todo terreno con cristales oscuros se dirigen a ciertos barrios, secuestran a las personas y se las llevan a lugares desconocidos”, recalcó la fuente, que pidió no ser identificada. (El Mundo 22/12/2010)

The following examples contain police sources. In those contexts it also seems comprehensible that the policeman or officer is not known by name because it is logical that information about a crime has a police source. However, as will be discussed in chapter 4.5.1, it is questionable how relevant the information ‘police source’ is for the reader. It is expected that the reader generally knows that information about a crime or, generally, police information must have originated from a police source: Según un portavoz policial, siete de los heridos se encuentran en estado crítico y otros seis en situación muy grave. De acuerdo con esa fuente, en el autocar viajaban 47 personas, la mayoría de las cuales jóvenes. (El País 29/09/2010) De acuerdo con la fuente policial, dos terroristas suicidas hicieron explotar la carga explosiva que portaban a las puertas de la base con el objetivo de romper el cerco de seguridad y facilitar el paso al resto de agresores. (El Mundo 15/11/2010) Al menos seis personas murieron este lunes y 12 resultaron heridas tras explotar un artefacto frente a un santuario sufí ubicado en la provincia oriental paquistaní de Punjab, informó una fuente policial. (El Mundo 27/10/2010) Un grupo compuesto por entre cinco y seis agresores abrió fuego contra los guardias, según una fuente policial, citada por la cadena privada ‘Express TV’. “Han llegado en coche. Han abierto fuego sobre la policía, antes de precipitar su vehículo cargado de explosivos contra el inmueble” explicó el ministro del Interior de la provincia [...] (El Mundo 13/11/2010) Nicolo Rizzuto, líder de la mafia de Montreal durante mucho tiempo, ha sido asesinado a tiros en su casa este miércoles por la noche, según ha confirmado una fuente policial al ‘Montreal Gazette’. (El Mundo 13/11/2010)

207 Un hombre ha muerto en Bangladesh tras ser golpeado por sus cuatro esposas, que montaron en cólera al descubrir que estaba casado con todas ellas, indicó este miércoles una fuente policial. Según la fuente, citada por el portal informativo ‘New Age’, todo comenzó el pasado lunes [...] (El Mundo 24/12/2010) Tras esta confesión, las mujeres le obligaron a que les llevase hasta la residencia de la cuarta esposa, donde según la fuente policial le golpearon repetidamente. [...] Bapari ingresó en estado crítico en un hospital de Faridpur, donde falleció en la madrugada del martes, concluyó la fuente. (El Mundo 24/12/2010)

As will be indicated in chapter 5.1.7, supuestamente is a common linguistic expression that is used to indicate reported information – especially in contexts of crime and police. But in the examples above the preferred expression to indicate the source of information is its literal nomination. The reason why supuestamente was not used to mark [p] evidentially may be that [p] in the examples above is certain. In contrast, the uses of supuestamente in the examples in chapter 5.1.7 are shown to convey not only a reportive reading but also an inferential one. Only the last phrase in the example above could be replaced by the adverb because [p] is inferentially marked by the verb concluir ‘to conclude’. In all the other cases [p] cannot be marked evidentially by supuestamente instead of by según la fuente policial, for instance, because the information transmitted in the examples above is ‘too certain’ for the use of supuestamente as the adverb is used to report on an inference drawn by a person other than the current speaker. The following examples contain ‘official sources’ which are of special interest from the perspective of the Gricean maxim ‘be relevant’ insofar as it is sometimes not even mentioned in which way they are official. However, at least it ‘sounds official’: Al finalizar el tiroteo, los infantes de la Marina dieron con el rancho y fueron localizados los cadáveres de las 72 personas, ha dicho una fuente oficial, que ha indicado que [...] (El País 27/08/2010) Funcionarios cubanos no estuvieron inmediatamente disponibles para hacer comentarios. Una fuente oficial se limitó a decir: “No está trabajando como primer secretario”. (El Mundo 19/11/2010) Un cooperante holandés fue secuestrado hoy lunes junto a su chófer por un grupo de hombres sin identificar cuando ambos viajaban por el norte de Afganistán, dijo a Efe una fuente oficial. (El Mundo 27/10/2010)

In the following examples the source is likewise nominated but at least specified by an adjective or prepositional group. Examples of this kind are also considered in chapter 4.5.1 from the perspective of the Gricean maxim of relation.

208 “El diseño que se hizo para acoger más de tres millones de visitantes y las prisas por terminarlo abocaron a una carrera inversora disparatada”, aporta una fuente próxima a las negociaciones para levantar el parque. (El País 26/08/2010) “La llave solo la tienen el alcalde y la presidenta de la asociación de la memoria histórica de Granada”, explicó una fuente municipal en alusión al “cuarto” que todo el mundo conoce en el pueblo. (El País 02/08/2010) La retención de los bienes no implica que la compañía pierda su propiedad o que sean embargados, sino solo que no puede venderlos, enajenarlos o entregarlos en garantía, explicó una fuente judicial. (El País 29/09/2010) En cualquier caso, no se debería repetir la agonía griega de la primavera. “Ahora estamos preparados, no tenemos que inventar el mecanismo”, explica una fuente comunitaria. (El Mundo 13/11/2010) Pyongyang, que ordenó ayer la alerta máxima a sus unidades, había efectuado también las maniobras que suelen anunciar un ataque, según dijo a la agencia Yonhap una fuente militar surcoreana [...] (El Mundo 24/12/2010)

This kind of source indication is typical for journalistic writing as (‘normal’) speakers in oral – or even written – conversation would speak about ‘a man/woman from the community’ instead of ‘a communal source’ (una fuente comunitaria) or about ‘someone from the court’ instead of ‘a judicial source’ (una fuente judicial). Imagine a conversation between two persons who do not work as journalists: indicating ‘a European source’ or ‘another source’ would provoke the obvious question ‘Who?’: Alemania prefiere aguantar lo máximo posible, e Irlanda ni siquiera tiene previsto emitir deuda hasta enero, pero la influencia negativa en otros países periféricos puede precipitar el rescate. Una fuente europea aseguraba en octubre que Irlanda ya calculaba sus necesidades en hasta 130.000 millones. (El Mundo 13/11/2010) Pero las dudas se multiplican. “Qué gran negocio es éste si uno puede disponer de un día extra de entrenamientos y la sanción se resume en 3.000 euros de multa”, señalaba una fuente por el paddock. “Seguramente pensaron que les saldría rentable”, añadía otra fuente. (El País 16/08/2010)

As there is always assumed to be a source for transmitted information in journalistic contexts, the indication of the source in terms of nominating it represents a special stylistic device in journalistic discourse and is suggested to be only accepted in journalistic contexts (if at all). Although the speaker-hearer convention or author-recipient convention seems to allow such information sources, not every reader is as tolerant as expected by the journalist. If ‘a source’ is mentioned as the source of information for [p], one might argue that the reader does not really know more than before since he knows that there is always an information source for the transmitted information. On the other hand,

209 it could be argued that this information is relevant insofar as it signals the reader that it is up to him whether to rely on this source or not. Even the specification of the fuente by an adjective as in una fuente diplomática, una fuente policial or una fuente oficial seems to be only slightly more relevant than una fuente since the reader knows from the context which kind of information source the transmitted information comes from. It is expected, for instance, that the reader generally knows that information about a crime or, generally, police information must have originated in a police source. In sum, reading online-news articles one gets the impression that interviewed persons as well as journalists are in a similar position to ‘evidentially justify’ the state of affairs they are telling or writing about as witnesses/suspects are in detective novels. Modal adverbs such as aparentemente or supuestamente and the quotative conditional represent only two of various linguistic means to evidentially justify oneself. They are common stylistic devices that are often used by journalists because the journalists’ task is to represent propositions which are not taken for granted or taken as unchallengeable by the hearer (cf. Givón 1982: 24): (66) “Creo que en España tenemos que empezar a hablar de ‘prohibido prohibir’, porque ya se prohíben demasiadas cosas”, reflexionó el líder del Partido Popular a la entrada al coso, aparentemente extrañado de la atención mediática que han recibido este año las corridas. “Yo no vengo aquí a hacer política, porque desde 1983 el día de la Peregrina siempre he estado aquí. [...]” (El País 11/08/2010) (67) En medios judiciales ha hecho mella la denuncia presentada contra varios medios de comunicación, entre ellos EL PAÍS, por revelar detalles sobre los ertzainas que supuestamente espiaron a favor de Aitor Telleria, así como la advertencia del presidente del ABB, Iñaki Gerenabarrena, para que se detenga la filtración de datos contemplados en los sumarios abiertos. (El País 03/08/2010) (68) Desde Berlín ha reiterado su protesta ante la posibilidad de que Assange sea extraditado a EEUU, algo que en su opinión sería “profundamente injusto”. Tal y como ha subrayado que su libro ‘Wikileaks Inside’, publicado en su versión alemana por la editorial Econ Verlag [...] (El Mundo 11/02/2011)

Supuestamente and sería are both used to attribute words (example 68) or information in general (example 67) to another speaker who is to be distinguished from the current one, or more precisely, from the journalist. Hence, both linguistic expressions are used to indicate reported information. Aparentemente in example (66), in contrast, conveys an inferential reading: the journalist indicates that the person he describes was ‘apparently surprised by the attention from the media’. So the journalist does not make a factual claim about the person’s amazement since his proposition is ‘evidentialised’ with the use of aparentemente.

210

4.1.3 The research phase After having gotten the impression from the pre-research phase (part II) that daily newspapers such as El País are a representative ‘information source’ – to speak in evidential terms – for analysing the epistemic and/or evidential use of certain linguistic devices, the work with GlossaNet represented a good opportunity to study the use of linguistic means such as Spanish modal adverbs, future tense, conditional forms, verbs of cognitive attitude etc. in a methodological and subtle way. The following search requests were registered based on the insights from the pre-research phase as they were found to be the most frequent ones: Table 12: Search requests of the present study MODAL ADVERBS VERBS OF COGNITIVE ATTITUDE FUTURE TENSE MODAL VERBS CONDITIONAL

aparentemente, evidentemente, obviamente, posiblemente, probablemente, seguramente, supuestamente creo, pienso, sé, supongo será puede ([ser] que), podría, debe(ría) sería, podría, debería

Due to the text type, modal adverbs, for instance, are used by both interviewed, quoted persons and journalists, whereas verbs of cognitive attitude are only to be found in direct (quoted) speech or in chat forums (cf. the remarks about the text type in the introduction). GlossaNet was set up to search for the verbs in 1st person singular of the simple present tense. This set up was not only chosen because it represented one option to control the amount of data but also because the analysis of verbs of cognitive attitude in their performative use is preferred, that is, “when they truly encode the subject’s [cognitive] attitude” (Cappelli 2007: 112), appearing in the 1st person singular of the simple present tense. Hence the verbs are non-descriptive and have a qualificational function (cf. Cappelli 2007: 112). Forms as the 3rd person singular/plural would have caused to get results that represent forms introducing or marking a quotation (which indeed is a subcategory of evidentiality), be it a direct or indirect form of quotation: (69) “España juega más a lo brasileño que Brasil”, ha dicho, para recordar que “el resultado es importante, pero si cada partido [del Mundial] lo ven 400 millones de personas, también hay que pensar en ellos”, es decir, en la estética del juego. Además, piensa que “es una buena noticia que España apueste por jugar bien” y gane, porque otros equipos copian ese estilo y también influye en los niños. (El País 12/07/2010)

211 (70) “[...] Pero los chinos son muy buenos trabajadores”, asegura. Juan, como empresario capitalista, cree que la competencia es el motor de las cosas. Hon Guan, como ex ciudadano chino, también piensa que el régimen comunista de allí vela por las personas que no tienen nada, algo que cree se ha descuidado en nuestra sociedad. (El País 12/07/2010) (71) Sí se aprecian, no obstante, “ampliaciones o reconstrucciones”, “por lo menos dos”, y por lo que queda se supone que el puente tenía tres arcos, el central muy alto, de unos 20 metros [...] (El País 16/10/2010)

Example (69) is a very interesting one because it is not usual to combine a construction that introduces indirect quoted speech (piensa que ‘he thinks that’) with an obviously direct quoted speech part as the quotation marks show. Wilson/Sperber (1993: 15) analyse similar examples as ‘speech reports’ where the current speaker attributes a certain inference to the cited person. Hence, these uses of verbs of cognitive attitude are ‘not non-descriptive’ but descriptive and excluded from the study because they cannot be said to truly encode the subject’s attitude, since in the verb’s descriptive use it is not the speaker himself who chooses the one verb or the other. Considering example (70), for instance, the journalist uses creer ‘to believe/to think’ and pensar ‘to think’ as speech-accompanying verbs but the speaker himself might have used saber ‘to know’ to back up his utterance and emphasise that he ‘really knows’ that [p] or even no verb because of the fact that he knows that [p]. The (inferentially used) future tense form será (cf. examples 72 and 73), the conditional forms (cf. examples 74 and 75) and the modal verbs (cf. examples 76 and 77) are used by both journalists and interviewed, quoted persons, which means that one part of the amount of language data is – if not ‘oral at all’, at least – oral in character. When the forms to be analysed appear in direct quoted speech, they are oral in character as they were uttered in, for instance, an interview or in a chat forum. If a conditional form whose use is not grammatically motivated (cf. chapter 7.1) is used by a journalist it is very likely to represent an instance of the journalistic conditional (cf. example 75): (72) [...] los problemas que encontraron para desmontar el módulo hicieron imposible el cambio, ha informado la NASA. Por eso, se va a realizar un nuevo paseo espacial, no antes del miércoles, y seguramente será necesario uno más para terminar la tarea. La situación en que se encuentra la estación es de normalidad, pero si se estropeara la otra mitad del sistema de climatización mientras no funciona la primera [...] (El País 11/08/2010) (73) [...] Zapatero ha advertido de que el tercer trimestre del año “será peor que el segundo”, cuando la economía española creció un 0,2%, según las cifras adelantadas por el Banco de España. Aunque el presidente ha recordado que la austeridad sigue siendo “una prioridad” para su gabinete [...] (El País 12/08/2010)

212 While the instance of será in example (72) is used by a journalist, será in example (73) is part of directly quoted speech such as sería in example (74). Sería in example (75) represents one instance of the journalistic conditional that is used to indicate reported information: (74) “Sacamos los cuerpos de las fosas, los metemos en cajas ¿y después qué?”, cuestiona Manuel Velasco, de Guerra- Exilio y Memoria. “Nunca queda más del 10% de familiares directos de los fallecidos. Mejor que hacer un mausoleo, sería enterrarlos en un nicho digno”. Su opinión la comparten tanto Todos los Nombres como Unidad y Memoria, que además del protocolo concreto de exhumaciones, proponen un banco único de ADN que favorezca la identificación [...] (El País 03/08/2010) (75) A la pregunta de qué pasará ahora con ese solar, responsables municipales señalan que los pisos se construirán y que corresponderá a Núñez decidir qué hace con la edificabilidad a la que tiene derecho. Una posibilidad sería, aclaran las mismas fuentes, acogerse a la sustitución de establecimientos que prevé el plan de usos. Consiste en la reconversión de pensiones y hostales en hoteles de categoría alta en determinadas circunstancias. (El País 11/08/2010)

Examples (76) and (77) contain puede que. Example (76), in contrast to example (77), is used by a journalist. So modal verbs are linguistic means that appear in both journalistic discourse and spoken discourse: (76) Al margen de sus dos secciones a competición, lo que convierte a Locarno en un festival de cine único en el mundo son sus proyecciones en la Piazza Grande, coronada por una enorme pantalla de 26 metros por 14. Puede que sea una de las más grandes del mundo. Pero seguro que es uno de los sitios más bonitos para ver una película. Las montañas, las casas de colores, las estrellas. (El País 12/08/2010) (77) “Todo es como en el ejército. Los métodos son conocidos: la denigración pública para influir en la sociedad”, afirmaba. “La campaña desatada en la prensa es asociada con usted, puede que usted no diera una orden directa, pero usted lo sabía y no ha parado esta arbitrariedad, y eso es lo importante”, sentenciaba el ex alcalde. El presidente Medvédev zanjó el lunes su conflicto con Luzhkov [...] (El País 01/10/2010)

GlossaNet was set up to search for the modal verbs puede and debe(ría) in 3rd person singular, as these are the forms speakers/writers mostly use if they want to ‘tinge’ their proposition epistemically and/or evidentially. In “their evidential use, they are restricted to an inferential reading. That is, they do not allow for readings that refer to other modes of knowing than inferentiality” (Cornillie 2007b: 115). In example (78) debería is used to convey an inferential reading: (78) Los analistas se muestran cautos sobre el impacto del nuevo CEO en el futuro inmediato de la empresa. “El nuevo jefe es un ingeniero de programas software.

213 Esto debería ayudar a Nokia dar el salto a Internet, como también a mejorar el portafolio de alta gama que la empresa piensa lanzar al mercado en EE UU. [...]” (El País 22/09/2010)

So the data used in this study are primary in nature. Croft (2003) distinguishes between “primary sources” and “secondary sources”. Below the former he subsumes native speaker elicitation, texts and descriptive grammars (cf. Croft 2003: 29), whereby the data in the present study belong to the primary source ‘texts’. In the context of this study the usually favoured primary source ‘descriptive grammars’ (cf. Croft 2003: 29-30) cannot be relied on as a primary source alone because in the descriptive grammar of Spanish, the Gramática Descriptiva, the use of expressions that are investigated here is not exhaustively dealt with. However, the respective chapters, i.e. the chapters which are concerned with a context-sensitive analysis of certain linguistic expressions, will deal with the findings that are presented in the different articles of the Gramática Descriptiva and the findings that are presented in the manual of the Nueva gramática de la lengua española.

4.2 The syntactic position of evidentemente & Co. In the following a few words should be dedicated to the syntactic position of the linguistic means analysed in the present study: roughly speaking, the syntactic position of the expressions under discussion will be ignored; however, where it appears fruitful to the analysis of the phenomenon under discussion, I will have a side glance at the syntactic position of the respective linguistic expression (cf. chapter 5.1.1 on aparentemente). In the following it should become clear that I am aware of the fact that, from a syntactic point of view, the different means appear in different parts of the sentence. However – as this subchapter shows – it has been found more appropriate to deal with evidentiality and its related categories from a functional-semantic point of view rather than from a syntactic one. Bally (1965: 56-75) divides a sentence into modus and dictum, that is, a sentence consists of these two complements. The former adds a qualification to the latter because the modus conveys the attitude of the speaker with regard to the dictum. “For example, in a sentence such as ‘I think that the accused is innocent’, ‘I think’ is modus, ‘that the accused is innocent’ is dictum” (Graffi 2001: 248). What Bally (1965) refers to by the notions of modus and dictum is, from a syntactic point of view in Functional Grammar, also dealt with in terms of predication (cf., for instance, Hengeveld 2005 in chapter 2.2). Böhm (94) differentiates between a “relationale Lesart” and a “konstruktionelle Lesart” of predication. While the ‘relational reading’ of predication refers to the relation be-

214 tween the subject of the sentence and the predicate, the ‘constructional reading’ of predication describes the relation between “Prädikator” (‘predicator’) and its complement(s), i.e. to the syntactic expression of the relation between the predicative head and its complements (cf. Böhm 94). Hence, one could state that the ‘relational reading’ of predication can be investigated at the modus-level alone and/or at the dictum-level alone, whereby the ‘constructional reading’ of predication is bound to the analysis of the interplay between modus and dictum. Consequently, the ‘constructional reading’ of predication is of importance for the present study. In this reading predication can also be defined as “a statement about an entity, that is, a statement which does not merely assert some fact as such but presents some fact as a property ascribed to an entity” (Sasse 1987: 554). So “[t]he act of predication is the act of ascribing a property to a referredto entity” (Searle 1969: 25). If this entity to which a property is applied is understood as a proposition, an adverb – provided that it is used as a sentence adverb – and a verb of cognitive attitude in its qualificational non-descriptive use represent the predicators because both means represent a statement about [p]: if [p], for instance, is supposed, the property of ‘being supposed’ is ascribed to the entity [p], or, to allege another example, if [p] is apparently the case, the property of ‘being apparent’ is ascribed to the entity [p]. Nevertheless, it should be emphasised that an adverb may also be used to modify (only) one part of [p]. Verbs in their qualificational non-descriptive use and adverbs in most of their uses belong, from a syntactic point of view, to the modus-part of the sentence, while the other expressions that are analysed in the present study are syntactically part of the (main) information that is transmitted, namely the dictum. So from a syntactic point of view, modals, the conditional form and the future form are, in contrast to the verbs and adverbs, parts of the propositions themselves. However, from a functional-semantic point of view, not only verbs of cognitive attitude and adverbs but also modal verbs, the future and the conditional analysed in the present study add a qualification to the respective proposition or to a part of the proposition. Thus, in the present study linguistic elements will be dealt with both at the modus-level and at the dictum-level because from a functional-semantic perspective all linguistic elements analysed here share the feature of qualifying [p] – regardless of whether they are part of the dictum or part of the predicator. Furthermore, if expressions of evidentiality are investigated in ‘real’ oral language data, a strict distinction between the different syntactic levels is almost impossible. In dialogic communication speakers often do not end the sentence (or rather utterance) they have begun or do not end it at all. From a syntactic point of view,

215 evidential expressions alone may even represent a sentence (cf. also chapter 4.5.2): (79) Regresó en los primeros días de mayo. Llegó al aeropuerto de Barajas y logró eludir los controles de los policías y los guardias antidroga, pese a venir con las entrañas llenas de bolas de cocaína. Tuvo suerte. Aparentemente. (El País 25/05/2010)

So within the framework of this study, it will not exhaustively be dealt with the different syntactic positions of all the linguistic elements that are the subject under discussion. Nevertheless, the syntactic position at the sentence level will be considered in so far as constructions that introduce a proposition such as supongo que [p] as well as constructions where the verb is to be found at the right periphery of the proposition such as [p], supongo are to be considered here. The same holds – to give another example – for the use of modal adverbs. Sometimes I will have a sideglance at the relationship between syntax and semantics, i.e. the relationship between an item’s syntactic position and the (possible) influence of that position on the meaning of the item (cf. the chapters on the adverbs).

4.3 The notion of procedural meaning With regard to the methodological and theoretical basis, a few words should be devoted to the notion of procedural meaning which is considered highly relevant and important for a few studies that will be referred to in the following section. Cornillie (2010a), for instance, states that it is not satisfactory to only examine the semantic and pragmatic meaning of epistemic/evidential expressions. Their meaning in a given context should always be the subject under discussion, and it should be examined “how the semantic profile allows for procedural meanings” (Cornillie 2010a: 303). In the following, whether and in how far the notion of procedural meaning is relevant for the present study will be outlined. Simon-Vandenbergen/Aijmer have also shown that adverbs of modal certainty “have both epistemic meaning (referential-indexical or conceptual) and procedural or indexical meaning” (2007: 56). They explain: […] it is possible that [adverbs of certainty] have both ‘contentful’ and procedural meaning. An adverb has a specific evidential or epistemic modal meaning. We will refer to this as the conceptual meaning of the adverbs and distinguish it from the procedural meaning. For example, certainly is contentful in that it means epistemic certainty and procedural when looked upon from the perspective of indexing the speaker’s or writer’s stance to the text or one of the participants (Simon-Vandenbergen/Aijmer 2007: 54).

216 So it should be put emphasis on the fact that the notion of procedural meaning Simon-Vandenbergen/Aijmer and Cornillie refer to, is obviously a somehow ‘non-canonical’ use of ‘procedural’. The modal adverbs, for instance, are not treated as having exclusively procedural meaning. ‘Real’ procedural items are often dealt with in a relevance-theoretic framework (cf. Blakemore 1987, 2004; Carston 2002). From a relevance-theoretic point of view procedural items (e.g. determiners such as ‘the’ or discourse markers such as ‘but’) encode instructions that constrain the inferential phase of comprehension, and are involved in the derivation of both explicatures and implicatures. Thus procedural items guide interpretive processes, such as the pragmatic enrichment of semantically underspecified terms […] (Bezuidenhout 2004: 125-126).

Carston (2002) also explains that in decoding procedural meaning the inferential phase of comprehension is primarily at work: Procedural semantics: the category of linguistic semantics whose domain is those linguistic forms whose encoded meaning does not contribute a concept but rather provides a constraint on, or indication of, the way some aspect of pragmatic inference should proceed. Subtypes are (a) constraints on pragmatic inferences involved in deriving the explicit content of the utterance, for example, pronouns and tense; (b) constraints on the derivation of implicatures (intended contextual assumptions and contextual implications), for example, discourse connectives such as ‘moreover’, ‘after all’, ‘but’, ‘so’ (Carston 2002: 379).

Blakemore’s Relevance and Linguistic Meaning. The Semantics and Pragmatics of Discourse Markers (2004) is a very in-depth study that analyses the procedural meaning of discourse connectives such as ‘but’ and ‘however’. In her opinion, procedural theory “deals with the way in which elements of linguistic structure map directly onto computations” (Blakemore 1987: 144). So those connectives are considered elements that constrain the inferential phase of comprehension. In highlighting the borderline between conceptual meaning and the procedural one, Blakemore explains the notion of procedural meaning. The former “contributes concepts to the logical form of a sentence, that is, it enters into the semantic representation. In other words, in conceptual encoding, linguistic forms encode conceptual information” (Huang 2007: 197). The notion of procedural meaning is explained in the following way: […] even when the definition of a concept proves controversial, there is a sense in which each speaker can bring it to consciousness and say whether two expressions encode the same concept without having to actually test whether they can be substituted for each other in all contexts. As anyone who has tried to analyse them will know, the situation is very different with expressions such as but and utterance initial well. Ask a native speaker what these mean, and you are much more likely to receive a description or illustration of their use than a straightforward paraphrase.

217 Moreover, native speakers are unable to judge whether two of these expressions – say, but and however – are synonymous without testing their inter-substitutability in all contexts (Blakemore 2004: 82-83).104

Applying these words to this study: ask what do evidentemente or al parecer, for instance, mean. Referring to the conceptual component of the expressions, the former could be said to be roughly distinguishable between the meaning ‘something which is evident’ and ‘as I can infer from facts that are evident to me’. The latter’s conceptual meaning could be paraphrased as ‘something that seems to be’ and ‘as was told/as I heard’. With respect to the conceptual meaning it is to distinguish between meaning aspects that are encoded and other ones that are contributed by the context. Considering their procedural meaning both evidentemente and al parecer can be said to signal the addressee how the content of the utterance they are part of is to be pragmatically processed. Mood, to pose another example, is also said to have (canonical) procedural meaning: Sperber/Wilson (1995, 1998), for instance, assign mood indicators procedural meaning because they provide a constraint on the pragmatic inference the addressee should proceed regarding the intended propositional attitude that is expressed by the mood indicator: […] illocutionary-force indicators such as declarative or imperative mood or interrogative word order merely have to make manifest a rather abstract property of the speaker’s informative intention: the direction in which the relevance of the utterance is to be sought (Sperber/Wilson 1995: 254).

So mood and discourse connectives are examples105 that have procedural meaning which is characterised by the fact that it

104 Wilson/Sperber also address the meaning of discourse connectives to highlight the distinction between procedural and conceptual meaning: If now or well encodes a proposition, why can it not be brought to consciousness? Why is it so hard for non-native speakers of German to grasp the meaning of ja and doch? [...] Conceptual representations can be brought to consciousness: procedures cannot. (Wilson/Sperber 1993: 16). 105 While discourse connectives are usually “analysed as encoding constraints on the implicatures of an utterance”, mood indicators “are analysed as constraints on explicatures”, as they guide the hearer towards the so-called “intended higher-level explicatures” (Ifantidou 2001: 198). But as this distinction between implicature and explicature will not be dealt with in the present study, and as the reference to Relevance Theory will be limited to an unavoidable minimum, the observation that discourse connectives and mood have procedural meaning suffices for the purpose of explaining the conceptualprocedural distinction.

218 does not contribute any concept but rather provides a constraint on, or indication of the way in which certain aspects of pragmatic inference should proceed, that is to say, it indicates particular computational processes (Huang 2007: 197),

whereby computational processes are understood as applying grammatical rules, inference rules etc. (Wilson/Sperber 1993: 16). But in argumentative line with Simon-Vandenbergen/Aijmer, Haßler (2010d) and me both meanings – the conceptual and the procedural one – must not exclude each other; the coexistence of both meanings is assumed for certain linguistic elements, which again illustrates that these linguists work with a ‘non-canonical’ notion of procedural meaning. It is non-canonical in the sense that it is co-existential with the conceptual meaning. For her study of epistemicity and evidentiality in English verbs of cognitive attitude, Cappelli observes that the distinction between conceptual and procedural meaning has also been equated with the distinction between truth-conditional and non-truth-conditional meaning. Conceptual meaning would be the only one capable of contributing to the truth conditions of an utterance. However, recent studies, such as Ifantidou’s (2001) research on evidentials, have cast some doubt on this assumption, showing that the situation is more complicated and that linguistic items encoding procedural meaning can in fact be truth-conditional (Cappelli 2007: 25; cf. Ifantidou 2001: 89).106

Indeed, Ifantidou focuses “on the claim made by many speech-act theorists that evidentials are illocutionary-force indicators, and are therefore non-truth-conditional”. She argues instead “on closer investigation, this claim turns out to be false” (Ifantidou 2001: 95). So Ifantidou shows for the Modern Greek particle taha, carrying the evidential meanings ‘maybe’107, ‘it seems’ and ‘apparently’ – “the latter also indicting information obtained through hearsay” (Ifandtidou 2001: 170) – that it is truth-conditional (2001: 202), beside being procedural (2001: 187). Concerning Ifantidou, [t]he meaning of a word or other linguistic construction is procedural if it constrains the inferential phase of comprehension by indicating the type of inference process that the hearer is expected to go through (Ifantidou 2001: 198).

106 Cram/Hedley confirm Cappelli’s statement that the notion of procedural meaning is often related to non-truth-conditional meaning because they explain that [t]he category of procedural meaning was initially introduced to deal with expressions which do not form part of the propositional meaning of an utterance, but serve as a procedural signal as to how the content is to be pragmatically processed (Cram/Hedley 2005: 187). 107 Usually, expressions carrying the meaning ‘maybe’ are described as modal instead of evidential.

219 In contrast to Haßler’s (2010d), Simon-Vandenbergen/Aijmer’s (2007) and my line of argumentation, Ifantidou thinks to have observed for English adverbs that they (only) encode conceptual information: “all these adverbials [evidently and allegedly, for instance], whether truth-conditional or non-truth-conditional, encode conceptual information” (Ifantidou 2001: 117; cf. also 120, 187-188, 202). But according to Haßler (2010d), Cornillie (2010a) and Simon-Vandenbergen/Aijmer (2007)108, certain adverbs are procedural in character – besides encoding conceptual meaning – because they are said to encode information about “how utterances containing these expressions should be processed” (Ifantidou 2001: 87). Thus, hearers are partly constrained to pragmatically infer “the intended interpretation using contextual assumptions and general principles of communication” (Ifantidou 2001: 59).109 Regarding myself as a kind of hearer or addressee – as I am the one who tries to determine the meaning, and thus the function, of certain evidential expressions – partly means behaving like a reader, but to a large extent as an analyst. Modal adverbs, for instance, indeed provide “a constraint on, or indication of, the way some aspect of pragmatic inference should proceed” (Carston 2002: 379). If the ‘non-canonical’ use of procedural is dealt with here, that is the notion of procedural meaning that can be combined with conceptual meaning, the following can be stated for modal adverbs: all modal adverbs are in so far procedural to a certain extent as they signal how the propositional content is to be pragmatically processed (cf. Haßler 2010d). So they are also considered truth-conditional (unless they serve discourse functions; cf. Cornillie 2010a, 2010b). Utterances containing a modal adverb at least signal the addressee that the content should not be regarded a factual claim. They signal the addressee that the content, instead, should be interpreted as a non-factual claim. The addressee should arrive at the right interpretation via, e.g., pragmatic 108 Cram/Hedley investigate the procedural meaning of pronouns and explain: “[w]hile they may not be purely procedural, as many discourse markers are argued to be, pronouns are procedural in a central, fundamental way” (Cram/Hedley 2005: 193). That means they – such as the linguists investigating modal adverbs – apply a ‘non-canonical’ procedural meaning to pronouns. 109 According to Ifantidou, in recovering the intended meaning, the addressee relies on the ‘widest context at all’: [it] is obvious that in recovering the speaker’s meaning, context plays an important role. [In Ifantidou’s study], context is not just the preceding linguistic text, or the spatiotemporal setting in which the utterance takes place. It includes any assumptions used to arrive at the intended interpretation, which may be drawn from the immediate linguistic and physical environment, but also from scientific, cultural, or common-sense knowledge, or any type of public or individual information that the hearer has access to at the time (Ifantidou 2001: 61).

220 inference. The same can be applied to the other linguistic forms that are the subject under discussion in the present study, that is, the same ‘to-a-certain-extentprocedural meaning’ can be applied to the meaning of verbs of cognitive attitude, modals, the (journalistic) conditional and the future form. Cram/Hedley (2005: 187) explain that linguistic elements are then procedural if they function “as an overt marker of how the semantic content of the rest of the sentence is to be understood”. This function is fulfilled by well in the following utterance with its ironic reading: Well, you’ve been a real help! What well “indicates is that the speaker intends the hearer to understand that she hasn’t been a help at all, and that is pretty much all it seems to mean” (Cram/Hedley 2005: 187). What all the linguistic items to be discussed in the present study ‘procedurally indicate’ is that the meaning of the utterance is not a factual claim. Hence, the meaning aspects applied to the linguistic elements analysed here can be represented in the following way: MEANING OF x x x x x

Modal adverbs Verbs of cognitive attitude in qualificational nondescriptive construal (Journalistic) conditional Synthetic future Modal verbs

PROCEDURAL

CONCEPTUAL MEANING(S)

MEANING

The addressee must go through the inferential phase of comprehension

Meaning aspect(s) encoded

The utterance is not a factual claim Fig. 13: Procedural meaning and conceptual meaning(s)

Meaning aspect(s) contributed by context

221 The different conceptual meanings (of a certain adverb, for instance) are to be dealt with in terms of determining which meaning aspects are encoded by a certain linguistic element and which meaning aspects are contributed by the context. In order to avoid ‘too subjective’ interpretations – as different readers, and more likely, different analysts may arrive at (slightly) different conclusions having gone through (slightly) different pragmatic inferences – in this study, attention to the linguistic context will be paid as much as possible, trying to make the analyses as ‘intersubjectively comprehensible’ as possible. Consequently, the consideration of the given context that (literally ‘literally’) surrounds every single linguistic means under discussion is highly relevant because […] un élément linguistique ne doit pas être étudié isolé, mais seulement en rapport avec les autres: la sémantique s’était donc orientée vers l’étude des oppositions entre mots voisins, telles qu’elles se manifestent à l’intérieur des “champs” notionnels, de façon à faire apparaître le découpage particulier que chaque langue opère à l’intérieur de ces champs. Ces diverses modifications conservaient cependant le caractère essentiellement paradigmatique reconnu à la sémantique. Décrire un terme, c’était toujours lui attribuer une certaine signification qu’il posséderait abstraction faite de son emploi dans le discours, et qu’il se contenterait d’ “apporter” dans les discours où il apparaît (Ducrot 1984: 47-48). [Il] est difficile de déduire le sens de l’énoncé à partir du sens des mots, si celui-ci n’est pas déjà décrit par rapport à la fonction du mot dans l’énoncé et […] une sémantique paradigmatique ne peut en aucun cas dispenser d’une sémantique syntagmatique (Ducrot 1984: 48-49).

Reyes (1996) illustrates very well that it is important to consider the function of an expression in a certain context – especially when studying their epistemic/evidential use. After giving the following two examples, (a) El asesino ha estado aquí hace poco. (b) Evidentemente, el asesino ha estado aquí hace poco.

she analyses: En [(a)] no hay ningún adverbio que indique probabilidad o duda, del tipo de quizá, ni se han usado tampoco construcciones del tipo es posible que. Hasta podría suceder que [(b)], con su adverbio evidentemente, le parezca a algunos de mis lectores una afirmación más fuerte que el mismo enunciado sin el adverbio. No es así, sin embargo. [(a)] es una afirmación plena, y en cambio [(b)] es una afirmación restringida, pues el adverbio evidentemente, al igual que muchos otros semejantes, alude a la inferencia que ha hecho el hablante para llegar a esa conclusión, y revela que lo afirmado es el producto de una inferencia (Reyes 1996: 28-29).

That means that the function of evidentemente often – but as will be shown in chapter 5.1.2, not always – is to express an inference. The meaning of a certain

222 linguistic item in a certain context always depends on contextually provided information.

4.4 The notion of utterance In this subchapter the opposition pair utterance vs. sentence is briefly dealt with. As in this study evidential expressions are analysed against the background of pragmatics, it is worth mentioning why all utterances under discussion are treated as utterances (Spanish enunciado) and not as sentences (Spanish oración): The term utterance is standardly used in contrast with sentence. A sentence, Sn, is what is characterized by the grammar as an abstract structured string with a semantic interpretation. An utterance of Sn is one instance (or token) of that sentence. […] utterances are instances of the use of sentences […] (Kempson 1977: 8).

So utterances are 1. concrete linguistic sequences which are realised in a certain communicative situation. On the one level, journalists – although doing so in a written way – communicate with their readership as well. On the other level, quoted persons have communicated with their interviewer, and hence realised utterances. 2. realised in a discourse situation, and regarding the two levels, journalist and reader as well as interviewee and interviewer are seen to (have) be(en) in such a situation. 3. to be analised against a pragmatic background, as they are pragmatic units. 4. evidentially marked if the speaker-hearer constellation or author-reader constellation requires it or is expected to require it. That means speakers mark their utterances evidentially if they see obliged to do so by the pragmatic situation. Escandell Vidal (1993) differentiates between sentence (oración) and utterance (enunciado) in the following way:

223 Table 13: The differences between oración and enunciado (Escandell Vidal 1993: 34) -

-

ORACIÓN entidad abstracta, teórica, no realizada se define dentro de una teoría gramatical, con arreglo a criterios de naturaleza gramatical unidad de la gramática su contenido semántico depende de su estructura, no de sus usos posibles se evalúa en términos formales: es correcta o incorrecta

-

-

-

-

ENUNCIADO secuencia lingüística concreta, realizada por un emisor en una situación comunicativa se define dentro de una teoría pragmática, de acuerdo con criterios discursivos unidad del discurso su interpretación depende de su contenido semántico y de sus condiciones de emisión se evalúa según criterios pragmáticos: es adecuado o inadecuado, efectivo o inefectivo...

As a matter of fact, journalists see themselves as being in a dialogue with the recipients of their news, albeit it is not a ‘typical’ oral dialogue that is characterised by an interaction of – at least – two interlocutors. Simon-Vandenbergen/Aijmer (2007: 42) also explain: “[…] when authors take up a position they do not always have a particular reader in mind, but they construct an imagined reader or audience with whom they engage in dialogue”. These words can also be applied to journalists. Even though they do not always clearly take up a position, they are in dialogue with an imagined readership. The ‘typical’ dialogue is described as a multidirectional communication because the dialogue participants represent speaker and addressee simultaneously. The journalistic communication, on the other hand, is unidirectional (cf. Schäfer 2008: 526) because author and recipient are separated – locally as well as temporally; that is why they do not share a communicative situation. The sending and receiving of information occur time-delayedly. The only multidirectional interaction between journalist and addressee that could take place is writing letters to the editor (cf. Schäfer 2008: 526). Nevertheless, the journalist sees himself being in a dialogue situation with the recipient, and that is why the journalist’s news has to be analysed as utterances instead of sentences. This, in turn, means that the statements are analysed against the overall background of pragmatics: “[...] la pragmática se ocupa de los enunciados, y la gramática, de las oraciones [...]” (Escandell Vidal 1993: 271). Sperber/Wilson (1995: 9-10) confirm: “The study of the semantic representation of sentences belongs to grammar; the study of the interpretation of utterances belongs to what is now known as ‘pragmatics’.” And Kempson (1977: 68) summarises that pragmatics is “a theory of communication separate from but dependent on a previously stated account of semantics”. The journalists’ utterances are analysed within the con-

224 text of pragmatics, also because in this study the linguistic domain of evidentiality is reinterpreted pragmatically as the concept of evidentiary justification is a pragmatic phenomenon. This justification can take place at two levels: in the first place, an interviewed person may indicate the source of information for a state of affairs via (inferential) phrases or expressions like supongo que [p], aparentemente [p], creo que será [p] etc.; secondly, if the interviewed person is quoted by the journalist, the quoted speaker represents the source of information for the state of affairs the journalist reports about.

4.5 The pragmatic background of the present study Pragmatically speaking, evidential claims to ‘knowledge’ can be made in a variety of ways through reference to: direct observation (see, hear, taste, smell, feel); indirect [retailed] observation/hearsay (variations on the theme of ‘it is said that’/’the word on the street is’/’rumour has it that’; allegedly, purportedly, reportedly, supposedly); (un)tested claims to knowledge (believe, guess, know, reckon, suppose, suspect think, understand; clearly, obviously); and inference (apparently, evidently, presumably) (Hoye 2008: 158).

‘Pragmatically speaking’, it should be summarised how pragmatics is to be understood in this study: it is commonly acknowledged that Charles Morris introduced “el término pragmática para designar ‘[...] la ciencia de los signos en relación con sus intérpretes’” (Escandell Vidal 1993: 7; cf. Allott 2010: 13). In his work Foundations of the Theory of Signs (1938), Morris draws a distinction between the following three concepts: syntax, the study of formal relations between the signs; semantics, the study of the meaning of signs and the objects they denote; and pragmatics, “the science of the relation of signs to their users” (Morris 1938: 29). Consequently, pragmatics occupies its own place besides semantics and syntax. Morris extended his theory in Signs, Language, and Behavior (1946), where he explains that pragmatics “deals with the origins, uses, and effects of signs within the total behaviour of the interpreters of signs” (Morris 1946: 219). That is why it is commonly acknowledged that the studies by Morris “are one origin of the very broad conception of pragmatics as the study of language use in general” (Allott 2010: 13), even though “it was the philosopher Paul Grice’s William James lectures at Harvard in 1967 that led to the real development of the field” (Sperber/Wilson 2005: 468). In contrast, Escandell Vidal does not understand pragmatics as another level of linguistic description, but rather as “una perspectiva diferente desde la que se pueden contemplar los fenómenos [...] En este sentido, la pragmática no pretende invadir el terreno de la investigación gramatical, sino, en todo caso, complementarlo [...]” (Escandell Vidal 1993: 10-11). Nuyts (1987) does also not accept pragmatics as a different

225 perspective; for him pragmatics is “the overall process of linking semantic and syntactic structures in concrete circumstances” (Nuyts 1987: 730). This means that pragmatics studies the contextualised meaning, whereas semantics studies the decontextualised meaning (basic meaning) (cf. Silva-Corvalán 2001: 131). And Calvo Pérez (1994: 52) explains that “un estudio adecuado del léxico no puede dejar de ser pragmático” because in a communicative situation – written or oral – “las frases pueden adquirir contenidos significativos que no se encuentran directamente en el significado literal” (Escandell Vidal 1993: 26). Hence, pragmatic explanations should not only be formal, but, above all, functional (cf. Escandell Vidal 1993: 47). To know the function of the Spanish adverb evidentemente, for instance, it has to be analysed from a pragmatic perspective. Although it etymologically derives from Latin evidens and videre, in many cases/uses its function has (nearly) nothing/little to do with its semantics (cf. also Reyes 1996: 28-29). If analysing the following examples, (80) “[...] Ahora los borregos que jaleaban y aplaudían a la calamidad cuando advertía que no recortaría los derechos sociales bajo ningun concepto le jalearan y le aplaudiran cuando anuncie el despido libre...Es el fanatismo en estado puro. Evidentemente no son socialistas sino zapateristas. [...]” (El País 30/05/2010) (81) [...] la Policlínica Tibidabo en Barcelona ofrece pastillas y tratamientos a sus pacientes para dejar de ser gays ha reabierto la polémica sobre una opción descartada en 1973, cuando los científicos rechazaron esta inclinación como trastorno psicológico. “Evidentemente, no se puede curar la homosexualidad. Estas terapias suponen mala praxis y están desautorizadas. Causan trastornos depresivos, conductas autodestructivas, ansiedad y pueden derivar en el suicidio”, afirma la psicóloga Silvia Morell. (El País 24/06/2010)

it becomes obvious that evidentemente “en muchos casos no tiene nada que ver con la evidencia verdaderamente visual, sino que sirve para subrayar la incontestabilidad de una conclusión” (Haßler 2004: 232). To sum up, here the notion of pragmatics is understood as “una perspectiva de análisis, un punto de vista, una manera de acercarse al estudio de cualquier fenómeno lingüístico” (Escandell Vidal 1993: 270) – first and foremost because the study deals with the question of evidentiary justification. Against this pragmatic background two pragmatic theories, Grice’s Cooperative Principle and Sperber/Wilson’s Relevance Theory, should be drawn attention to. As the theories often form part of the theoretical and methodological background of studies dealing with evidentiality, they will be the focus in subchapters 4.5.1 and 4.5.2. The theories will be examined within the pragmatic context of evidentiality, whereby the examination will reveal that the applicableness of the latter to the study of evidential expressions is questionable.

226

4.5.1 The Cooperative Principle – Grice’s maxims of conversation Grice’s Cooperative Principle […] is the claim that in conversation participants try to make their contributions suitable to the shared purpose of the ‘talk exchange’ that they are engaged in: that is, they cooperate with each other in the strong sense that they have a shared goal beyond understanding and being understood (Allott 2010: 51).

Although within the context of newspaper articles one does not deal with a typical talk exchange, but rather with a unidirectional communication (cf. chapter 4.4), journalists have the same object: they want to be understood by their recipients. Grice’s Cooperative Principle (1967/1989: 26-27) consists of conversational maxims. He suggests a set of norms, four maxims that operate during a talk exchange: maxim of quantity, maxim of quality, maxim of relation and maxim of manner. Especially the maxim of relation, also called maxim of relevance, is of great importance for the present study. ‘Be relevant’ is the core message of this maxim, which means that a speaker (author) should only contribute relevant information – information being relevant for the addressee (recipient). If analysing example (82) and example (83), it is concluded that – within the context of evidentiary justification – it is relevant for the journalist as well as for the recipient that the former indicates the information source for the respective utterance: (82) Frente a sus acólitos, Edgar Álvarez posaba pisando el balón para que le hicieran fotos. Su entrenador, el seleccionador Reinaldo Rueda, observaba la escena con satisfacción. “En principio, está todo perdido”, explicó Rueda después; “cuando te enfrentas a un rival tan superior, la preparación es más mental que táctica. [...]”. (El País 23/06/2010) (83) Para escenificarla, ayer inauguró en la plaza de Callao la fuente que ha donado al Ayuntamiento. “En vista de que no las coloca, se la hemos regalado, a ver si emprende un plan de recuperación de fuentes”, explicó Marcos Montes, presidente de A Pie. (El País 23/06/2010)

But not all information sources represent a relevant contribution if analysed from the recipient’s perspective. Some indications of the information source can only be interpreted (and understood) from the journalist’s perspective: this is the case with utterances containing information sources for the one and only reason that the journalist can distance himself from the information he transmits (c.f. examples 84 and 85). Obviously, the author wants to transmit information without being responsible for the content or without being able to be responsible. In example (84), for instance, the only possibility for the journalist not to assume

227 responsibility is to indicate a ‘non-identified’ but ‘official’ source of information (cf. also chapter 4.1.2): (84) [...] Afganistán, mientras los atentados siguen siendo una constante. Diez civiles murieron hoy y otros 23 resultaron heridos en tres explosiones de bombas camineras en distintos puntos de la conflictiva provincia de Kandahar (sur de Afganistán), según una fuente oficial. El suceso más grave tuvo lugar en el distrito de Maywand [...] (El País 03/08/2010) (85) Sin embargo, la publicación estadounidense cita a una fuente que asegura que se trata de una clínica al sur de California. Mientras tanto, al parecer los choques de la actriz con la ley empiezan a afectar su carrera. (El País 01/10/2010)

Here the evidential value, “el valor probatorio o evidencial”, of the evidential expression seems to be minimised because the recipient does not know more than there is/was a source for the transmitted information. But this is what the recipient always knows for the news he reads. Hence, it does not seem to be relevant that journalists emphasise having a source for the state of affairs they write about because after indicating a source the recipient does not know more than before. In contrast to those examples above, there are cases in which the (exact) source simply cannot be mentioned due to, for instance, political reasons. The following example represents such a case: (86) Desde Reino Unido, las agencias de seguridad han confirmado la existencia de un compló, que se encontraba en una etapa temprana, y que se sigue trabajando para determinar todas las ramificaciones que pueda tener. Una fuente británica ha dicho que el plan de los terroristas era realizar ataques del tipo Bombay, que en noviembre de 2008 dejó unos 170 muertos tras la toma de varios edificios, entre ellos un hotel de lujo [...] (El País 01/10/2010)

Examples (84) and (85) show that the maxim of relevance proposed by Grice is sometimes violated. The possible violation of his maxims is a famous point of criticism: Most pragmatic theorists agree that conversation (and communication more generally) is governed by principles. Some accept the [Cooperative Principle] and maxims as Grice proposed them, but other theorists have proposed alternative systems, mostly with fewer principles and rules. A common criticism of Grice’s theory is that there are too many maxims […], including blatant violations of maxims, apparent violations and clashes between maxims (Allott 2010: 10).

Thinking the other way around could result in the following idea: the maxim of relevance – and one should keep in mind that the conversational maxims are unsaid/implied – may be violated (more or less) on purpose. If the journalist cannot/does not want to do more than only emphasise having a source of informa-

228 tion for the state of affairs he reports on, he may nevertheless do so because he wants to distance himself from the state of affairs or the source. And if the recipient is (sub)consciously aware of the (violated) conversational maxim, he could be able to interpret it. As in a few studies on evidentiality (Ifantidou 2001, Cappelli 2007, Cornillie 2007a110) Sperber/Wilson’s Relevance Theory (1986) is dealt with, a few words should also be dedicated to this theory if applied to the study of evidential expressions.

4.5.2 Relevance Theory – how relevant is it? In contrast to Grice’s maxims that may be violated or not (Sperber/Wilson 1995111: 162), Sperber/Wilson (1995) affirm that their principle of relevance explains human cognition in general, in the sense that relevance is the only criterion that forces the speaker (or writer) to produce certain information. Hence, they claim that a speaker/writer always estimates his contribution of information as being relevant: To communicate is to claim an individual’s attention; hence to communicate is to imply that the information communicated is relevant. This fundamental idea […], that communicated information comes with a guarantee of relevance, is what in the First Edition was called the principle of relevance and what we would now call the Second, or Communicative Principle of Relevance […]. We argue that this principle of relevance is essential to explaining human communication, and […] it is enough on its own to account for the interaction of linguistic meaning and contextual factors in utterance interpretation (Sperber/Wilson 1995: vii). […] there is a single property – relevance – which makes information worth processing for a human being (Sperber/Wilson 1995: 46).

So Yus Ramos (1997) is right when he comments that “Sperber and Wilson (1986) developed a cognitive approach to interpretation from their initial proposal of the unification of Grice’s (1975) classic maxims in a single principle of relevance” (Yus Ramos 1997: 235). Johnson-Laird (1988: 349) puts it even simpler: “Sperber and Wilson (1986) attempt to reduce Grice’s conventions to one: be relevant”. Sperber/Wilson (1995: 161) themselves clarify that their “principle of relevance is much more explicit than Grice’s co-operative principle and maxims” because 110 Cornillie (2007a: 9) roughly refers to Ifantidou’s relevance-theoretic analysis of evidentials but he himself does not adopt the framework of Relevance Theory. 111 Sperber/Wilson’s work Relevance. Communication and Cognition originally is from 1986. I worked with their second edition that luckily contains at the end of the book a chapter that is entitled “Revisions”.

229 [a]chieving optimal relevance […] is less demanding than obeying the Gricean maxims. In particular, it is possible to be optimally relevant without being ‘as informative as is required’ by the current purposes of the exchange (Grice’s first maxim of quantity): for instance by keeping secret something that it would be relevant to the audience to know. It seems to us to be a matter of common experience that the degree of co-operation described by Grice is not automatically expected of communicators (Sperber/Wilson 1995: 162).

According to Escandell Vidal (1993: 143) the principle of relevance acts like a principle that organises the production of utterances. In my opinion, Relevance Theory is thought to be the innovative answer to the co-operative principle proposed by Grice because it is said to represent “an attempt to work out in detail one of Grice’s central claims: that an essential feature of most human communication is the expression and recognition of intentions” (Wilson/Sperber 2004: 607). Herrero Cecilia (2006: 66-67) describes relevance as “un principio general que dirige la actividad comunicativa del locutor y la actividad interpretativa del interlocutor o destinatario”. Sperber/Wilson (1995) themselves explain that an utterance is only then relevant in a certain context if it modifies the existing context, that is, if the utterance has a ‘contextual effect’: The notion of a contextual effect is essential to a characterisation of relevance. […] having contextual effects is a necessary condition for relevance, and that other things being equal, the greater the contextual effects, the greater the relevance (Sperber/Wilson 1995: 119).

In my view, it would only make sense to speak of a ‘contextual effect’ in connection with evidentiality – if at all – if it were not only applied to whole utterances that transmit information but also to information pieces within an utterance. In the following examples (87, 88, 89) it is obvious that the adverbs modify the conveyed information because they add the relevant information that [p] is/was probablemente/evidentemente/aparentemente the case. From this it follows that the transmitted information is not ‘[p] is/was the case’: (87) No hace falta ser doctor en Sociología para darse cuenta del poder narcotizante que este juego ejerce sobre las masas. Probablemente no fue casual que Zapatero anunciase su tijeretazo contra funcionarios y pensionistas el mismo miércoles en que el Atlético de Madrid ganaba la Europa League [...] (El País 17/06/2010) (88) [...] pero veo que hoy soy la oveja negra. Evidentemente hay cosas mas importantes ahora en este pais! Vale! Y tampoco están ilegalizando la prostitución, nadie ha dicho eso, cada uno es libre mientras no te obliguen!! Solo quieren regularizar publicidad […] (El País 16/05/2010) (89) Regresó en los primeros días de mayo. Llegó al aeropuerto de Barajas y logró eludir los controles de los policías y los guardias antidroga, pese a venir con las

230 entrañas llenas de bolas de cocaína. Tuvo suerte. Aparentemente. (El País 25/05/2010)

With regard to Relevance Theory, Sperber/Wilson (1995: 181) explain: “A speaker who communicates that she believes that P does not automatically communicate that P”. Examples (90)-(94), which contain verbs of cognitive attitude, illustrate this fact: (90) [...] sin citarle jamás, se ocupó en criticar punto por punto la actual política del actual presidente de la República. “Me comprometo en esto porque pienso que los franceses necesitan otra voz”, dijo el político. (El País 23/06/2010) (91) “[...] Cuando veo el debate suscitado por las posibles primarias para elegir al mejor candidato socialista para presidir la Comunidad de Madrid -a veces excesivamente descalificatorio y, por ende, fuera de lugar-, creo que se está perdiendo lo que, a mi juicio, es lo esencial: disputar democráticamente la mayoría, con una oferta atractiva para los ciudadanos madrileños que permita desarrollar una nueva política más social […]” (El País 26/08/2010) (92) “[...] Acabo de leer que, según un reciente sondeo, el 61% de los vascos aprueba la prohibición de fumar en lugares cerrados. Me incluyo en ese grupo. Y creo que el oponerse a esa medida se podrá argumentar, tal vez, sobre la base de los intereses, pero, desde luego, no se puede justificar sobre la de los principios. [...]” (El País 26/08/2010) (93) “[...] El problema es cuanto de ese estatuto es constitucional. La gente habla mucho, pero la complejidad jurídica es muy grande. El PP no recurrió todo el estatuto, sino parte (creo que fueron 110 artículos, y supongo que tendrá unos 200). Y hay gente que se cree que se puede declarar todo inconstitucional (primer error, porque no está recurrido todo), cuando lo que se declaran inconstitucionales son artículos en concreto. [...]” (El País 12/06/2010) (94) “[...] Hoy el modelo general es que haya una intervención normativa media, que se combina con el papel de la negociación colectiva, que debe ser creciente. La contratación temporal creo que es uno de los grandes errores del sistema español. Se optó por ella como sistema de flexibilizar el mercado, y se inoculó un veneno, porque somos el país europeo con más temporalidad y más tasa de paro. [...]” (El País 27/08/2010)

If a speaker/writer expresses that he thinks/believes/supposes that [p], he does not express that [p] is the case. Consequently, insofar as a speaker/writer transmits the information that he thinks/believes/supposes that [p], he transmits relevant information in addition to the content of his utterance, namely that [p] does not represent a fact. In other words, the verbs of cognitive attitude in the examples above are procedural because they signal to the hearer how the rest of the utterance (or even more information) is to be understood (cf. Cram/Hedley 2005). They signal the hearer that he must go through the inferential phase of comprehension in order to understand that the utterance he is confronted with is

231 not a factual claim as the speaker ‘only’ says that he thinks/believes/supposes that [p] instead of [p]. So from a relevance-theoretic point of view, verbs of cognitive attitude serve the same function as modal adverbs do: they have a ‘contextual effect’ within the proposition (actually this was not originally meant by ‘contextual effect’). Based on this, it would be interesting to analyse why a speaker/writer explicitly says that he really knows that [p]: (95) Lo ve como una oportunidad de mostrar otra cara, más optimista, del continente: “Mi gran deseo es que en África tengamos la oportunidad de demostrar que podemos trabajar igual que los demás. Y sé que estaremos a la altura. Estoy convencido”. (El País 21/06/2010) (96) [...] granadino Enrique Morente, a las 22.30. “Esta es mi primera vez en la Noche Blanca del Flamenco. Y sé que es una noche muy flamenca, muy urbana. Es una noche muy, muy especial”, afirma el cantante. (El País 21/06/2010) (97) “[...] Él no mató a tu sobrino, pero ordenaste su muerte. Tú y tus hombres. Lo matasteis mientras estaba desarmado, mientras jugaba con un niño. ¿Es esto honorable? Yo he crecido en tierra de la Camorra y sé cómo piensas. Consideras débil a quien le tiene miedo a la muerte, a quien le tiene miedo a la cárcel. Sabes que si quieres mandar en la vida de la gente, debes pagar este poder. [...]” (El País 21/06/2010)

Is it relevant that a speaker/writer explicitly indicates that he knows the information he transmits? Is there a ‘contextual effect’? Might it be relevant because accentuating that one knows that [p] serves a particular pragmatic function? In which contexts do speakers accentuate that they know that [p]? These questions are dealt with in chapter 5.2.4. Up to this point – but only after the slightly modified application of the notion of ‘contextual effect’ – it could be confirmed that it indeed can make sense to deal with evidentiality within the relevance-theoretic framework, albeit an overvaluation of that theory should be tried to be avoided. Relevance Theory may be additionally – and not exclusively – relied on. Cappelli (2007) explains that the relevance-theoretic approach to communication is compatible with a theory of lexical complexity which she adopts for her description of the semantic complexity of the verbs of cognitive attitude (cf. Cappelli 2007: 15). As Sperber/Wilson state that utterance interpretation partly relies on inference processes and “that understanding an utterance is not simply a matter of linguistic decoding” (Ifantidou 2001: 60), Cappelli (2007) and Ifantidou (2001) rely on this assumption because “evidential information [is said to be] partly linguistically encoded and partly pragmatically inferred” (Ifantidou 2001: 59). Cappelli (2007) also explains in her approach that

232 […] utterances are considered linguistically coded pieces of evidence, which are decoded and enriched through non-demonstrative inference until the intended communicator’s meaning is retrieved (Cappelli 2007: 16). In order to get to the right interpretation, the hearer needs to perform several “actions”: he/she needs to disambiguate ambiguous expressions, assign reference and recover ellipsed material. All these “actions” are the domain of pragmatics and are carried out via inferential processes. Such inferential processes are guided by contextual assumptions and some general principles of communication (Cappelli 2007: 16).

But why is Relevance Theory especially relevant if dealing with evidentiality? In order to get the intended interpretation of an utterance, is it not always the hearer’s task to combine “explicit content, contextual assumptions and implications, and [hence] the speaker’s intended attitude to these” (Ifantidou 2001: 60)? Cappelli herself explains that “understanding the meaning of an utterance is not exhausted by decoding the linguistically-encoded meanings” and that “a significant amount of pragmatic effort on the part of the hearer is necessary in order to arrive at the communicator’s meaning” (Cappelli 2007: 18). This can truly be applied to the understanding of an evidentially marked utterance. But is this not also true for utterances that lack evidential expressions? However, in ‘discovering’ the speaker’s intended meaning and attitude, the context plays a highly important role (cf. Sperber/Wilson 1995: 132-142, Ifantidou 2001: 61, Cappelli 2007: 17). ‘Context’, in Relevance Theory, is defined as follows: The set of premises used in interpreting an utterance […] constitutes what is generally known as the context. A context is a psychological construct, a subset of the hearer’s assumptions about the world. It is these assumptions, of course, rather than the actual state of the world, that affect the interpretation of an utterance. A context in this sense is not limited to information about the immediate physical environment or the immediately preceding utterances: expectations about the future, scientific hypotheses or religious beliefs, anecdotal memories, general cultural assumptions, beliefs about the mental state of the speaker, may all play a role in interpretation (Sperber/Wilson 1995: 15-16).

Carston (2002: 376), referring to ostensive communication, also adds that the “subset of mentally represented assumptions […] is not pre-given but is selected by the hearer”. To summarise it is context that is highly subjective and every listening – as opposed to communicating – individual brings his own personal, subjective context when recovering the speaker’s meaning. The subset of assumption selected by the hearer that Carston and Sperber/Wilson refer to is selected “on the basis of the utterance and his bid for an interpretation consistent with the second principle of relevance” (Carston 2002: 376; my emphasis). The second principle says that “[e]very act of ostensive

233 communication communicates a presumption of its own optimal relevance” (Sperber/Wilson 1995: 260) and it is to be distinguished from the first principle that says: “Human cognition tends to be geared to the maximisation of relevance” (Sperber/Wilson 1995: 260). This should be kept in mind for a moment. In the chapter containing “Revisions” Sperber/Wilson want to clarify that by ‘Principle of Relevance’ they originally referred to the second principle only. Then – simultaneously drawing attention to the fact that in their work from 1986 they speak about ‘two claims’ and not about ‘principles’ – they explain: However, many readers, even careful ones, have used the term ‘Principle of Relevance’ to refer to claim (1). This is a straight misreading, but an understandable one. Claim (1) is more fundamental and general than claim (2), and at least as worthy to be called a principle. We originally called claim (2) a principle to contrast it with other pragmatic ‘principles’ proposed in the literature: in particular Grice’s Co-operative Principle. We failed to foresee that when our book was read and interpreted – as we wanted – in the context of wider cognitive concerns, this use of the term ‘principle’ would seem rather arbitrary, cause unnecessary effort, and hence (as we should have predicted on relevance-theoretic grounds) lead to misinterpretation. We have decided to remedy the situation by talking in future of two Principles of Relevance: the First (or Cognitive) Principle […] and the Second (or Communicative) Principle […]. Throughout this book, the term ‘Principle of Relevance’ refers to the Second […] (Sperber/Wilson 1995: 260-261).

To conclude, in the entire book if speaking about ‘Principle of Relevance’ this claim “Every act of ostensive communication communicates a presumption of its own optimal relevance” is referred to. Cappelli (2007) and Ifantidou (2001) refer to the importance of Relevance Theory in their studies on verbs of cognitive attitude and evidentials, respectively. While the former only devotes a few pages to the relevance-theoretic background (Cappelli 2007: 15-27), more relying on inferential communication in particular, the latter even calls her study Evidentials and Relevance. In order to deal with Ifantidou’s study it is important to know how she defines evidentials. In her view, “evidentials have two main functions: they indicate the source of knowledge, and the speaker’s degree of certainty about the proposition expressed” (Ifantidou 2001: 5; my emphasis). This broad understanding of evidentials seems to become even broader as she also analyses adverbs such as happily or sadly under the heading of ‘evidentials and relevance’ (cf. Ifantidou 2001: 98). This, in turn, might help to understand why she would like to examine “how relevance theory would answer one of the questions raised in this book: how are degrees of speaker commitment pragmatically inferred” (Ifantidou 2001: 67). So it is clear by now why Ifantidou – even though dealing with evidentials – is concerned with speaker commitment: it is due to her starting point, her definition of evidentials.

234 How are propositional attitudes recovered? Ifantidou explains and illustrates her explanation by the example given below: […] the process of enrichment involved in utterance interpretation goes beyond the recovery of the proposition expressed; it includes identifying the speaker’s attitude towards the proposition expressed. Thus, the speaker [of the following example] may intend to communicate not only the proposition expressed but also the propositional-attitude information in [(a), (b) and (c)] (Ifantidou 2001: 75): Susan won the elections. (a) The speaker believes that Susan won the elections. (b) The speaker regrets that Susan won the elections. (c) The speaker feels sad that Susan won the elections (Ifantidou 2001: 76).

Up to here it was assumed that the interpretation, the recovering of the speaker’s propositional attitude clearly depends on the (relevance-theoretic) context ‘created’ by the hearer as “[a] context is a psychological construct, a subset of the hearer’s assumptions about the world” (Sperber/Wilson 1995: 15), which is selected by the hearer only (cf. Carston 2002: 376) in order to serve as a context for a particular situation. It depends on so many facts and circumstances how the hearer – being confronted with the utterance ‘Susan won the elections’ – may recover the speaker’s propositional attitude that I cannot list them all here. How well do speaker and hearer know each other? Does the hearer anything know about the speaker’s relationship to Susan? If yes, how much does he know? Etc. But Ifantidou answers her question “Which of all the possible interpretations is the hearer justified in choosing?” (Ifantidou 2001: 76): The answer follows from the relevance-theoretic comprehension procedure: he is justified in choosing only the first, i.e. most accessible, interpretation which satisfies his expectation of relevance […] (Ifantidou 2001: 76).

Then she admits that paralinguistic features like facial expression or tone of voice may help the hearer to determine the propositional attitude (cf. Ifantidou 2001: 76). This is exactly the point which is crucial: the Communicative Principle of Relevance relies on ostensive (!) communication. It deals “with ostensive stimuli, produced in order to attract our interlocutors’ attention and lead them to turn it to the meaning we want to convey” (Cappelli 2007: 18). Therefore it is not appropriate to interpret a ‘contextless’ utterance as Ifantidou (2001: 76) does – especially if claiming to rely on Relevance Theory. This chapter will be closed with a few words from Sperber/Wilson, who ask: Does the principle of relevance apply to all forms of communication? No: it applies only to ostensive communication [!], not to straightforward coded communication. For instance, a telegraph employee who communicates messages by

235 encoding them is expected to be accurate in her encoding […] (Sperber/Wilson 1995: 158; emphasis in the original).

Consequently, Relevance Theory or rather the Communicative Principle in particular is ‘not relevant’ for the present study. In newspaper articles there are not any (or more cautiously formulated, ‘hardly any’) ostensive stimuli – “designed to attract the audience’s attention” (Wilson/Sperber 2004: 611) – to be found that would force me to select a set of assumptions that represents my context based on “expectations about the future, scientific hypotheses or religious beliefs, anecdotal memories, general cultural assumptions” (Sperber/Wilson 1995: 15-16). As there is no ostensive communication in newspaper articles, there is no inferential communication on my side.112 And it is to challenge where the ostensive communication in Ifantidou’s (2001) and Cappelli’s (2007) studies are.

4.5.3 The present study’s approach to evidentiality in Spanish The preliminaries concerning the methodological and theoretical background of the present study show that the adopted approach to evidentiality in Spanish is context-sensitive and functional. “A functional approach to language means, first of all, investigating how language is used” (Halliday 1973: 7). More precisely, a function-to-form approach will be combined with a form-to-function approach. Nuyts’ study of epistemic modality as a semantic category, for instance, “adopts a radical function-to-form approach to the matter: taking the semantic category as its starting point, it looks into the range of its linguistic manifestations” (2001b: xvi). This is what was already done during the pre-research phase. While reading Spanish detective novels I came across different linguistic means that are used to express evidentiality. Hence, the function-to-form approach is part of the present study. In a later study Nuyts (2005) explains: […] one specific kind of empirical approach, which is absolutely dominant in current functionalist linguistics[,] [is] what may be called a ‘form-to-function’ approach. In such an approach, the issue of the relationship between forms and meanings or functions is addressed from the perspective of the forms (i.e. the modal auxiliaries); meanings are handled as they impose themselves in the course of one’s 112 “Inferential communication and ostension are one and the same process, but seen from two different points of view: that of the communicator who is involved in ostension and that of the audience who is involved in inference” (Sperber/Wilson 1995: 54).

236 concern with forms and exclusively in function of the analysis of the forms. While there is obviously nothing wrong with this approach per se, it potentially leads to a biased view of the ‘matching’ problem in language – hence of the way language ‘functions’ – unless it is systematically complemented by a ‘function-to-form’ approach, i.e. one in which one takes the perspective of the meaning categories expressed in language and looks at the range of possibilities to realise them formally in the language (Nuyts 2005: 14).

With reference to the present study, the function-to-form approach can be considered the ‘precursor’ of the form-to-function approach. Since evidentiality can be conveyed by various linguistic expressions, GlossaNet was set up to search for those expressions that were regarded as being used evidentially. Then adopting the form-to-function approach means taking a particular linguistic form as its starting point and determining whether it is really used to convey evidentiality and if so, what kind.113 For example, if the form sería can be used to convey an inferential reading as well as a reportive one, it is to find out which function it fulfils in a particular context. However, if forms fulfil functions that can in no way be termed evidential, these uses are after a first analysis excluded from an in-depth analysis. They will only briefly be considered in order to provide counter-examples or rather to provide an overview of the other functions a particular form fulfils. To provide another example, chapter 5.2.3 represents an excursus on the different functions of pensar in its qualificational non-descriptive construal. Pensar is used as an example to show the various functions of verbs of cognitive attitude. But in chapter 5.2.2, in contrast, only those instances of pensar which express an inferential reading are studied. So in the present study only forms that convey an evidential reading are included, whereby the meaning may overlap with meanings of other linguistic categories (cf. chapter 2.2). However, I will sometimes take a sideglance at other meanings the expressions under discussion may convey. In a few studies where a function-to-form approach is adopted the approach is combined with the framework of Cognitive Grammar (see, for instance, Cornillie 2007a). Cornillie (2007a), adopting a cognitive-functional approach to evidentiality and epistemic modality in Spanish (semi-)auxiliaries, for example, explains:

113 Bondarko (1991) also explains that “[t]he correlation between means and functions can be most exhaustively described by [a] combination of approaches from the point of view of means and from the point of view of functions” (Bondarko 1991: 6). Furthermore, the form-to-function approach corresponds to the point of view of the hearer, while the function-to-form approach corresponds to the point of view of the speaker (Bondarko 1991: 7).

237 The strength of Cognitive Grammar lies in the way it accounts for meaning nuances between apparently similar linguistic expressions: the construal of a linguistic expression is a representation of an interplay of prominent and less prominent units of meaning (Cornillie 2007a: 13).

As already indicated, the present study will also be concerned with prominent and less prominent units of meaning of a certain linguistic expression. Nevertheless, it is preferable to speak about a context-sensitive approach instead of a cognitive approach. The reason is that what the speakers/journalists were cognitively processing when expressing a certain utterance simply cannot be found out. It can only be tried to trace back why a particular speaker in a particular situation used a particular linguistic expression and speculated what the goal behind using this particular expression was. This will be done with the help of the context. However, there is ‘something cognitive’ about the approach. To study language from the cognitive-linguistic perspective “is to study patterns of conceptualisation” (Evans/Green 2006: 5), which will obviously be done here since the conceptualisation of evidentiality in certain linguistic expressions is investigated. Nevertheless, this ‘something cognitive’ of the present approach is not regarded enough to speak about a cognitive approach. In any case, the term ‘cognitive’ would be misleading if it were applied to the approach adopted in the present study. I will stick to contextually-provided information in order to find out which function(s) a particular item fulfils. That is why the present approach is termed functional and context-sensitive. The analysis of the meanings of the linguistic expressions under discussion will concentrate on the conceptual meanings, whereby it will be tried to differentiate between meaning aspects that are encoded by a particular item and meaning aspects that are contributed by the context. The procedural meaning of the expressions to be analysed is that they signal the hearer that they must go through the inferential phase of comprehension in order to understand that the utterance they are part of is not a factual claim. Their procedural meaning will thus not be further considered in the analysis (cf. chapter 4.3). I subscribe to Cappelli’s (2007), Ifantidou’s (2001) and Papafragou’s (2005) view that linguistic expressions which encode procedural meaning can be truthconditional. Papafragou shows with her (re-)analysis of a few modal items that they contribute to truth-conditional content. She provides the following examples (a) My grandfather must be sick. (b) My grandfather may be sick. (c) My grandfather is sick. and argues:

238 [if] epistemic modal verbs do not contribute to the proposition expressed, all three utterances should express the same proposition (that the speaker’s grandfather is sick), albeit with different degrees of speaker commitment. It follows that, if in fact that person is very healthy, what the speaker has said in [(a)-(c)] is false – and furthermore, false for exactly the same reasons throughout (a-c). However, most people would agree that, in these circumstances, the speaker has said something false only in [(c)]. In [(a)] or [(b)], the speaker has simply said that, as far as she knows, it is necessary/possible that her grandfather is sick. This fact suggests that epistemic interpretations of modals belong to the propositional content of the utterance – for which the speaker can be held accountable at later stages of the conversational exchange (Papafragou 2005: 1693).

This line of argumentation can be followed very well so that the consideration of linguistic expressions which encode procedural meaning being truth-conditional should be intersubjectively comprehensible. As already indicated, against the methodological and theoretical background of the present study, a context-sensitive and functional approach to lexical and grammatical evidential expressions in Spanish is adopted. The following chapters thus represent an in-depth analysis of the use of verbs of cognitive attitude, modal adverbs, modal verbs, the synthetic future and the conditional form. The analysis will be a qualitative one, even though it is worked with a quantitatively remarkable amount of language data, which I received by email from GlossaNet in the time span 25/04/2010-27/04/2011. All the data are taken from the Spanish daily newspapers El País and El Mundo. As already mentioned, the programme does not allow working with asterisks to replace verbal endings so that GlossaNet was set up to search for entire lexical items. In the case of the conditional and the future I had to decide in favour of particular lexical forms as well: ser and deber in their conditional form were chosen, that is, sería and debería. So debería is expected to serve two purposes: on the one hand, it may appear as the journalistic conditional and thus in its reportive use; on the other hand, it may appear as the form to express an inference based on uncertain knowledge and to convey conditioned necessity in contrast to the indicative form debe that is expected to express an inference based on objective (or rather: intersubjective) knowledge and to convey unconditioned necessity (cf. Pietrandrea 2005: 74-76). Hence, debe represents one form of the modals GlossaNet was searching for. The other indicative form is puede. In case examples are quoted which contain forms that are not listed in the table above, I came across them incidentally. This means that they appeared in the surroundings of one of the forms GlossaNet was searching for. The dates that are bracketed below the examples are the dates of receipt. Usually, GlossaNet runs two days late. I received the results in text format. The form searched for was supplied with 250 characters on either side. Sometimes a few characters

239 were ‘lost’ because of ‘information’ like the following one: “Reuters Reuters | Manchester Actualizado jueves 23/12/2010 20:01 horas * Disminuye el tamaño del texto * Aumenta el tamaño del texto Comentarios 0” or “Publicidad Sin_Cruz [IMAGE] * Todos * Mejor valorados * Te mencionan * Tu red 4 » Comentarios ¿Quieres comentar? Entra o regístrate 1. Cerrar ventana Redactar mensaje privado Enviar mensaje”. Examples that suffered too much from ‘Reuters’ or ‘Disminuye el tamaño del texto’ so that the context was downsized to an ‘unanalysable’ minimum were excluded from the study. In contrast, examples that contain many more characters will be analysed as well because sometimes it was possible to merge one end of the quote with the beginning of another.114 If examples contain grammatical and/or spelling mistakes, which may even sometimes happen in journalistic discourse, they are mostly to be found in chat forums, where speakers – fortunately – write as they speak so that the use of forms extracted from chat forums can indeed be termed ‘oral in character’. Newspaper chat forums were incorporated in the search by GlossaNet and the examples found in chat forums are – exactly as they appear – included in the present study. As already indicated, the study in its entirety is guided by the assumption that language can only be described adequately if it is done against the background of pragmatics and if the language data are not gained introspectively.

114 So the data which I received by email represent the basis for the analysis.

240

5 A context-sensitive and functional approach to the use of verbs of cognitive attitude and modal adverbs in Spanish The works “I reckon I know how Leonardo da Vinci must have felt…”. Epistemicity, evidentiality and English verbs of cognitive attitude (Cappelli 2007) and The Semantic Field of Modal Certainty. A Corpus-Based Study of English Adverbs (Simon-Vandenbergen/Aijmer 2007) are two studies that treat verbs of cognitive attitude and adverbs of certainty in epistemic-evidential terms (cf. chapters 3.10 and 3.9, respectively). They are of great relevance for the present chapter and its subchapters. Cornillie’s studies (2010a, 2010b) of Spanish epistemic and evidential adverbs are also of great importance for the following subchapters.115 In contrast to Cappelli, who describes the epistemic and evidential use of British English verbs of cognitive attitude in a very detailed way, Simon-Vandenbergen/Aijmer explicitly examine the context-sensitive use of English modal adverbs. In Spanish verbs of cognitive attitude as well as modal adverbs also represent linguistic devices which are not only used to express the speaker’s epistemic stance but also evidentiality. So the aim of this chapter is to determine which meaning aspects of the verbs and modal adverbs can be said to be encoded by the linguistic item under discussion and which ones are contributed by the context. In Cappelli’s line of argumentation, for Spanish verbs of cognitive attitude I also expect that each and every one of the members of the class has its own place within the system, even though contextual constraints can force the single verbs into the semantic space of the others (Cappelli 2007: 301),

and that “[n]ot all evidential forms inherently convey epistemic meaning, and not all epistemic expressions bear reference to evidence” (Cappelli 2007: 130). In this study the context will be used to determine which meaning aspects of an element are encoded and which meaning aspects are contributed by the context. The adverbs evidentemente ‘evidently’ and obviamente ‘obviously’, for example, encode the meaning aspect that something ‘is evident/obvious’ – that something is visually perceivable. Both adverbs seem to indicate ‘what is known from visual evidence’. Evidentemente may also be used if the evidence is inferential in nature, as example (98) indicates. Example (99) contains obviamente 115 This chapter, including the corresponding subchapters, is based on the paper “The epistemic and evidential use of Spanish modal adverbs and verbs of cognitive attitude” (Hennemann 2012b).

241 with its meaning ‘it is obvious’, that is, it conveys the meaning which is encoded. In example (98) the inferential meaning is contributed by the context, while in example (99) the encoded meaning aspect of obviamente is verified by the context: the fact that the boot is broader is visually perceptible: (98) [...] la Policlínica Tibidabo en Barcelona ofrece pastillas y tratamientos a sus pacientes para dejar de ser gays ha reabierto la polémica sobre una opción descartada en 1973, cuando los científicos rechazaron esta inclinación como trastorno psicológico. “Evidentemente, no se puede curar la homosexualidad. Estas terapias suponen mala praxis y están desautorizadas. Causan trastornos depresivos, conductas autodestructivas, ansiedad y pueden derivar en el suicidio”, afirma la psicóloga Silvia Morell. (El País 24/06/2010) (99) Una protección que también se ha puesto en la pierna izquierda, a petición del propio Rossi, que tendrá así las mismas sensaciones en ambas piernas. También la bota ha sufrido modificaciones: es obviamente más ancha para garantizar un mejor confort de la pierna y para permitir una mejor inserción de la misma; además, se ha añadido una segunda cremallera, una apertura lateral para facilitar [...] (El País 19/07/2010)

The examples indicate that, because of the context-sensitiveness, I will have to go beyond the sentence as “logical […] relations which are expressed within the sentence in one case may indeed be expressed between sentences in other cases” (Simon-Vandenbergen/Aijmer 2007: 82). Verbs of cognitive attitude were extensively studied by, for instance, Cappelli (2007) and Nuyts (2001b). Cappelli (2007: 301) describes them as “one of the means to give voice to one’s own epistemic-evidential stance”. This means that Cappelli recognises a relation of dependency between the linguistic categories evidentiality and epistemic modality, whereas Squartini (2004) tries to disentangle this relationship in the Romance languages. Cappelli also points out that even though [e]pistemic evaluation changes according to the evaluation of the evidence on which it is based [and] we can conclude that evidentiality and epistemicity are intimately interwoven semantic domains that are sometimes very difficult to split […] (Cappelli 2007: 130),

epistemic modality and evidentiality are distinct categories (Cappelli 2007: 130). According to Nuyts, the two linguistic domains overlap in every “epistemic use of the mental state predicates” (Nuyts 2001b: 108). He explains that every speaker using an epistemic predicate relies on the information source he has for the state of affairs (Nuyts 2001b: 111). But the question is whether this reliance on the information source is linguistically encoded or not, or whether it can be verified with the help of the context. It should be distinguished between meaning aspects that are encoded by a particular lexical item and other ones that are

242 contributed by the context. For evidently, for instance, Simon-Vandenbergen/Aijmer explain: “[e]vidently seems to have two senses, ‘in a way that is easy to see or understand’ and ‘according to what we know, especially from hearsay’” (Simon-Vandenbergen/Aijmer 2007: 161). For the Spanish equivalent two meanings could be determined so far as well. They are slightly different though. The adverb is used to express both visual evidence (evidentemente[vis]) and inference (evidentemente[inf]): (100) [...] menos que hace un año, 20.079 bolivianos menos o 11.713 argentinos menos. Pero es equívoco afirmar que todas estas personas “regresaron a casa”. Lo que nos dicen los datos del padrón es que se ha producido un cambio de ciclo y, evidentemente[inf], llegan muchos menos extranjeros. (El País 09/05/2010) (101) ¿Fueron los neumáticos o fue Lorenzo el culpable de la caída? “¿Estaba rodando demasiado al límite porque le estaba presionando mucho?”, preguntaron al mallorquín. “Evidentemente[vis], no iba de paseo, pero los Bridgestone tienen este problema”, respondió. “El problema viene cuando no esperas que el neumático esté frío, como cuando paras para dejar pasar a alguien. [...]” (El País 08/06/2010) (102) [...] conforme nos aproximamos a esa boca tenebrosa, las distancias se deforman y las horas comienzan a tartamudear, como si de repente habitáramos en el interior de una pelota de papel arrugada por un niño. Evidentemente[inf], existe una frontera invisible en los alrededores de un agujero negro más allá de la cual la curiosidad es sinónimo de aniquilación. Si nos acercamos demasiado al abismo veremos cosas que jamás nadie creería [...] (El País 18/05/2010)

It is reasonable to think that in example (101) evidentemente is used to express visual evidence because the interviewing and the interviewed person speak about a competition both persons or at least the interviewee have/has witnessed. The uses in examples (100) and (102), on the other hand, seem to be inferential ones. In example (100), the affirmation that recently less foreigners are coming is due to statistical dates available to the speaker, which he draws the conclusion from. The speaker in example (102) mentions an ‘existing invisible frontier’ that firstly cannot be experienced visually and that secondly represents a kind of description of the current situation. The uttered inference is due to his experience, or rather perception of the situation. So these three examples and two different meanings are one reason to only analyse utterances which are embedded in context: Evidential meanings are not simply restricted to single utterances (utterance is used here to cover pragmatic instantiations of language, be these spoken or written) but are perhaps best explored and understood across connected utterances, in identifiable contexts, at levels coherent with more extended instantiations of language use, where the role of evidentiality in discourse is most readily apparent (Hoye 2008: 152-153).

243 Kärkkäinen, who studied the linguistic encoding of epistemic stance in English conversation, explains the high importance of the context illustrating this with the use of I think:116 When a token of I think appears in context, it becomes impossible to determine its precise semantic meaning out of context, on its own and independent of the utterance in which it occurs. […] the linguistic context of the speaker’s current turn provides clues as to which aspect of meaning, doubt or the speaker’s (strong) commitment to the truth value, is foregrounded (Kärkkäinen 2003: 111).

The following example containing a form of Spanish pensar emphasises this fact: (103) El rival será el Fulham, la gran sorpresa de esta competición y un equipo del que Quique no se fía: “yo desconfío siempre de los equipos que llegan a una final. Pienso que seguramente será un equipo muy competitivo y muy inglés, es decir, muy difícil”. (El País 02/05/2010)

The presence of seguramente and the future form será, which are found at the sentence level, help to determine that pienso que is used inferentially. Squartini (2001: 321) has shown that Spanish será is a prominent linguistic device to express inference. So here pienso que is surrounded by a context that is inferential in nature. The utterance before the one containing pienso que may underline this: the speaker always distrusts (socker) teams that get to the final (yo desconfío siempre de los equipos que llegan a una final). That means the speaker draws a conclusion out of his live experience. In Willet’s terms it is an inference belonging to the subcategory ‘reasoning’ (1988: 57).117 Hence, it could be concluded that pensar is not inherently inferential. Its inferential meaning is actually contributed by seguramente, será and the utterance before. Cappelli (2007: 185) developed the following table118 for English think, which could also be applied to Spanish pensar:

116 From a semantic-functional perspective, verbs of cognitive attitude are comparable with adverbs since both qualify [p] or a part of [p] epistemically/evidentially (cf. chapter 4.2). 117 Plungian (2001: 354) differentiates between “synchronous inference”, “retrospective inference” and “reasoning”. He describes the first one with the words “P, because I can observe some signs of P [P at T0]”, the second one as “P, because I can observe some traces of P [P before T0] and the latter as “P, because I know Q, and I know that Q entails P”, what example (103) illustrates. 118 The tables proposed by Cappelli summarise the “semantics” (Cappelli 2007: 224) or “conceptual dimensions” lexicalised by a certain verb (Cappelli 2007: 185).

244 Table 14: The conceptual dimensions lexicalised by think (Cappelli 2007: 185) EVIDENTIALITY think

non-inherent

EPISTEMICITY Likelihood Certainty positive degree highly variable according to the context

COGNITION computational process over available evidence

As Cappelli points out the English verb does not inherently lexicalise evidentiality. Concerning this category, it can be said that Spanish pensar is used in a similar way as English think (cf. chapter 5.2.2). Analysing the epistemic-evidential use of modal adverbs and verbs of cognitive attitude also means to look at the mood of the verbal forms because […] the subjunctive is often used to express […] doubt, especially with adverbs, in contrast with the indicative which expresses greater confidence: Tal vez venga mañana, Tal vez vendrá mañana (Palmer 1986: 66; cf. also Haverkate 2002: 35-37).

Furthermore, as indicated above, the analysis should go beyond the sentence level due to the context-sensitiveness of this study. Finally, in the analysis of individual items an attempt will be made to determine which meaning aspects are encoded and which ones are contributed by the context. Is it really possible – in the realm of the use of verbs of cognitive attitude and modal adverbs – to differentiate between evidentiality and epistemic modality? Dendale/Tasmowski (2001: 345) do not state without reason that in several languages certain linguistic expressions are used to express both linguistic categories, although not (always) simultaneously. These expressions then blur the conceptual distinction (Squartini 2004: 874). The chapter’s main contribution is two-fold: it describes the evidential and epistemic meaning aspects of Spanish verbs of cognitive attitude and modal adverbs in particular contexts and it highlights the importance of taking the context into account when analysing the meanings of these linguistic elements. Working with naturally occurring data, this study intends to demonstrate that the context shows which meaning aspects of a particular lexical item are encoded and which ones are contributed.

5.1 The evidential use of Spanish modal adverbs In her article for the Gramática Descriptiva Kovacci (1999) distinguishes the Spanish modal adverbs in two main groups which, in turn, are distinguished in two subgroups: Los adverbios del modus conforman dos grupos: 1) los relacionados con la modalidad como a) actitud del hablante frente al dictum (aseverativa, dubitativa,

245 etc.) [...], y b) el valor de verdad del dictum [...]; 2) los relacionados con la actitud del emisor frente a la enunciación: a) su propia disposición como hablante [...] y b) su interpretación del código empleado [...] (Kovacci 1999: 755).119

Modal adverbs such as probablemente, seguramente or tal vez are said to belong to group (1a) (Kovacci 1999: 755). These adverbs – as posiblemente, difícilmente or quizá(s) – are said to express doubt; in this connection, Kovacci speaks about “modalidad dubitativa” (1999: 755). That is why she matches the different adverbs of group (1a) with different points of the ‘scale of doubt’: Desde el punto de vista semántico todos conforman una escala continua de duda, desde su grado máximo, que se aproxima a la negación (difícilmente) hasta la aproximación a la certeza (seguramente) (Kovacci 1999: 755).

Unfortunately, Kovacci does not consider the pragmatic use of the adverbs. The adverbs are only correctly applied to the different points of the ‘scale of doubt’ from a semantic point of view. Reyes (1996; cf. also Haßler 2005), for example, has shown that speakers often use evidentemente if they speak of a certain state of affairs that is – literally – not evident: “[...] cuando el evidencial indica una inferencia, la afirmación se restringe aunque los adverbios utilizados parezcan, contrariamente, reforzarla (es el caso de evidentemente, sin duda, etc.)” (Reyes 1996: 31). The present study will also show that seguramente is especially used if the speaker is not certain. The “adverbios restrictivos del valor de verdad de la aserción” represent one part of group (1b) and comprise adverbs like supuestamente, presuntamente, presumiblemente, aparentemente, virtualmente, prácticamente or the adverb verosímilmente (Kovacci 1999: 758). Such an adverb is used if the speaker wants to restrict the truth value of his proposition. The “reforzadores del valor de verdad de la aserción”, in contrast, are said to reinforce the truth value of a certain proposition. So according to Kovacci (1999: 760), adverbs like “indudablemente, indiscutiblemente, incuestionablemente, innegablemente, ciertamente, verdaderamente, evidentemente, obviamente, y locuciones como sin duda, en verdad, en realidad” represent the second part of group (1b). Aparentemente and evidentemente – belonging to different subgroups of group (1b) – are (more or less) commonly acknowledged to convey an evidential meaning, but Kovacci does not refer to the evidential use of certain adverbs. The inferential meaning of evidentemente, for instance, was already shown by Reyes (cf. also chapter 4.3): [...] el adverbio evidentemente, al igual que muchos otros semejantes, alude a la inferencia que ha hecho el hablante para llegar a esa conclusión, y revela que lo afirmado es el producto de una inferencia (Reyes 1996: 29). 119 The second group – including its subgroups – is not important for the present study.

246 Anyway, Kovacci’s (1999: 755, 758-760) remarks on the Spanish modal adverbs may be summarised by the following scheme. For the sake of clarity, I added a few adverbs to the rather technical terms: LOS ADVERBIOS RELACIONADOS CON LA MODALIDAD COMO

ACTITUD DEL HABLANTE

EL VALOR DE VERDAD

FRENTE AL DICTUM

FRENTE AL DICTUM

seguramente probablemente tal vez

Adverbios restrictivos del valor de verdad de la aserción

Adverbios reforzadores del valor de verdad de la aserción

aparentemente

evidentemente

Fig. 14: Kovacci’s kinds of adverbs (1999: 755, 758-760; my scheme)

Even if the evidential use of certain modal adverbs is not recognised in the Gramática descriptiva, other linguists do not only speak about an ‘evidential use’ of certain adverbs but even distinguish semantically between epistemic and evidential adverbs. In doing so, Cornillie (2010b) applies the general feature of qualifying a proposition to these adverbs, whereby the kind of qualification varies according to whether the adverb is epistemic or evidential: Epistemic and evidential adverbs are generally defined as invariable expressions that qualify (parts of) a proposition in terms of the degree of likelihood that the event expressed may be real (in the case of epistemic markers), or in terms of the reference to the information that leads to a proposition (in the case of evidentials) (Cornillie 2010b: 309).

In this chapter the use of the modal adverbs aparentemente, evidentemente, obviamente, probablemente, posiblemente, supuestamente and seguramente is analysed and it will be shown that several adverbs

247 […] express reference to knowledge and, by doing so, verbalize the evidential justification for the speech act: aparentemente […], evidentemente, obviamente, supuestamente […]. These expressions convey evidential values such as various types of inferences and hearsay (Cornillie 2010a: 300).

So on the one hand, it is to show that certain adverbs can be used evidentially. On the other hand, it is to test whether seguramente, probablemente and posiblemente may also be used to convey evidential values. Because of the use of probablemente in example (10) or (11) which indeed can be interpreted as being used inferentially (cf. chapter 2.1.2.1), more examples should be looked at. As posiblemente is – like probablemente – said to be only modal it should also be analysed regarding its inferential use. Seguramente is included because it is expected to convey an inferential reading as well. Hence, the chosen abverbs that are worth being further investigated comprise evidential as well as epistemic ones. When it is required, special attention is paid to the mood of the verbal form used in the utterance containing the adverb under question because [m]odal variation in sentences where the dubitative adverb precedes the verb has been generally explained in terms of the greater or lesser degree of uncertainty of the speaker concerning the truth value of the statement (Haverkate 2002: 35).

That means in certain cases a closer look at the use of the subjunctive seems to be necessary, which may underlie the function of the adverb it occurs with because […] in contemporary Spanish, the mood system is a device through which speakers can evaluate the information value of clauses. Further, because speakers may take an evaluative stance with respect to any proposition, the identification of the Subjunctive with syntactic subordination is breaking down (Lunn 1995: 429).

As will be shown below (and as was already shown by examples (100), (101) and (102) containing evidentemente), certain modal adverbs are sometimes used to express an inference, sometimes hearsay, sometimes visual evidence and sometimes their use is not the one nor the other but a mixture of inference and visual evidence. Therefore, it will be worked with the following subscript abbreviations: ‘[inf]’ for inference, ‘[vis]’ for visual evidence, ‘[hs]’ for hearsay120 and ‘[inf/vis]’ for a mixture of inference and visual evidence.

120 For the sake of simplicity in the present study a reported reading and a hearsay reading will not be differentiated between, whilst nevertheless being aware of the difference or rather the opposition known source vs. unknown source (cf. Olbertz 2007: 154).

248

5.1.1 aparentemente As the data show, aparentemente has four meanings, whereupon the context determines the (more or less) exact meaning. Firstly, the meanings ‘hearsay’, ‘visual evidence’ and ‘inference’ will be considered: (104) Los medios de comunicación rusos, sin embargo, se refieren a cuatro condenados por espionaje y traición al Estado y de ese supuesto parte, aparentemente[hs], el decreto de indulto firmado por el presidente Dmitri Medvédev sobre el que informó el viernes la página oficial del Kremlin. (El País 12/07/2010) (105) Hoy, el técnico de La Oranje se encuentra a un partido de hacer el Mundial perfecto y ganar todos los partidos desde que en 2008 comenzara la clasificación para Sudáfrica. Van Marwijk ha transformado al equipo en una cosechadora de resultados aparentemente[vis] infalible, cuya última prueba será la final del Mundial que disputará ante España. (El País 11/07/2010) (106) Vaz ha publicado La otra mujer (2003), Leña (2004) y Metástasis (2006), entre otros. La poeta se ha dedicado, además, a asuntos tan aparentemente[vis] contradictorios como la gimnasia rítmica, la filosofía -es licenciada por la Universidad de Sevilla-, el periodismo y las artes plásticas. (El País 04/06/2010) (107) Sin embargo, según las imágenes mostradas en Al Arabiya en una conexión en directo, el presidente iraní apareció en el campo deportivo donde dio un discurso ante la multitud. Estaba aparentemente[vis] bien y no se hizo mención al ataque ni por parte del mandatario ni la televisión. (El País 06/08/2010) (108) Con 87 años, sorda y aislada, no se habla con su única hija, Françoise, a causa de la denuncia que ésta interpuso contra su amigo, el fotógrafo François-Marie Banier. La millonaria apareció el viernes en televisión: aparentemente[inf/vis] lúcida a pesar de la sordera, se fue por la tangente a la hora de responder sobre su presunto delito de evasión fiscal. (El País 08/07/2010)

While example (104) obviously conveys a hearsay reading as contextually provided information shows, examples (105), (106), (107) and (108) contain the construction modal adverb + adjective (infalible ‘never failing/faultless’, contradictorio ‘contradictory’, bien ‘well’ and lúcido ‘lucid’). In these examples aparentemente is used to express visual evidence because the characteristics described by the adjectives are visually perceptible, although they may be accompanied with an inferential moment as example (108) demonstrates: the speaker infers from visual evidence – from the appearance of the woman – that she is still a lucid person. The following example could be analysed in similar terms as example (108): (109) En sus nuevas producciones, acompañado por la soberbia banda que le asiste en vivo, Wilson sigue como siempre (la expresión algo perdida, pero

249 aparentemente[inf/vis] feliz), creando música fascinante con la que evadirnos de la gris realidad. (El País 20/09/2010)

In example (109) the expression of the person’s face is described as ‘somehow lost but apparently happy’ so that the person’s ‘apparent happiness’ is the output of an inferential process the speaker/journalist had gone through when the visual evidence, the expression, was cognitively processed. It may therefore be reasonable to argue that although an adverb is primarily used to indicate visual evidence, it is never exclusively used to indicate visual evidence. Visually perceived information is always cognitively processed and to a certain extent interpreted (cf. chapter 2.1.3), as Volkmann also explains: Sinneswahrnehmung bleibt nicht nur auf die bloße Aufnahme von Reizen beschränkt, sondern führt auch zur Interpretation des Reizes, Umwandlung in Wissen und dessen Weiterverarbeitung. […] Der Übergang von Interpretation des Reizes, über Verarbeitung bis zur Informationsschaffung ist fließend (Volkmann 2005: 105).

Cornillie (2010b: 315) has shown that aparentemente is used to perform not only the inferential reading but also hearsay. This may be confirmed by example (104), whereby the information of an explicit web page is reported so that the use, strictly speaking, performs the reported reading. It is usually differentiated between “cases in which the context allows for the source of the proposition to be identified, and those in which this is not the case, using the labels ‘reported’ and ‘hearsay’ […], respectively” (Olbertz 2007: 154). Example (104) contains references to possible information sources (underlined phrases) that may go hand in hand with aparentemente.121 The following examples illustrate that aparentemente may indeed be used to indicate reported information: (110) “Tenemos una mujer residente en el condado de Collier que ha sido infectada por el cólera en Haití”, confirmó en un correo electrónico el portavoz del Departamento de Epidemiología de Florida, Rob Hayes. Aparentemente[hs], la mujer desarrolló los síntomas días después de regresar de Haití. Las pruebas de laboratorio fueron concluyentes y la hospitalizaron aislada de los demás pacientes. Según fuentes médicas está en buenas condiciones. (El Mundo 20/11/2010) (111) Obtuvo un rescate pecuniario de 3,7 millones de euros, según algunos diarios canadienses, y la excarcelación, en abril de 2009, de cuatro reos de la prisión de Kati (Malí). Belmokhtar quedó, aparentemente[hs], contento de la labor de Chafi, el consejero presidencial. Por eso volvió a solicitar sus servicios cuando capturó a los tres catalanes. (El País 10/08/2010)

121 I admit that it could be argued that ‘the official Kremlin web page’ (example 104) is also visually perceivable.

250 (112) Cuatro de los heridos se encontraban esta mañana en estado grave. La Policía desconoce las causas del tiroteo, que aparentemente[hs] se habría producido sin ninguna discusión previa, informaron los medios locales. (El País 04/07/2010)

The underlined phrases in each text passage which represent information sources for the respective state of affairs and which appear together with aparentemente illustrate that [p] can be attributed to the source mentioned before or afterwards and that aparentemente is ‘apparently’ used to express that [p] belongs to another voice – and not to the one of the journalist. The following instance of aparentemente is considered ambivalent with respect to its reading: it may be said to indicate hearsay and it may be said to indicate visual evidence which was formerly perceived by the journalist: (113) [...] deambulan todo tipo de vendedores de baratijas que intentan ganar su sustento a cuenta de los que hacen fila frente a la comisaría y la alcaldía en busca de su documentación. Pero las filas son enormes, muchos no están seguros de poder votar. Aparentemente[hs/vis], miles de haitianos han dejado para el último momento la recogida de los documentos. No están impedidos de votar, de hecho si están haciendo fila es porque consultarán los listados colocados en enormes murales [...] (El Mundo 29/11/2010)

Even though there is no further information source provided which aparentemente may refer to, a hearsay reading seems highly likely and the journalist obtained the information from the government agency, for instance. The other possible explanation is that it is used to indicate visual evidence. The journalist may use aparentemente to indicate what he himself had visually perceived. So this use is ambiguous. What is obvious from the above listed examples is that if aparentemente conveys a hearsay reading it usually has scope over a clause (cf. example 112) or a whole sentence (cf. examples 104, 110). Aparentemente usually has smaller scope if it has the meaning ‘visual evidence’, especially when it is used to qualify (only) an adjective as in example (106) or (108), for instance. The scope of the adverb is equally small if it is used to convey the introductorily mentioned fourth meaning, viz. ‘by all appearances/indications’ – whereby the appearances do not have to be visually perceptible – or ‘seemingly’. While the meanings ‘by all appearances’ and ‘by all indications’ could be described in terms of weak inference, the meaning ‘seemingly’ expresses an ‘even weaker inference’, more precisely, an assumption where the modal component is more prominent than the inferential one (cf. Boye 2006: 66). So in contrast to example (108) or (109), where the speaker infers that [p] on the basis of visual evidence, that is, where visual evidence and inference go hand in hand, the following examples contain aparentemente conveying the inferential meaning ‘by all appearances’, ‘by all indications’ or (simply) ‘seemingly’. Hence, the

251 uses may be interpreted as conveying an inferential reading without any explicit reference to the kind of evidence or as expressing an assumption in the sense of ‘something is seemingly the case’. These meanings are hardly distinguishable, but aparentemente usually has one of these meanings if the adverb’s scope is a relatively small one and if visual evidence can be excluded. In the following two examples aparentemente (only) has scope over adjectives (incompatible and inocuo). The adjectives that are qualified by aparentemente in examples (114) and (115) denote concepts that are abstract (cf. the underlinings). This leads to the assumption that in these examples aparentemente means ‘by all appearances/indications’ or ‘seemingly’: (114) [...] de comprometerse con Portugal, debería preparar el partido contra el Deportivo, el 3 de octubre, al tiempo que configura la lista de los 22 portugueses que se llevará a los encuentros de clasificación. Dos tareas aparentemente incompatibles. En el club creen que el malestar del técnico se debe, principalmente, a su deseo de que el club le negara su derecho a ir con Portugal de manera más explícita. (El País 23/09/2010) (115) [...] también sea capaz de reactivar en pacientes con estrés postraumático ciertos recuerdos sin ningún relación con dicha situación, alterando las asociaciones normales entre recuerdos de uno y otro signo. De manera que incluso memorias aparentemente inocuas, o situaciones del día a día, pueden ayudar a rememorar el trauma. (El Mundo 24/12/2010)

In the following examples aparentemente is either used to indicate visual evidence as the adverb has scope over one or various adjectives that denote characteristics which are visually perceivable or it means ‘by all appearances/indications’ or ‘seemingly’. It seems to be impossible to determine the exact meaning: (116) Argentina, Brasil, Italia e Inglaterra perecieron bajo el yugo de estos individuos aparentemente frágiles, omnipresentes, impredecibles, difíciles de clasificar, tan incómodos para los rivales como problemáticos para entrenadores que buscan respuestas simples. (El País 19/09/2010) (117) “¿Cómo nos podemos quitar ese miedo? Es imposible”, dijo a este diario, aparentemente sereno, minutos antes de exponer su clase. Tras afirmar que no es una persona insegura, se posicionó sobre la teoría de que el miedo tiene una relación directa con la responsabilidad por estar frente al público. (El País 18/09/2010) (118) “[...] El Gobierno está siendo portavoz de la parte fuerte y aparentemente agresora, que es Marruecos, y no está prestando ninguna voz y atención a la parte débil, que es el Sahara”, apostilló. Por último, Juan Diego Botto calificó de “desliz” las manifestaciones de González-Sinde. (El Mundo 22/11/2010) (119) Las víctimas tomaron fotos mientras despejaban un área para refugiarse del fuego (abajo a la derecha), “convencidos de que no pasaría nada”. Aparentemente

252 tranquilo, uno de los fallecidos se autoretrató (arriba a la derecha) segundos antes de la tragedia. (El País 26/07/2010)

In the following two examples aparentemente has scope over the construction ser + adjective. Here, the adverb may either mean ‘by all appearances/indications’ or ‘seemingly’, whereby it is considered highly likely that aparentemente in example (120) means ‘seemingly’ rather than ‘by all appearances/indications’ because it appears in a negative context. If something seems to be for free but it is not, it is only ‘seemingly for free’: (120) En el fondo, esta frase sugiere que, a determinados niveles -un ejemplo de los más recurrentes es el de la industria farmacéutica-, no existen regalos desinteresados y que algo que aparentemente es gratuito, al final se paga de alguna forma. (El País 01/05/2010) (121) Y como presumiblemente harán ahora. Los dos grandes partidos se escudan en la búsqueda del pluralismo, lo que aparentemente es loable. También podrían alegar que es una particular manera de luchar contra el sedentarismo, tan malo para la obesidad y las enfermedades coronarias. (El País 26/07/2010)

Especially example (122) is interesting in so far as aparentemente in the clause al menos aparentemente ‘apparently at least’ is posposed to [p] and its use seems to be ironic: even if it is not [p], [p] at least seems to be: (122) En la pugna jurídica y política, el Estado portugués está notoriamente solo frente al comprador español, la mayoría de accionistas de PT y la Unión Europea. Solo, pero firme, al menos aparentemente. (El País 04/07/2010)

In summary, while ‘inference’ and ‘visual evidence’ are assumed to represent meaning aspects that are encoded by aparentemente, the ‘reported’ meaning is considered to be contributed by the context as the adverb – if used to convey reported evidence – needs to appear with further references to information sources such as the underlined phrases in the examples. One could conclude that if aparentemente does not appear together with other references to information sources the meaning conveyed is either visual evidence or inferential evidence or both. Even though it is possible sometimes to apply a hearsay reading to the adverb that is obviously not used with a further reference to an information source (cf. example 113), it is not a reading that can doubtlessly be verified with the help of the adverb’s contextual surroundings. If aparentemente means ‘by all appearances/indications’ or ‘seemingly’ it is a meaning aspect that is verified by the context but in a special way: if the adverb is not used to convey one of the other readings, it has to be one of the last mentioned, whereby it is sometimes not possible to determine whether aparentemente is used to indicate visual evidence or used to express ‘by all appearances/indications’ or ‘seemingly’, if the adverb qualifies adjectives which describe characteristics that are visually per-

253 ceivable. The meanings ‘by all appearances’ and ‘by all indications’ could be described in terms of weak inference; the meaning ‘seemingly’ seems to express an ‘even weaker inference’, which may be described in terms of assumption because the modal component is more prominent than the inferential one. Hence, even an ‘apparently’ evidential adverb such as aparentemente may be used to convey a modal reading (cf. Aikhenvald’s (2004) remarks on evidentials which may have extensions related to epistemic modality in chapter 2.1.1).

5.1.2 evidentemente The adverb evidentemente seems to have similar functions as aparentemente, whereby it is not used to indicate reported/hearsay evidence: (123) [...] confrontación por un proyecto integrador, en el que todos, incluidos los nacionalismos democráticos, puedan sentirse comprometidos [...] con el desarrollo del Estado democrático y corresponsables con el proyecto europeo de la España plural. Pero, evidentemente[inf], la tarea no es fácil. Entre otras razones porque ese proyecto federal implica una importante reforma constitucional. (El País 23/07/2010) (124) [...] en el hecho de que el porcentaje de españoles capaz de comunicarse fluidamente en inglés sea muy inferior al de la población nórdica, la germánica o incluso la de algunos países eslavos. Los motivos son de tipo cultural, educativo y, evidentemente[inf/vis], también lingüísticos. Comencemos por estos últimos. El español, y nuestras otras lenguas románicas (a las que habría que añadir también el vasco), no pertenecen a la familia lingüística del inglés, que es el grupo de lenguas germánicas [...] (El País 21/07/2010) (125) “[...] Nosotros recomendamos dedicarle a las tareas de 30 a 40 minutos diarios, de lunes a viernes, como en el calendario escolar”, explica Ortiz. La franja de edad de estas tareas incluye incluso a los pequeños de tres años que, evidentemente[inf/vis], aprenden a partir de juegos. La mejor hora es después del desayuno, cuando el niño está descansando intelectualmente y cuenta con todo el día por delante. (El País 09/06/2010) (126) Este fenómeno sería como un subproducto que fabricaba la cultura burguesa para anestesiar al burdo proletariado. Todos los medios de comunicación de masas desarrollaban la función de hipnotizar a las masas. El fútbol, evidentemente[vis], en uno de los primeros puestos. Pero ¿qué ocurre, sin embargo, ahora cuando la Wikipedia es el centro del saber [...] (El País 21/07/2010)

Evidentemente in (123) is clearly used inferentially as the context indicates: […] la tarea no es fácil. Entre otras razones porque…‘this task is not easy. Among other reasons because...’. The speaker infers the state of affairs (the difficulty or non-easiness) and justifies his utterance giving the reason (porque). Reyes (1996: 28-29) has already shown that evidentemente is often used when a state

254 of affairs is not visually perceptible. The use of the adverb in example (124) and (125) can be interpreted as both visual evidence and inference: the fact that Spanish is a Romance language and English a Germanic one is evident or obvious but it has to be inferred that this is one reason to explain why Spanish people often have difficulties with speaking English (example 124). The use of evidentemente in text passage (125) can be explained in similar terms: Ms or Mr Ortiz surely monitored children when they learned from playing, but she/he had to infer that what the children did was learning (by playing). In (126) we clearly deal with a use of evidentemente to indicate visual evidence because that media use football to hypnotise people represents a phenomenon that is visually perceptible. For the English equivalent, Simon-Vandenbergen/Aijmer found out that evidently also has a ‘hearsay’ meaning (2007: 161). The ‘hearsay’ meaning for evidentemente could not be verified in the course of this study, as I did not come across uses that could be compared with the use of aparentemente to mark reported information. The meaning of evidentemente could be paraphrased – despite being translated as ‘evidently’ – as ‘obviously’, ‘seems to be’, ‘something is visible’, ‘evident from certain indications’ or even ‘evident from the situation I can observe’, whereby in every case it is not easy to determine which meaning is the more/most prominent one as the following examples illustrate: (127) “[...] Una persona puede cobrar allí mil o dos mil yuanes al mes, que es lo mismo que cobrar 200 euros. Evidentemente, aquí ganan mucho más”. Los negocios chinos, cuenta David Zhou, sufrieron el año pasado. En parte, por la caída de las importaciones de productos de su país, una de las principales actividades del colectivo en Valencia. (El País 15/09/2010)

In example (127) it is highly likely that evidentemente means ‘evident from the situation I can observe’ because the speaker refers to a fact that obviously surrounds him as he uses aquí ‘here’. In example (128), in contrast, the exact meaning seems even more difficult to determine (but it has to be one of the meanings mentioned above), whereby in example (129) the meaning of evidentemente could be translated as ‘logically’, thus meaning that ‘it is obviously difficult to see democracy if there is so much poverty and disparity’ (evidentemente es difícil ver la democracia cuando hay tanta pobreza y desigualdad). (128) Se trata de una confesión personal cuyo personaje central mantiene significativas semejanzas con la autora real. Aunque el nombre de la protagonista, Antonia, no es, evidentemente, el mismo de la autora, creo que podemos hablar de una modalidad novelística que se ha denominado autoficción, mezcla de autobiografía y ficción que permite trasladar información del autor al personaje y viceversa. (El País 13/09/2010)

255 (129) P. Con 1.100 millones de habitantes, la India se ha ganado el título de la mayor democracia del mundo. Un tanto sarcástico en un mundo de castas. R. Son las paradojas en las que vive la India porque evidentemente es difícil ver la democracia cuando hay tanta pobreza y desigualdad. La historia es cómo lograr salir de la miseria sin optar por recetas de monoteísmo político o religioso. (El País 01/08/2010)

The following use of evidentemente can be compared to example (129). The adverb is used to qualify [p] (Los agricultores más eficientes reciben mejores subsidios ‘the most efficient farmers receive more subsidies’). Indeed, this may be seen as being paradoxical so that it is ‘obviously (logically) unfair’ (no es justo): (130) [...] y las ayudas a los productores de los países más veteranos, que se calculan con un método que tiene en cuenta datos históricos y beneficia a los que más produjeron entre 2000 y 2002. Los agricultores más eficientes reciben mejores subsidios y eso “evidentemente, no es justo”, han admitido fuentes europeas consultadas [...] (El Mundo 20/11/2010)

The use of evidentemente in the following two examples (131 and 132) is not easy to determine: even though in both text passages visually perceivable subjects/objects are referred to, viz. cárceles ‘prisons’ and andaluces ‘Andalusians’, the respective speaker refers to a fact with reference to these subjects/objects which is ‘obvious’, ‘evident from certain indications’ or even ‘evident from the situation he can observe’: (131) Estos delincuentes corrompen a los custodios o los amenazan, todos tienen seres queridos que pueden ser sus víctimas, evidentemente existen cárceles de alta seguridad para los narcotraficantes, pero hay que diferenciar entre los sicarios que no son más que asesinos y los narcotraficantes. (El País 13/09/2010) (132) Sobre las polémicas declaraciones en las que dijo que en Andalucía no paga impuestos “ni dios”, el líder de ERC ha dicho: “Evidentemente hay andaluces que pagan impuestos, como hay catalanes que no pagan”, pero ha remarcado que él defiende su tesis con cifras desde hace diez años, a pesar de la “opacidad absoluta” del Ministerio de Hacienda. (El Mundo 20/11/2010) (133) [...] parece un tanto sesgado al no considerar los correos en un cliente de correo como outlook, yo tengo 6 cuentas en outlook y de ellas 4 en la blackberry y no saldría en el estudio. evidentemente al aparecer los sistemas de comunicación en tiempo real debe producirse una caida del correo, pero de ahí a afirmar que el correo está muerto... feliz navidad. (El Mundo 25/12/2010)

Example (133) is from a chat conversation. The speaker/writer describes a fact that could be termed ‘world knowledge’ as it is an obvious fact that can be observed by (nearly) everyone; it is thus ‘evident from the situation’: with the appearance of the modern communication systems the letter has died out.

256 The following examples contain evidentemente, where the adverb is part of a proposition that contains information about the speaker himself. In other words, the speaker uses evidentemente to evidentially mark information about himself. While the utterance in example (134) may be said to be only ‘unfortunately formulated’ because evidentemente can only refer to the second part of the clause, that is, to Cobo lo sabe ‘Cobo knows it’ (and not also to yo no soy del PP ‘I am not from PP’), evidentemente in example (135) clearly refers to estoy en contra de una intervención del ejército español ‘I am against an intervening of the Spanish army’ because the preceding clause represents an insertion: (134) “Le conozco hace muchísimos años. Hacemos deporte juntos y somos seguidores de José Tomás. Evidentemente, yo no soy del PP y Cobo lo sabe. Pero casi nunca hablamos de política. También tengo amistad con Toño Alonso [José Antonio Alonso portavoz socialista en el Congreso]”. (El País 03/08/2010) (135) [...] documéntate un poco sobre el ejército español y sobre el marroquí y verás que la diferencia es abrumadora a favor de España, sencillamente Marruecos no tiene marina de guerra, y es fundamental. Evidentemente, y aunque estoy a favor del Sahara, estoy en contra de una intervención del ejército español [...] (El Mundo 13/11/2010)

Does the speaker infer from his own thinking and attitude that he ‘is evidently against an intervening of the Spanish army’? In my opinion, here evidentemente has a discursive function: it seems to summarise and refer back to what the speaker has already said. So ‘evidently – and taking into account what he has already said – he is against an intervention of the Spanish army’. Evidentemente in example (134) could possibly be said to fulfil a very similar discursive function. However, in example (135) evidentemente may also serve an appealing function; it may appeal to the imaginative power and inferential process of the other speaker/writer of the chat forum. Cornillie (2010b) draws attention to the function(s) that certain adverbs fulfil with respect to discourse planning. About evidentemente he says: […] the evidential adverb [evidentemente] mainly involves inferences from reasoning. Yet, the reasoning is not so much triggered by sensorial, visual or auditive stimuli but is part of the discourse planning. It can be argued that there is a semantic shift from a purely evidential reading (i.e. something that can be clearly perceived) to a discourse related reading (i.e. something that is clear according to discourse patterns) (Cornillie 2010b: 327).

In order to analyse discourse and interactional functions of epistemic/evidential adverbs, the study of speech data retrieved from dialogue situations seems to be indispensable. That is why the use of evidentemente in examples (134) and (135) is only assumed to fulfil a further function that goes beyond the function of adding an evidential value to the propositions the adverbs go with. Simon-Van-

257 denbergen/Aijmer (2007) also adopt an interactive approach to modal adverbs. In so doing, they take into consideration the dialogical nature of all verbal communication. Their study reveals that modal adverbs are not only used to express the speaker’s commitment to the truth value of the proposition but also to serve interactive, conversational functions. And these functions depend on the context or rather on the dialogical situation. The following instances of evidentemente obviously convey an inferential reading. The inference is drawn from formerly made experiences (example 137) or from judging a current situation (example 136): (136) #13 Anónimo 14.Nov.2010 | 12:08 PONCISTA DICE: Evidentemente Zapatero fue un error,.... no hay la menor duda, y tampoco cabe la menos duda que el Partido Popular va arrasar con sus votos y menos mal, porque esta situación solo la puede solucionar un gobierno del Partido Popular!!!! (El Mundo 17/11/2010) (137) Además, se ha repetido una costumbre que se debe cambiar. Evidentemente, la celebración de una cumbre del G-8 justo antes de una cumbre del G-20, como ocurrió en Canadá el pasado mes de junio, solo sirve para prolongar el mantenimiento de dos clubes por separado, cosa que resulta insostenible. (El País 13/09/2010)

Concerning the question which meaning aspect could be assumed to be encoded by evidentemente and which one could be contributed by the context, ‘visual evidence’ appears to be encoded and ‘inferential evidence’ contributed by the context because the inference is often based on visual evidence or at least on something that is perceivable like the situation described in example (136). However, evidentemente – as well as aparentemente – “can also denote conclusions which are not based on visual perception” (Haßler 2010d: 99; cf. Haßler 2004). Evidentemente may have scope over a whole sentence (examples 132 and 137) or, for instance over an adjective only (cf. the underlining in example 124). This means that the adverb may function as a sentence adverb as well as it may be used to qualify only one part of [p]. If used as a sentence adverb, it may be separated by a comma (example 137) or not (example 132). ‘Evidently’ – evident from the examples above, in the majority of cases evidentemente is used as a sentence adverb.

5.1.3 obviamente “Obviously means ‘as evidence shows’. Hence, obviously, even in its extended evidential meaning of ‘as knowledge of the world shows’ […] is evidential, meaning ‘as the evidence shows’” (Simon-Vandenbergen/Aijmer 2007: 220). But sometimes ‘obvious evidence’ has to be interpreted by the speaker, and then

258 it is also used inferentially, although these are not two different meanings or functions of obviamente. It seems really difficult to differentiate: (138) Sentado en el sillón de su despacho en el centro de Washington, Moreno, colombiano de 57 años, está exultante de felicidad el martes por la tarde, minutos después de haber sido renovado en su cargo para un segundo periodo de cinco años. Su alegría, obviamente, responde al reconocimiento por el éxito de su gestión. Pero él la justifica por “el honor de ocupar este puesto en un momento particularmente positivo para este continente”. (El País 10/07/2010)

In example (138) the speaker/writer tries to explain the reason for Mr. Moreno’s happiness. The speaker/writer could see that the person under question is happy but the reason for Moreno’s happiness is an inference, as the context shows. (139) Aira recordó sus inicios como traductor de best-sellers (“me decanté por ellos cuando descubrí que pagaban lo mismo por traducir buena que mala literatura; y que la mala resultaba obviamente más fácil”) y cuánto estos influyeron en su inclasificable estilo. (El País 13/07/2010) (140) Si hay un pueblo que carece de sentido de Estado es el italiano. El único periodo en el que se produjo un breve y ficticio sentido de Estado fue bajo el fascismo [...]. Obviamente, las dictaduras son algo terrible, que eliminan las libertades [...], pero representan una fuerza unitaria cohesionadora [...] (El País 15/07/2010)

Both adverbs seem to indicate what was already quoted: ‘as evidence shows’ or ‘as knowledge of the world shows’, while the first would apply to example (139), the latter applies to example (140). In example (139) a translator speaks about his personal experience that badly written literature can be translated obviously with less effort. The fact that is expressed by example (140) is indeed something that knowledge of the world shows. Simon-Vandenbergen/Aijmer have shown for obviously that “it ‘inherently’ refer[s] to the fact that the information can be gathered from available evidence” (2007: 38), that it “convey[s] certainty based on some kind of evidence which is there for everyone to see” (2007: 147), that its meaning ‘obvious to everyone’ makes it more “intersubjective rather than subjective” (2007: 150) and that it is consequently used to present information as common ground so that its function “is bonding: the creation of shared attitudes, a common world”, whereby the speaker conveys ‘I know that you also know this’ (2007: 154). These characteristics are all applicable to the examples cited above. The examples that follow will show that obviamente is used to mark very logical and commonly known facts as well as to indicate that [p] is an inference which cannot always be termed ‘intersubjectively comprehensible’. In example (141) obviamente is used in the sense of ‘as knowledge of the world shows’, more precisely, to mark [p]

259 as the most logical fact at all: a dead person will not be alive again. This ‘not being here’ (ella no va estar aquí) is marked by obviamente: (141) [...] que perteneció a su madre, la fallecida Diana de Gales, que entregó a su novia. “Lo llevé en la mochila durante tres semanas, me lo llevaba a todas partes”, dijo el novio. “Es el anillo de compromiso de mi madre y pensé que sería bonito porque obviamente ella no va estar aquí para compartir toda esta diversión y emoción y ésta ha sido mi manera de tenerla cerca de todo lo que está ocurriendo”, dijo Guillermo sobre la elección de la joya. (El Mundo 18/11/2010)

The following example contains obviamente whose use is similar to the one in example (141): it is simply commonly known that the average salary of a German ‘has nothing to do’ with the average salary of a Spanish employee, and furthermore, if they are compared, it is obvious: (142) Claro que también habría que pedirles a los políticos españoles una responsabilidad, compromiso y capacidad para gestionar el país parecidos a los que de seguro le exigen a los políticos alemanes. Y a lo que va 232, el sueldo medio de un alemán obviamente no tiene nada que ver con el de un español, y además seguro que está mucho más ajustado a la realidad alemana e incluso de su clase política. (El Mundo 27/10/2010)

Example (143) is slightly ambiguous as the context does not provide sufficient information that would allow one to disambiguate the adverb’s use: the deictic demonstrative pronoun esta ‘this’ indicates that, while uttering [p], the speaker must stand in front of the house itself or in front of a picture of the house which is ‘obviously very expensive’ (obviamente carísima). On the one hand, the house could display a sign showing the price; on the other hand, the house could simply look very expensive and the architecture, for instance, causes the speaker to infer that the house must be very expensive: (143) “Un día, paseaba por las Ramblas, me topé con este palacete del siglo XVIII y sentí que me llamaba” -tuve que adquirirlo-, “Encontramos esta casa justo en el centro de Madrid” -de 500 metros cuadrados y obviamente carísima-. Como si los pobres y mileuristas vivieran en pisos pequeños, oscuros, periféricos o pobremente decorados porque tienen mal gusto o no han sabido “encontrar” esas joyas perdidas... (El País 16/10/2010)

As already indicated with regard to subjectivity, even visual evidence (in that case, the sign showing the price of the house) is to a certain extent inferential in nature: the price is cognitively processed and then evaluated as being a high price. So the fact that the house is ‘obviously very expensive’ is inferred from the numbers. However, the visual component is more prominent than the inferential one. By contrast, the cognitive effort to infer that the house must be expensive is higher when based on the appearance of the house itself; in this case, the inferential component is more prominent.

260 The four examples that follow contain an inferential use of obviamente. Example (144) illustrates that its use is not always more intersubjective than subjective and that it does not always convey certainty based on evidence that could be seen by everyone. Instead, the speaker (subjectively) infers which age his chat-partner refers to when speaking of ‘other times’ (otros tiempos): (144) Lo importante es la situación actual, en la que España tiene reponsabilidad jurídica y peso político para proteger vidas que están siendo atropelladas por reclamar derechos que la ONU reconoce (el referéndum) y Marruecos no otorga. a “otros tiempos” obviamente te refieres a que España era una potencia y ahora no lo es. Fuera lo que fuera, no le da derecho, ni a España ni a ningun otro pais, a ocupar territorios que no les pertence. (El Mundo 17/11/2010)

The use of obviamente in the following example is also not considered intersubjectively comprehensible. The speaker judges that the person he talks about ‘was obviously affected by the critics during the World Cup’ (Obviamente le afectaron las críticas que tuvo durante el Mundial) and then utters another supposition that is introduced by por supuesto: (145) Cuestionado por las causas que han impedido estar a Torres en plena forma, Hodgson respondió: “No lo sé. Estuvo muy bajo, muy bajo cuando regresó del Mundial. Obviamente le afectaron las críticas que tuvo durante el Mundial y por supuesto todo lo que ha rodeado al club y al equipo, a mí, y a todos los que estamos relacionados con el Liverpool en el inicio de la temporada [...]” (El Mundo 26/10/2010)

The last two uses of obviamente are highly likely more intersubjective than subjective because the adverb marks an inference or judgement that is based on the evaluation of the current situation – based on the observation of the current situation: (146) [...] los mitos a menudo nos dice más sobre las inquietudes y deseos de aquellos que participan en este cometido que de la naturaleza del propio mito. En este sentido, la actual multifacética entente cordial en torno a la memoria del primer lehendakari obviamente no ha eliminado las discrepancias de interpretación respecto a su actuación y su legado. Pero sí puede ser entendida como una potente señal enviada por una sociedad que, a punto de superar la lacra del terrorismo [...] (El País 17/10/2010) (147) [...] economía financiera, inversión y consumo, gasto público y estímulos estatales, ajustes internos y globalización, problemas urgentes y mejoras a largo plazo, amén del consiguiente y abultado desempleo que es la antinomia del Estado de bienestar. Obviamente, hay que restablecer el crecimiento, pero ¿cómo? En la economía de mercado, es este el que impera, es decir, en otras palabras, hay que proteger al capital, dar preferencia a lo privado sobre lo público [...] (El País 18/07/2010)

261 Even though it could be illustrated that obviamente is – besides its intersubjective use – also used to mark subjective statements, it is concluded that the meaning of obviamente is not as much context-dependent as the meaning of aparentemente, for instance. Obviamente nearly always seems to express ‘as evidence/knowledge of the world shows’ or ‘as observable evidence shows’. If the adverb is based on observable evidence, the use is consequently partly inferential. In sum, obviamente clearly is an evidential adverb. Even if no contextual information about the kind of evidence is provided, it is evidential. Haßler (2010d: 98), for instance, lists an example where a speaker uses the Italian adverb ovviamente to merely mark the conclusion as such, hence, where “[n]o explicit reference is made to what was behind it or what it was based on” (Haßler 2010d: 99; cf. also Haßler 2003). The adverb’s scope varies: for instance, in text passage (140) it functions as a sentence adverb and in example (143), by contrast, it qualifies an adjective only (see the underlining of carísima ‘very expensive’).

5.1.4 posiblemente Posiblemente expresses a possibility (of a state of affairs); its nearest English equivalent is ‘possibly’. Probablemente ‘probably’, on the other hand, expresses a likelihood (of a state of affairs). And a likelihood of a state of affairs is more probable than a possibility. “Possibly and probably are not antonyms, but they are at different positions in a gradience of polarity” (Greenbaum 1969: 151). So we could conclude that a probability expresses a greater possibility, and that a possibility is less probable than a probability: (148) La localidad de Rothbury, un pequeño pueblo del noreste de Inglaterra cuya población no llega a los 2.000 habitantes, ha vivido esta semana posiblemente algunos de los días más intensos de su historia. Desde el martes, decenas de agentes británicos, apoyados por helicópteros, acordonaban la localidad para intentar apresar a Raoul Moat, al que llevaban buscando desde el pasado sábado [...] (El País 12/07/2010) (149) [...] fue “como volver a casa, porque los tres venimos del mismo sitio, somos músicos de jazz, pero nunca habíamos tocado en acústico. El año que viene tenemos la idea de montar una gran fiesta, posiblemente en San Francisco, aprovechando que fue allí donde nos reunimos por primera vez. Esa ciudad tiene un sonido especial que viene de los tiempos del Fillmore y las bandas de rock de los sesenta”. (El País 12/07/2010) (150) Tras 45 minutos de maniobras, se confirmó el fallecimiento del paciente por parada cardiorespiratoria, posiblemente secundaria a golpe de calor al presentar el joven una temperatura corporal de 41 grados. La causa final de la muerte, tal y como

262 establece la normativa vigente, ha de ser dictaminada por el forense a quien se avisó para que certificara [...] (El País 10/07/2010) (151) El fotógrafo ve las cosas de manera diferente y habla de “libertad recobrada” y de toma de distancia con la primera parte de su carrera, pero sin renegar de ella en absoluto. Posiblemente dan fe de ello sus extraños trabajos jacksonianos y el fresco en homenaje a un África exangüe que ilustran estas páginas. (El País 13/07/2010)

Considering these examples, one may conclude that posiblemente is a pure epistemic adverb. None of the given contexts leads to the conclusion that the use of posiblemente is evidential. So the meaning aspect which is encoded by this adverb is clearly that a certain state of affairs ‘is/was possibly the case’. The only example where the evidential dimension may be present is text passage (150): it could be an inference from the situation and it may be even reasonable to think that the patient died because of the heat, as he had a body temperature of 41 degree. In that case the inferential meaning would be contributed by the context. But generally speaking, posiblemente expresses that [p] is possibly the case, often without referring to the reason(s) why the speaker assumes that it is possible that [p]. This is illustrated by the following examples: (152) Está previsto que a lo largo del día Ahmadineyad presida una reunión en la que hablará el líder de Hezbolá Hassan Nasrallah, posiblemente por video. A bordo del todoterreno, Ahmadineyad ha saludado a las personas que se agolpaban a ambos lados de la carretera y en los puentes, y le lanzaban flores mientras coreaban gritos de bienvenida. (El País 16/10/2010) (153) Pero parte es eso, parte. Y de ahí que desde la compañía se explique que, de alguna forma, ese encarecimiento acabará por afectar a las tarifas, tanto de gas, como posiblemente de la electricidad que pagan los usuarios. Frente a ese análisis está el que puede hacer Industria, con los acentos colocados en diferentes lugares respecto a la posición de Gas Natural. (El País 14/10/2010) (154) Hemos ampliado espacios democráticos y la democracia sólo arroja luz, participación y transparencia. La democracia en todos los niveles es mucho más barata que su ausencia. El Estado autonómico molesta a los centralistas de don Pelayo y posiblemente a los nacionalistas que obtenían privilegios en función de hechos diferenciales que hoy ya no son argumentos presupuestarios. (El País 14/10/2010) (155) ‘En el valle de Elah’ (2007): Tommy Lee Jones es un militar retirado que acude a una base militar para investigar la desaparición de su hijo. Una reflexión sobre la guerra y el patriotismo ideada por Paul Haggis, posiblemente el mejor guionista del Hollywood actual, que lleva al progenitor a replantearse muchas de sus convicciones. (El Mundo 20/11/2010) (156) Diversos medios informaban ayer de que Sauerland pretende aguantar hasta que los concejales lo depongan, posiblemente en octubre. Así conservaría sus

263 ingresos hasta 2014 y no echaría en saco roto su bonita jubilación. (El País 02/08/2010)

What all the examples above reveal is that posiblemente has a small scope and that it does not function as a sentence adverb, that is to say, it does not qualify a whole sentence but only parts of it (see the underlined phrases). However, posiblemente may function as a sentence adverb as example (151) shows, whereby it is not separated by a comma. “Nevertheless, it should be noticed that in Contemporary Spanish posiblemente is showing a growing tendency to select the subjunctive, which is true in particular of newspaper style” (Haverkate 2002: 36), whereby posiblemente + subjunctive is not only used by journalists but also by interviewees, whose answer has the form of direct quoted speech (cf. example (159), which represents a respuesta ‘answer’), even though not marked as such: (157) Posiblemente, la creación del Servicio sea un paso decisivo en la configuración de la Unión Europea como actor internacional. (El País 21/07/2010) (158) Esa misma fuente habla de la “fatiga” de los materiales. “No siempre las roturas se producen de forma inmediata”, explica. Y apunta a que posiblemente el “traqueteo continuado” de la atracción [...] haya hecho disminuir su resistencia. (El País 21/07/2010) (159) R[espuesta]- Cataluña es tierra de comercio y no sólo sabrá mercadearse, sino que posiblemente se convierta en uno de los competidores más fuertes de España. (El País 19/07/2010)

The speaker expresses an even lower committed epistemic evaluation that a certain state of affairs is the case when using posiblemente + subjunctive, as if he uses posiblemente + indicative. If the adverb is followed by the subjuntivo, the use seems to be purely epistemic: (160) También les recuerdo que el gastó en educación no llega a los 2 digitos del pib, y que las indemnizaciones de estos 33 hombres y los otros empleados de esta mina, serán aplazados hasta la eternidad y posiblemente no vean más que una migaja misera, mientras los acreedores de la misma se hagan ricos, por las nuevas vetas encontradas durante el millonario sondaje [...] (El País 16/10/2010) (161) Así lo ha explicado a The Daily Telegraph un portavoz de St. James Palace, la residencia del príncipe Carlos en Londres. “Tanto el Príncipe de Gales como la Reina posiblemente contribuyan a los costes de la boda. Será una contribución familiar, pero no se ha tomado ninguna decisión”, explicó. (El Mundo 20/11/2010) (162) A mitades de este año, el coche japonés ha recibido un nuevo motor, esta vez en diésel de 2.5 litros que rinde 190 caballos. Posiblemente el nuevo Murano adopte estos motores o derivados aunque por el momento no hay nada confirmado. (El Mundo 18/11/2010)

264 (163) Los que entendemos de fútbol vemos a Casillas desorientado, desmotivado y falto de concentración. Posiblemente la distracción en la banda esté tumbando al mito. (El Mundo 20/11/2010) (164) Aunque Gómez ha obtenido varios cientos de avales más que Jiménez, uno y otra han superado las 5.000 firmas, por lo que posiblemente el vencedor no obtenga una gran diferencia de sufragios. Gómez se presentó a las 10.40 en la agrupación socialista de Parla, municipio del sur de Madrid donde ha sido alcalde tres mandatos consecutivos. (El País 06/10/2010)

There is neither a hint in the conceptual meaning of the adverb itself nor in the contextual surroundings of posiblemente that could lead to the assumption that the adverb is (also) evidential. So posiblemente is a pure epistemic adverb. If the adverb conveys an evidential meaning, it is a meaning aspect that is contributed by the context (cf. also example 150): (165) Roddick se impuso con resolución en el primer set (3-6), pero en el segundo, García López le castigó con varios intercambios a los ángulos de la pista, y en especial con una soberbia dejada, a la que el estadounidense intentó llegar, y en la que posiblemente se causó la lesión. Con 2-2 Roddick pidió asistencia médica en la pista, y tras perder el juego siguiente decidió no continuar. (El País 16/10/2010)

The journalist reports on a tennis match he had observed and he seems to express an inference based on his observation. There must be a reason why the journalist assumes that the tennis player was hurt in that situation and not in another one. This analysis based on empirical data is also confirmed by Lunn: The Spanish Subjunctive is primarily expressive of epistemic modality. Numerous studies, following Bolinger (1968) and Terrell and Hooper (1974), have shown that the best synchronic description of the Spanish mood system links choice of the Indicative […] to assertion and choice of the Subjunctive […] to non-assertion (Lunn 1995: 429-430).

Nevertheless, it should be emphasised that, if used with an adverb such as posiblemente, even the indicative is not used to express an assertion. Indicative mood only signals that the possibility of [p] being the case is higher than if the possibility were expressed with the subjunctive. The Manual of the RAE explains that the choice of the subjunctive mood in connection with a modal adverb, forming the structure adverb + subjunctive, depends on whether the information is thematic or rhematic: Los adverbios de duda y de posibilidad, como quizá(s), tal vez, acaso, a {lo – la} mejor, posiblemente, probablemente, seguramente se caracterizan por inducir indicativo o subjuntivo dentro de su propia oración. El subjuntivo puede aparecer en estas construcciones si el adverbio precede al verbo y no está separado de él por una pausa. [...] Cuando ambos modos son posibles, la elección depende en gran medida

265 de la manera en que se interprete la estructura informativa de la oración. Se suele preferir el subjuntivo si la información introducida no es focal o no se presenta como nueva [...] (Manual 2010: 490).

It is to question that the opposition of rhematic vs. thematic information represents the only selection criterion. It is certainly right to associate the subjunctive, roughly speaking, with old information and the indicative with new information. The fact that information which the speaker shares or assumes to share with the hearer is marked by the subjunctive is also addressed by the Manual of the RAE: [...] la información que aporta el sujeto de la subordinada en Tal vez sea cierto que la naturaleza pone en marcha sus propios mecanismos de supervivencia (Regàs, Azul) se presenta como si fuera compartida por el oyente, a diferencia de lo que sucedería con indicativo (Tal vez es cierto...) (Manual 2010: 490; cf. also Lunn 1995: 430).

Nevertheless, the consideration of a further pragmatic aspect – namely the speaker’s degree of commitment – should not be ignored: the degree of commitment is lowered when using posiblemente + subjunctive as if using posiblemente + indicative. The opposition of thematic vs. rhematic information does not represent the only criterion that determines the speaker’s choice of the subjunctive or indicative. In the following examples posiblemente is used with the future tense, whereby it appears in both positions, future + posiblemente and posiblemente + future: (166) “El Barça es un equipo que podría jugar en la NBA, sin ninguna duda”, opina Bryant Jackson se toma unos segundos antes de evaluar la posibilidad de que los Lakers alcancen su tercer título consecutivo esta temporada. “Esta será posiblemente mi última batalla y confío en no acabar como el general Custer en Little Big Horn, precisamente en Montana, donde yo nací”. (El País 10/10/2010) (167) Tampoco jugará Reyes, sancionado, por lo que Quique Sánchez Flores deberá cambiar parte de su once, en el que entrará Fran Mérida, ya sea por la banda o como delantero junto a Diego Forlán, y al que posiblemente volverá Antonio López al lateral izquierdo, tras su suplencia en los dos últimos choques. Esas son las probables novedades en la alineación del Atlético, en la que David de Gea defenderá la portería [...] (El País 06/10/2010) (168) Jagland, asimismo, señaló frente al ‘Aftenposten’ que la reacción de China tras el anuncio del galardón no le sorprendió. El premio 2010 se convertirá posiblemente en “uno de los más importantes” en la historia del Nobel, agregó el presidente del comité noruego, que hizo sus declaraciones al rotativo durante una visita en Estados Unidos. (El Mundo 20/11/2010)

Posiblemente + future tense could be said to be used to express 1. a higher degree of certainty if compared with the construction posiblemente + subjunctive,

266 and 2. an inferential reading that is supported by contextually provided information: the synthetic future form. But generally, the adverb is clearly more epistemic than evidential which is emphasised by the fact that it can be intensified by muy ‘very’, which is logically linked to the choice of the indicative: (169) La nueva generación, definitivamente, no tiene remedio. Lo curioso es que la generación anterior, la mía, tampoco tuvo remedio. Y, muy posiblemente, la próxima tampoco lo tendrá. (El País 02/08/2010)

Evidential adverbs such as aparentemente or supuestamente, for instance, are not gradable with the help of intensifiers.

5.1.5 probablemente “[…] probablemente [expresses] a low degree of uncertainty, which, in regard to the use of the former [posiblemente], is reflected by the fact that it never takes the subjunctive” (Haverkate 2002: 36). If probablemente is used to express “a low degree of uncertainty”, it must express a relatively high degree of certainty (and later will be shown that, on the contrary, it can take the subjunctive): (170) Ese grupo es el de mayor implantación en la remota región paquistaní de Waziristán, donde también reciben protección y entrenamiento, según los servicios secretos norteamericanos, los jefes talibanes de Afganistán y de Al Qaeda, probablemente el propio Osama Bin Laden. (El País 12/05/2010) (171) El juez del caso Millet ha sido etiquetado como el caracol, por la presunta lentitud con que avanza el caso. Al golfista Sergio García lo etiquetaron como el niño, probablemente por su aspecto, como también es el niño Fernando Torres, jugador de fútbol. A la modelo australiana Elle McPherson la etiquetaron como el cuerpo, y a Naomi Campbell, como la pantera negra. (El País 15/06/2010) (172) “Qué quieres que te diga, qué te voy a decir”. “Pues que me digas cómo estás”, la aclaró ella. Casillas señaló que era un “momento muy feliz”, que estaba “muy contento” y repasó lleno de emoción su lista de agradecimientos a su familia, amigos... Probablemente se mordió la lengua para no citar a su novia, a la que tenía delante, se calló, se rió, miró para otro lado, ella le dijo “no pasa nada, vamos a hablar un poquito del partido” y él contestó que no con el dedo. (El País 14/07/2010)

In examples (170) and (172) the uses of probablemente simply indicate suppositions, whereas the use in (171) seems to be inferential: it is inferred that the golfer was caricatured as a child because of his childlike appearance (probablemente por su aspecto). Nevertheless, as posiblemente, probablemente is an adverb used to express (pure) epistemic modality. The adverb is usually used to express that [p] probably is the case, often – as in the case of posiblemente –

267 without referring to the reason(s) why the speaker assumes that it is probable that [p]: (173) [...] Novak Djokovic, el segundo mejor jugador del planeta, y que buscará en el Abierto de Estados Unidos, a partir del 30 de agosto, el único título grande que le falta. “Tengo confianza en mi tenis”, siguió Nadal antes de su debut individual en Canadá, probablemente el miércoles, al que llega “justito” porque solo se ha entrenado tres semanas: “Ahora pienso en mi tenis, no en mi cuerpo. Estoy en perfectas condiciones”. (El País 11/08/2010) (174) [...] ha decepcionado particularmente a Estados Unidos, que ha tratado desesperadamente de estrechar la colaboración con ellos. Con su voto negativo, Brasil y Turquía incursionan en un terreno político desconocido y, probablemente, de significativas consecuencias. La resolución ha tenido, no obstante, el apoyo de las principales potencias occidentales, de China y de Rusia [...] (El País 12/06/2010) (175) Los familiares se quedarán en casa de un amigo. Cosmin era un cliente habitual de Pícaro. Le conocía todo el mundo porque no faltaba ni un fin de semana. Esta discoteca es probablemente la más famosa entre los rumanos que residen en la región. El propio Alin, el encargado del local, conocía a Cosmin. (El País 11/08/2010) (176) El programa se completa con Fancy free, de Jerome Robbins, y VIII, de Christopher Wheeldon, inspirado en la vida de Enrique VIII, sobre una inquietante música de Benjamin Britten. Wheeldon es probablemente el mejor y más prometedor coreógrafo joven del panorama internacional y VIII fue creada para el Ballet de Hamburgo y luego repuesta en Nueva York. (El País 03/08/2010) (177) Medalla asegurada. El oro lo perdió la española en la ría. Más bajita que Zarudneva, con menos técnica sobre los obstáculos, Marta retrocedía siempre un poquito en cada valla y lo recuperaba en el llano. Probablemente confiaba en su sprint final, en ese apretar los dientes donde lo saca todo que le valió el oro mundial hace justo un año, pero esta vez no le alcanzó (9m 17,74s). (El País 02/08/2010)

In summary, probablemente in the examples above does no more than express a probability that a certain state of affairs is/was the case. The following example, in contrast, is similar to example (171), where the inferential meaning is contributed by the context: (178) Se llama Boule y fue descubierto, en 2001, como gen humano. “Esta es la primera evidencia clara que tenemos que indica que nuestra capacidad de producir esperma es muy antigua, probablemente originada en el surgimiento evolutivo de los animales, hace 600 millones de años”, afirma el científico Eugene Xu, líder del equipo autor del descubrimiento. (El País 18/07/2010)

The scientist explicitly refers to evidence (esta es la primera evidencia clara) for the fact that ‘our capacity to produce spermatozoa is very old’ (nuestra capacidad de producir esperma es muy antigua). He then adds that this capacity ‘probably originated with the emergence of animals’ (probablemente originada

268 en el surgimiento evolutivo de los animales). It is consequently reasonable to assume that he also has evidence for [p] marked by probablemente. Comparing the examples above, the adverb’s scope is different. In examples (175), (176) and (177), for instance, probablemente functions as a sentence adverb because it has scope over the whole sentence, even though it is not found at the beginning of the sentence. In examples (173) and (174), by contrast, probablemente qualifies only one part of [p] (see the underlined phrases). However, probablemente always seems to mean that [p] (or a part of [p]) is probably the case. Probablemente – in contrast to Haverkate’s (2002) statement – may select the subjunctive, which seems to be typical of newspaper style, and if so, it expresses a lower committed epistemic evaluation that a certain state of affairs is/was the case, as if the adverb were used in connection with the indicative: (179) ¿Practicaban alguna disciplina esotérica? ¿Qué se propusieron? Probablemente el asunto se archive y nunca sepamos exactamente lo que ocurrió ayer en el estadio Loftus Versfeld. Probablemente los miles de hinchas japoneses y paraguayos que asistieron a la cita, expectantes, olviden pronto estos acontecimientos y solo retengan la consecuencia burocrática. (El País 02/07/2010) (180) El PP y sus amigos, mal que les pese. Los demás están todos. La paradoja del caso es que probablemente la presidenta del TC tenga razones para pensar que ha salvado el Título VIII de la Constitución [...] (El País 15/07/2010) (181) Por eso, las pymes que constituyen su red de proveedores deberán moverse en idéntico sentido y de forma inmediata, si no quieren perder un tren de vanguardia que probablemente no vuelva a pasar. (El País 18/08/2010) (182) Lacourt hizo la que probablemente sea la mejor marca de la competición al batir el récord de Europa de 100 espalda en 52,11s. (El País 18/08/2010)

The use of probablemente + subjunctive could thus be compared to the use of posiblemente + indicative. Probablemente and posiblemente are generally differentiated by the fact that the former expresses a higher degree of certainty than the latter. But the probability that [p] is lowered if probablemente is found with the subjunctive and in those cases its use is similar to posiblemente + indicative. Hence, it could be assumed that probablemente + subjunctive could be substituted by posiblemente + indicative. The following examples furthermore show that probablemente + subjunctive is not only used by journalists but is also to be found in direct quoted speech or in chat conversations: (183) “España cuenta actualmente con el mejor rating posible y no puedo decir donde se situará en el futuro, pero probablemente baje un poco” ha declarado esta mañana Steven A. Hass, responsable de crédito de Moody’s. (El País 02/08/2010)

269 (184) “Se pueden descubrir bastantes más cosas”, apunta Baceta “este Castro tiene 2.500 años, ha sufrido muchos cambios, hay zonas que probablemente haya desaparecido pero aún así creo que en Arrola todavía queda bastante por dar”. (El País 11/08/2010) (185) Desde el patio de butacas se escucharon piropos y aplausos, aunque muchos, probablemente, sigan sin comprender qué es lo que están viendo. La fórmula Mercé Una hora y media después de que Galván saliera a escena llegó el turno a José Mercé, segundo recital de la noche. (El País 11/08/2010) (186) Lo que parece claro es que un periódico digital es un bien para el medio ambiente, transporte 0,00 gasto en papel 0,00 y ese ahorro en las cuentas de gastos de los medios. Probablemente las cuentas sean diferentes cuando el medio digital sale también en papel. Desde mi punto de vista la lectura no es la misma que un diario, porque los medios digitales se actualizan en tiempo real. (El País 04/07/2010) (187) Yo le recomendé que volviera durante la semana grande de las fiestas, y sobre todo para estar aquí mañana, el día de la gran cabalgata; no podía venir en estas fechas. Seguro que el gentío será tan grande como en años anteriores, y probablemente pase desapercibida para la mayoría la ausencia del camión engalanado que tenía intención de enviar (y pagar) el Ayuntamiento de Tel Aviv [...] (El País 04/07/2010)

Just like posiblemente, probablemente can take the indicative future. In those cases the use of probablemente may be said to be accompanied by an inferential moment, even though the context does not refer to any kind of evidence that would have caused the speaker to infer that [p]. More precisely, it could be said that the speaker expresses a relatively high degree of certainty concerning a future event: (188) Aunque el portavoz de la compañia no ha ofrecido detalles de la conferencia, que se celebrará en la sede central de Cupertino, probablemente tratará de iPhone 4, que sigue acumulando críticas de consumidores y analistas por sus problemas de cobertura, minimizadas hasta ahora por Apple. (El País 18/07/2010) (189) Esta semana, tanto el presidente de Afganistán, Hamid Karzai, el de EE UU, Barack Obama, como el jefe del Estado Mayor Conjunto norteamericano, Mike Mullen, han asegurado que la filtración de los documentos de Afganistán provocará, probablemente, las muertes de soldados y colaboradores afganos. Mullen ha ido más lejos. Ha dicho que Wikileaks y sus gestores “pueden tener ya las manos manchadas por la sangre de algún joven soldado o de alguna familia afgana”. (El País 03/08/2010)

Probablemente was furthermore found taking the conditional. In those cases the adverb is used to express a hypothesis the speaker judges as probable, whereby the speaker is far away from being certain that [p]:

270 (190) [....] para saber que si éstas tuvieran que abonar la limpieza y los desperfectos que generan indirectamente en el medio urbano adyacente, indemnizar a quienes lo habitan por la falta de descanso y costear los policías que requiere mantenerlo tranquilo probablemente no serían viables económicamente. Pero mientras tales conceptos no entren en su contabilidad, los problemas descritos se reproducirán de modo cíclico por un motivo obvio: ensuciar es gratis. (El País 03/08/2010)

In sum, the use of probablemente is highly similar to the use of posiblemente – provided they take the indicative. Regarding their conceptual meaning inherent in the adverbs, the former expresses a higher probability that [p] than the latter. For both adverbs, the inferential meaning may be contributed by the context, even though they are not evidential adverbs.

5.1.6 seguramente For the use of seguramente I expect similar results as were shown for the use of no doubt and undoubtedly by Simon-Vandenbergen/Aijmer: “Although both no doubt and undoubtedly literally mean that ‘there is no doubt’ and hence express certainty, they have developed different pragmatic functions” (Simon-Vandenbergen/Aijmer 2007: 146). Speakers use these adverbs when they have doubt. So seguramente is expected to be used if (and only if) the speaker/writer is not totally sure: (191) Santos va a enfocarse mucho más que Uribe en el desarrollo social y económico [...]. Seguramente le prestará mucha más atención a la realidad y a la percepción interna y externa del déficit de respeto a los derechos humanos: entiende el problema, y lo quiere atender. También puede -y probablemente lo hará con gusto- librar [...] (El País 09/07/2010) (192) Casi todos los expertos en turismo estarían de acuerdo en que en Europa hay un selecto puñado de ciudades de primerísima categoría. Roma. París. Londres. Probablemente Viena y Praga. Seguramente ninguna más. (El País 14/07/2010) (193) El pragmatismo chino parece haberse impuesto de momento. Consciente del enojo que podría haber generado en una buena parte de los 400 millones de internautas con que cuenta el país – entre ellos, seguramente muchos funcionarios del Gobierno y empresas chinas –, consciente del mal efecto que habría supuesto para la imagen de China y para el clima inversor extranjero, Pekín ha decidido renovar la licencia del buscador de Google […] (El País 11/07/2010) (194) Los jugadores se las han ingeniado todas con tal de que se les pasara el mes lo más rápido posible. Camino del paraíso, se evaporaron las dos primeras semanas, seguramente porque la porra organizada por Puyol fue reñida -se la llevaron Piqué y Mata; Capdevila fue segundo- y porque había partidos a todas horas por tele. Algunos siguieron los partidos de Pau Gasol. Otros la final de Wimbledon de Rafa Nadal. (El País 14/07/2010)

271 In text passage (191) the adverb is clearly used to express an inference because it is followed by a verb in synthetic future form (prestará), and the futuro sintético was shown to express inference (Squartini 2001).122 Furthermore, going beyond the sentence level, probablemente + synthetic future (hará) also seem to express at least (if not an inference) a relatively high committed epistemic evaluation that a certain state of affairs is the case. Example (192) contains the information that ‘Rome, Paris and London belong to the first class tourist cities. Probably (also) Vienna and Prague. Certainly not more cities.’ Cornillie (2010a) explains that [modal adverbs] in –mente unlike manner adverbs, can also be transformed in a construction formed by a copula and an adjective. The finite subordinate clause introduced by que is then the subject of the sentence […] (Cornillie 2010a: 304).

But trying to paraphrase the adverbs with es probable que [p] and es seguro que [p], the sense of the information would be modified (beside the fact that we had to add a verb): Es probable que Viena y Praga son/sean ciudades de primerísima categoría. Es seguro que ninguna ciudad más es de primerísima categoría. The transmitted information would be modified by paraphrasing the adverb seguramente, whereas the information containing probablemente would not. The reason is that, using seguramente, the speaker wants to express that he is fairly sure that [p], not totally sure that [p]. The adverb’s use in text passage (193) should be analysed in similar terms compared to the use in (192). The use or function of seguramente in example (194), on the other hand, seems to be an inferential one because it is part of a subordinate clause indicating the reason with porque. However, the inferential meaning does not seem to be encoded by seguramente; it is contributed by the context. Hence, it seems reasonable to conclude the same what Simon-Vandenbergen/Aijmer found out for certainly: “The use of certainly […] occurs in a context of uncertainty rather than certainty” (2007: 211), whereas in Byloo/Kastein/Nuyts’ (2006) view, a (relatively) strong epistemic use exists for certainly. In their paper “On certainly and zeker” Byloo/Kastein/Nuyts investigate “the meanings and uses of the English and Dutch adverbs” (2006: 45). They find out that the adverbs expressing ‘certainty’, which they also call “strong” adverbs “clearly tend to be used in a much more ‘flexible’ way than the (relatively) weaker ones (Byloo/Kastein/Nuyts 2006: 45). The main difference between the English adverb and the Dutch one is that a weaker and a stronger epistemic use exists for zeker, while certainly “only features the stronger one” (Byloo/Kastein/Nuyts 2006: 66). For English certainly may only exist a stronger epistemic use, because surely features the weaker one 122 That is why this future form also received the name futuro de probabilidad (Volkmann 2005: 287).

272 (Byloo/Kastein/Nuyts 2006: 67), that is, English provides one form for the stronger use and another one for the weaker use. Dutch zeker, in contrast, ‘has to’ convey a stronger as well as a weaker epistemic use because Dutch provides only one form. If one applies this explanation to the use of seguramente, the Spanish adverb, as the Dutch one, is assumed to convey both a stronger and a weaker epistemic – and even evidential – use, as Spanish provides only one adverb for featuring (weak or strong) certainty.123 This in turn implies that the translation of seguramente into English should vary according to the degree of certainty: if seguramente features a stronger epistemic/evidential use certainly represents the right choice; if the Spanish adverb features a weaker epistemic/evidential use surely should be the appropriate translation. So the theory goes. However, in the majority of cases, it must be hard to distinguish between one use and the other. Furthermore, I wonder which contextual criteria should be considered to be able to decide whether a certain use of seguramente can be termed ‘stronger’ or ‘weaker’, and what if one comes across a use that is stronger than a weak one, but not a real strong one? When is a use strong and when is it not? However, Byloo/Kastein/Nuyts admit: “Of course the fact that the speaker uses this adverb [certainly/zeker] instead of an unqualified assertion may sometimes indicate that conceptually (s)he is not really completely certain” (2006: 51). That is why in the present study for the translation of seguramente certainly was simply kept to. In the following examples seguramente is also used to feature a stronger – stronger in the sense of ‘stronger rather than weaker’ – evidential use because it is used with porque indicating the reason for ‘seguramente [p]’. The following two examples are interesting in so far as seguramente does not appear in the same clause as the state of affairs it modifies. Its use is somehow anaphoric, that is, it refers back to [p]. In example (195) Seguramente porque X ‘Certainly because X’ refers to ‘All this did not shock me’ (Todo eso no me asustaba) and in example (196) Seguramente es porque X ‘Certainly it is because X’ also refers back to the earlier utterance: (195) De aparcar los largos vestidos de noche en favor de coquetas prendas de día que no piden disculpas por su pragmática feminidad. “Todo eso no me asustaba. Seguramente porque siempre he sentido que soy el hombre apropiado para el puesto. Y porque mi ropa es fácil y apetecible, sin ser aburrida. [...]” (El País 17/08/2010) 123 However, it could be assumed that sin duda ‘beyond doubt, no doubt, undoubtedly’ and indudablemente ‘without doubt, undeniably’ are used to convey a higher degree of certainty than seguramente. Against this backdrop, seguramente – never expressing absolute certainty – could be translated as ‘surely’, while sin duda and indudablemente could also be paraphrased as ‘certainly’. But this is only an assumption and needs further investigation.

273 (196) Siempre me he preguntado por qué Ángel Cristo entraba en la jaula incluso con leones que le había pasado un colega al que no le comprarías un coche usado. Seguramente es porque no sabía hacer otra cosa. (El País 08/05/2010)

The use of seguramente in the following example can also be described as inferential and as fairly certain because the information ‘certainly in the Middle East and over a short time […]’ (seguramente en Oriente Próximo y durante poco tiempo [...]) is based on models of population: (197) [...] explica Pääbo, del Instituto Max-Planck de Antropología Evolutiva (en Alemania). “Esto indica que la hibridación se produjo después de que el Homo sapiens empezara a salir de África, seguramente en Oriente Próximo y durante poco tiempo, antes de que evolucionaran las distintas ramas euroasiáticas”. Se basa en que los modelos de población indican que cuando una población colonizadora se topa con una población residente [...] (El País 08/05/2010)

Even though it can be said that seguramente is generally used if the speaker is not (totally) certain, the degree of certainty expressed by the adverb may be ‘affected’ by its surroundings such as verbal mood. In the following four examples an inferential reading is applied to the adverb due to contextually provided information, more precisely, due to the inferentially used future forms. In example (202), in contrast, the adverb is used with the subjunctive, which clearly lowers the degree of certainty of seguramente when compared to the degree of certainty that is conveyed in the other examples: (198) Pese a las sólidas razones para llevar a cabo la modernización, los observadores de Rusia ven con escepticismo la sinceridad de las intenciones del presidente. La modernización se quedará seguramente en palabras, y sirve para ocultar que la clase dirigente prefiere el statu quo. El plan del Kremlin adopta un enfoque restrictivo: solo se limita al progreso tecnológico. (El País 11/08/2010) (199) [...] los problemas que encontraron para desmontar el módulo hicieron imposible el cambio, ha informado la NASA. Por eso, se va a realizar un nuevo paseo espacial, no antes del miércoles, y seguramente será necesario uno más para terminar la tarea. La situación en que se encuentra la estación es de normalidad, pero si se estropeara la otra mitad del sistema de climatización mientras no funciona la primera [...] (El País 11/08/2010) (200) Prohibir la tortura es algo en lo que el 99% de la humanidad ya está de acuerdo. El porcentaje de los que creemos que no importa cuál sea la especie del maltratado o torturado va en aumento, y seguramente en un futuro más o menos lejano se equiparará al consenso actual cuando las víctimas son humanas. (El País 10/08/2010) (201) Así hasta 616 veces con 616 pollos. Sólo 10 tuvieron que ser trasladados a un centro especializado por una rotura en las patas o las alas. En pocos días

274 seguramente podrán volver a unirse a la colonia más numerosa del Mediterráneo occidental, con más de 8.118 pollos nacidos este año. (El País 10/08/2010) (202) Mientras tanto, los seguidores de Perdidos prácticamente han abandonado la esperanza. Y los fieles a la noventera Friends, por su parte, seguramente se queden para siempre a la espera de esa ansiada dosis extra tantas veces anunciada y nunca hecha realidad. Quizá sea mejor así. (El País 11/08/2010)

Generally speaking, if used with the indicative (future), the reading conveyed is inferential, while if used with the subjunctive, the reading is more modal and the speaker’s degree of certainty is definitely lowered as if the adverb were used with the indicative. Seguramente may also be used with a verb in the past tense: (203) [...] de que hay familias con discapacidades muy duras, difíciles, y creo que es bueno para todos relativizar esas situaciones. Para nosotros no ha sido ningún sufrimiento. Su tranquilidad ante el éxito y el fracaso tendrán que ver con ese aprendizaje. Seguramente que eso nos ayudó a poner las cosas en su sitio. Cuando nació fue un dolor, pero a los pocos días nos dimos cuenta de que era un niño como los otros. (El País 10/08/2010) (204) Le dijo usted a Zapatero en el programa de Ángels Barceló: “No se olvide que siempre que llueve escampa”. Es un dicho castellano, pero es verdad. La situación es complicada, pero no es solo española, sino mundial. Seguramente nos han arrastrado otras economías, pero vendrán tiempos mejores. Si superamos los años cuarenta y los cincuenta, con las dificultades que hubo, cómo no vamos a superarlo ahora. (El País 10/08/2010)

The inference expressed by seguramente + verb in the past tense clearly indicates a high degree of certainty because the speaker evaluates a fact from the perspective of the current situation. It is nevertheless no maximum degree of certainty because [p] is marked by seguramente. The following instance of seguramente is also used with a verb in the past tense. However, it should be analysed differently: (205) Tres científicos han identificado en Tenerife una nueva especie de pájaro -el verderón de pico fino- que sobrevivió hasta hace 13.000 años y seguramente despareció con la llegada de la especie humana a las islas y la introducción de roedores. Este paseriforme (del orden de los gorriones) ayuda a comprender la historia de la evolución y la morfología [...] (El País 30/09/2010)

It seems as if seguramente is not used from the perspective of the journalist but from the perspective of the three scientists mentioned at the beginning of the text passage. The journalist obviously transmits information marked by seguramente for which the source is the scientists. Hence, the use of seguramente is partly inferential/not totally certain and partly quotative: the speaker who marks [p] by the adverb is not totally certain that [p] is the case and simultaneously in-

275 dicates that the information originated in another origo than his. The last text passages, in contrast, do not contain any reference to certain evidence or reasons why [p] is seguramente the case for the speaker: (206) Reina y Marchena son los jugadores más expresivos en el banquillo español Siempre fue así. “El espíritu del grupo sobrevive, se ha mantenido pese a los cambios”, avisa José Manuel Ochotorena, entrenador de porteros hace dos años y ahora. Seguramente, él es quien mejor lo sabe, aunque solo sea porque la incorporación de Víctor Valdés se interpretó como un problema y no ha existido ni un atisbo de roce. (El País 04/07/2010) (207) Sin embargo, a pesar de lo detallado de la información y de las numerosas fuentes que cita, obvia un dato que creo que es importante y seguramente hubiera sido relevante para sus lectores: la Unión Europea contribuye a la financiación de estas infraestructuras de Alta Velocidad entre Madrid, la Comunidad Valenciana y la Región de Murcia con casi 2.000 millones de euros [...] (El País 30/09/2010) (208) Pero he de decir que Auxerre tiene el estadio pequeño más bonito de Francia. Igual que los vestuarios del equipo visitante. Seguramente los jugadores del Madrid no lo van a encontrar muy grande ni muy confortable, pero hemos pintado las paredes ¡y ha quedado como nuevo! (El País 30/09/2010)

In those cases the examples are perhaps more appropriately translated as surely instead of certainly as surely conveys a slightly lower degree of certainty than certainly. However, there are uses of seguramente that do not go hand in hand with explicit reference to evidence, but the speaker is nevertheless very certain that [p]: (209) Es evidente que esa conspiración se puede atribuir sin duda alguna a María Dolores de Cospedal, secretaria general del Partido Popular. Poner recortes de periódicos, con noticias, seguramente inducidas desde ellos mismos, en manos del fiscal es una forma poco decente y desde luego calumniosa de actuación. (El País 12/06/2010)

As the context shows (Es evidente que and sin duda), the speaker does not have a high opinion of María Dolores de Cospedal. For him ‘it is evident that (Es evidente que) the conspiracy can be undoubtedly (sin duda) assigned’ to her. The speaker is thus highly convinced of the fact that the news was induced by her and other members of the Partido Popular. In comparing, for instance, examples (206), (208) and (209) it becomes obvious that seguramente may have scope over sentences (example 208) (or at least whole clauses as in example 206) as well as over phrases only (example 209).

276

5.1.7 supuestamente The adverb supuestamente is often to be found in contexts treating crimes or juristic questions, which can be explained by the fact that this adverb “excludes personal experience and reports the observation made by someone else” (Cornillie 2010b: 313): (210) Skripal, antiguo coronel del Ejército ruso, fue condenado en 2006 a 13 años por realizar actividades de espionaje para Reino Unido. Supuestamente compartió información sobre decenas de sus antiguos colegas que trabajaban encubiertos en Europa, llegando incluso a revelar los lugares en los que tenían lugar reuniones secretas, así como sus direcciones, y contraseñas. (El País 11/07/2010) (211) En aquella operación de 2006 también fueron detenidos en diversos puntos de la provincia de Alicante otros ocho miembros de la organización, entre ellos Isabelle R. y el abogado J. J. S. quien, según la Policía, supuestamente colaboraba en las actividades [...] (El País 22/07/2010) (212) La policía nacional detuvo el jueves a 13 personas, la mayoría de un clan familiar, a las que se acusa de controlar, supuestamente, el tráfico de hachís al menudeo en el barrio de El Cabanyal. Este clan familiar estaba perfectamente estructurada, según fuentes policiales, y utilizaba a personas de origen sudamericano como correos. (El País 12/07/2010) (213) La última víctima, Andrea Zambrano, ecuatoriana de 28 años y embarazada, había denunciado a su pareja en dos ocasiones, y ayer murió estrangulada supuestamente por él, Marcelo Iván C. E., de 33, en la habitación que compartían en el barrio barcelonés de Sants. (El País 10/07/2010) (214) [...] FBI han detenido en la Gran Vía de Madrid a un estafador internacional buscado por Estados Unidos. El arrestado, R.L, un boliviano de 70 años, estafó 1, 3 millones de dólares a través de una falsa organización caritativa [...] que supuestamente ayudaba a niños con necesidades, especialmente en Sudamérica. (El País 30/05/2010) (215) El agredido confesó el paradero de algunas sustancias y los imputados fueron a comprobar la veracidad de lo declarado, además de que supuestamente se apropiaron de dinero en metálico y sustancias tóxicas cuya existencia no se pudo determinar en el juicio. Entre tanto, uno de los agresores aguardó en el bosque junto al enmanillado para evitar que este intentara huir. (El País 18/07/2010) (216) [...] donde trabajaba como portera para el periodo de vacaciones. La policía trata de localizar a un hombre que ejercía de portero en un inmueble muy próximo al lugar de los hechos y con el que supuestamente convivía la mujer, han informado fuentes de la Delegación del Gobierno en Aragón. (El País 18/07/2010)

As the examples given above indicate, supuestamente is often used to give information about a crime when the state of affairs is not really clear at the time of writing/uttering the information. The state of affairs that is communicated is still

277 not really certain but supposed. Sometimes the adverb does not qualify [p] as a whole but one part of [p]. So the adverb may have scope over whole sentences (cf. example 210) as well as over one part of [p] (cf. example 211). Cornillie (2010b: 313, 315) has shown that supuestamente can only perform the reported reading but I would like to draw attention to the fact that it is very often accompanied by an inferential moment, slightly different though: if we deal with a ‘classical’ inferential use of a certain linguistic expression, the speaker himself is the person who drew the conclusion, while in the case of supuestamente the speaker/journalist transmits the inferentially gained information from another person (inference + reported). The evidential meaning aspect is thus assumed to be inherent in the adverb. This may be the reason why supuestamente is notably frequently found in contexts treating crimes or juristic questions, as the following instances of this adverb again confirm: (217) [...] tras detectar una serie de empresas del sector del alcohol que eludían el pago de los impuestos especiales mediante la declaración de falsos envíos de ese producto a vinagrería. El objeto final del fraude era el desvío de alcohol supuestamente empleado en la fabricación de vinagre, para destinarlo a otros usos sin pagar del Impuesto Especial del Alcohol y las Bebidas Derivadas, así como del IVA sobre éste. (El Mundo 27/11/2010) (218) Mustafa Salma, supuestamente secuestrado por el Frente Polisario en septiembre de 2010. “Estoy aquí para pedir al Estado español que ayude a mi hijo”, ha afirmado esta mañana. (El Mundo 27/11/2010) (219) El Ministerio de Exteriores saudí anunció hoy la detención de 149 personas supuestamente vinculadas a la red terrorista al Qaeda en Arabia Saudí. En una rueda de prensa retransmitida por el canal de capital saudí ‘Al Arabiya’, el portavoz del Ministerio de Exteriores, el general Mansur Al Turki, explicó que los detenidos [...] (El Mundo 27/11/2010) (220) La periodista deja claro, no obstante, que “nadie debería creer que una mujer que no lleva velo no es una buena musulmana”. Una opinión que no comparten muchos en Oriente y algunos aquí, como el imán de Cunit (Tarragona), quien supuestamente acosó a una correligionaria por vivir a la occidental, por no cubrirse, por conducir un coche. (El País 25/04/2010) (221) [...] Joana Barceló, quien esta mañana ha remarcado que el Ejecutivo no se siente “en absoluto” deslegitimado pese a haber formado gobierno con UM, partido del que se investiga una supuesta compra de votos, al tiempo que ha destacado que, supuestamente, estos votos se consiguieron mientras UM “gobernaba con el PP”. Muy contraria es la postura del portavoz ‘popular’, quien asevera que “en todo caso, quien se benefició de estos votos fue Antich [...]” (El Mundo 27/11/2010) (222) [...] Superior de Justicia de Cataluña (TSJC) ordenó, tras la absolución de la acusada por el magistrado Castelló, que se volviera a repetir el juicio, con lo cual el caso regresó a la fase de instrucción por la aparición de nuevos testigos, a quienes

278 supuestamente Antonio M. confesó que participó en el apuñalamiento de la víctima a cambio de una moto y medio kilo de cocaína. La causa: Discusión por impago de la pensión. El asesinato de Antonio S.O., que murió a consecuencia de las 18 puñaladas recibidas, ocurrió tras una acalorada discusión mantenida con su ex esposa por el pago de la pensión alimenticia de la hija y porque supuestamente el padre pretendía llevársela a vivir consigo a Granada. Según la Fiscalía, los dos acusados entraron la noche del 6 de diciembre en el domicilio de la víctima [...] (El Mundo 27/11/2010)

All these examples again indicate that supuestamente is often used to give information about a crime when the state of affairs is not really clear at the time of transmitting the information to the readership. The last example in particular reveals that the information marked by the adverb is still not really certain but supposed in contrast to the last utterance introduced by Según la Fiscalía: as will be shown later (cf. chapter 7.1), there are two kinds of the según X construction – según X + indicative and según X + conditional. If the former is used, the transmitted information is assumed to be more certain.124

5.2 The evidential use of Spanish verbs of cognitive attitude Nuyts (2001b) says that verbs of cognitive attitude “are notoriously difficult to deal with, much more so than the adverbs and adjectives, mainly because of their complex semantic structure and their mysterious linguistic behaviour” (Nuyts 2001b: 107). Nevertheless it will be tried to describe in detail the evidential and epistemic meaning aspects of Spanish verbs of cognitive attitude and to find out which meaning aspects are encoded by the verbs and which may be contributed by the context. The findings will be compared with the conclusions on the English verbs of cognitive attitude drawn by Cappelli (2007), who takes […] verbs of cognitive attitude [to] conceptually encode the speaker’s evaluation that he doesn’t know that p is actually the case, but that he has a preferred hypothesis: he doesn’t know if p is true but he knows that p is possible (Cappelli 8125).

The forms of the verbs to be analysed here are qualificational and non-descriptive, that is, “they truly encode the subject’s attitude” (Cappelli 2007: 112). 124 For the study of the English equivalent of supuestamente see also Celle’s analysis of the use of supposedly (2009: 288). 125 The page numbers for the source “Let’s assume it is both conceptual and procedural… A hypothesis about the information encoded by verbs of cognitive attitude” (Cappelli) are taken from this pdf file: http://www.gloriacappelli.it/blog/media/cappelli-assumeconceptual-and-procedural.pdf.

279 That is why only verb forms in the 1st person singular are considered, and therefore quoted speech or utterances found in chat forums are dealt with, as verbs of cognitive attitude in the 1st person singular are not used by journalists themselves.

5.2.1 creer [Believe] encodes an affective dimension. By using this verb, the speaker signals that he/she is ready to support his/her assertion with the strength of his/her subjective, affective commitment. The possibility for the verb to lexicalize this dimension means that, besides pointing towards the epistemic domain, it also involves evidentiality. The speaker is highly committed to the likelihood that a state of affairs is the case on the grounds of some sort of affective evidence that he/she has for it (Cappelli 2007: 170).

How affective-evidential is the use of Spanish creer? (223) [...] fenómenos culturales, como la película [...] que provocó una explosión de interés por el Pinot Noir. “Yo creo que el verdadero poder está en el arte de saber contar historias. Esa es la mejor manera de vender, ya sea un vino o una información. Creo que por eso he tenido éxito, porque soy un buen contador de historias”. (El País 11/07/2010) (224) P[regunta]: ¿De quién se ha nutrido usted? R[espuesta]: De muchos, creo que una de las personas que más me ha influenciado es Martín Caparrós y una autora como Susana Orlean que tiene una mirada muy sutil y muy sofisticada que me interesa muchísimo. (El País 11/07/2010) (225) “[...] Soy optimista de cara a mañana, creo que tenemos opciones de lograr un podio que sería muy bueno. Me gustaría poder poner en aprietos a los Red Bull, aunque creo que será complicado”, ha dicho Alonso [...] (El País 12/07/2010) (226) “[...] Nos han metido dos goles, uno de rebote y el otro, de contra. Hemos estado sólidos. Yo creo que la base del éxito es una buena defensa, porque luego, con la calidad que tenemos, siempre creo que uno u otro jugador marcará. [...]” (El País 12/07/2010)

It seems that it could be verified that the verb creer – as its English counterpart believe – may lexicalise the evidential domain beside the epistemic one. The speaker indeed is “highly committed to the likelihood that a state of affairs is the case on the grounds of some sort of affective evidence that he/she has for it” (Cappelli 2007: 170). Let us consider the examples in detail: in examples (223) and (224) the speaker uses creo que marking information epistemically/evidentially concerning his own person. That is why the affective dimension is present. In text passage (225) the verb is combined with the synthetic future form será, which emphasises that we deal with an inferential use. The use in (226) seems to be inferential as well because the subordinate clause, following

280 the clause containing creo que, gives reasons for the speaker’s believing. Hence, the table for English believe developed by Cappelli (2007: 172) seems also to be valid for Spanish creer: Table 15: The conceptual dimensions lexicalised by believe (Cappelli 2007: 172) believe

EVIDENTIALITY Affective Evidence affective dimension present

Likelihood positive degree

EPISTEMICITY Certainty the evaluator is not certain but fairly committed

The fact that the use of creo que indeed is highly affective is illustrated by the following examples as well. Examples (227) and (228) represent judgements: in example (227), the speaker makes a judgement about herself indicating that she believes that ‘she swam well’ (he nadado bien) and in example (228), the speaker judges that he ‘learned a lot of marvellous things’ (he aprendido cosas maravillosas): (227) La más rápida en los 100 braza ha sido la rusa Yuliya Efimova, con 1:07.96. “Creo que he nadado bien, pero podía haber sido mejor. Yo he intentado darlo todo, pero supongo que al final he querido reservarme un poco. Esto quiere decir que esta tarde puede ser mejor. El año pasado en el Mundial era todo nuevo para mi, me sorprendió todo un poco. [...]” (El País 12/08/2010) (228) “[...] Yo quiero seguir trabajando porque creo que nací para morir amarradito al yugo, como digo yo. La vida a mí me ha dado cosas muy lindas... me ha tratado muy mal, me ha tratado duro muy duro, pero ¿les digo algo...? Creo que he aprendido cosas maravillosas y a tomar los buenos caminos de la vida”. (El País 17/10/2010)

So the use of creo que in the examples above may be described as indicating tentative conclusions. The speakers believe that [p] because they themselves were involved in [p]. Equally affective-inferential is the following use of creo que. The speaker is an optimist and that is why he believes in [p]: (229) El lehendakari volvió a situar en septiembre la posible transferencia de las políticas activas de empleo. “Soy optimista y creo que se conseguirá en septiembre, que no es una mala fecha”, indicó un poco forzado [...] (El País 12/08/2010)

The next examples neither contain contextually provided information which indicates that the use of creo que is clearly inferential nor contextually provided information which shows that the speakers believe that [p] because they themselves were/are involved in [p]. There are consequently various reasons why the speakers may use creo que. However, it should be emphasised that it is used with the personal pronoun yo. In Spanish, the use of the personal pronoun is optional. So there must be a reason why the speakers say yo creo que instead of

281 creo que. Yo is one of the most subjective words a speaker may use. So the following instances of yo creo que may be seen as opinion-indicators which speakers use to mark [p] as highly subjective they affectively believe in without explaining why: (230) Ya no hay trabajo para ellos y pasan a situaciones de retiro y, a diferencia de sus compañeros chilenos, sin la vida asegurada para el resto de su vida. Yo creo que este es el momento de defender los puestos de trabajo, denunciar las situaciones límite que hacen que los trabajadores tomen medidas tan drásticas como encerrarse en una mina para luchar por su empleo. (El País 17/10/2010) (231) Aquellas personas que tienen la posibilidad de hablar con su esposo antes de hacer cosas indebidas. O el esposo que tenga la posibilidad de hacer cosas indebidas, antes de engañar a su esposa, yo creo que antes de hacer eso tienen que hablarlo. No terminar las cosas así como así, nada más. Yo creo que el amor es lo más hermoso que puede existir en la vida. (El País 17/10/2010) (232) “Hay dos cosas diferentes. Por un lado yo creo que es casualidad que todos estos artistas hayan hecho estos álbumes. Lo que no creo que sea tan aleatorio es que todos se hayan publicado en el último trimestre del año. Muchos están destinados a ser producto navideño. [...]” (El País 17/10/2010) (233) Lo esencial, dice Tina, es “tener una historia de 90 minutos, una cámara y un micro”. Solo ha necesitado cinco actores, una peluquera y una maquilladora. “Yo creo que en el futuro es que la autoproducción sea un pilar más y fundamental de lo que se puede hacer en el cine”. Es realista, “lo peor es el momento de la distribución, antes yo controlaba todo, ahora no depende de mí”. (El País 12/08/2010) (234) ¿Qué queda ahora de Manara que participó a la rebelión de 1968? R. Mucho, muchísimo. Pero creo que con respecto al ‘68 he dado un paso más. Entonces se creía que todo era política, que el privado fuese política, que todo se podía reconducir a la dimensión política. Ahora yo creo que todo es cultura. Y que cualquier elección política es también una elección cultural. Creo que la cosa más importante es la educación, preocuparse de saber, de conocer, de darse una propia cultura. (El País 17/10/2010)

The next example contains instances of creo que where the inferential component is clearly more prominent than the epistemic one. The speaker explains that he believes/thinks that it is a cruel question (creo que es una pregunta cruel) and then adds that it is (a cruel question) because X and because Y (Lo es porque [X] y porque [Y]), that is, the speaker indicates the reason for his believing. According to the speaker, the answer to this cruel question should be obviously negative (La respuesta suele ser obviamente negativa). Thus, X and Y represent the basis for another conclusion that is marked by obviamente: (235) [...] las víctimas del terrorismo, y creo que es una pregunta cruel. Lo es porque parece cuestionar una respuesta que suele ser obvia y porque expone a la

282 víctima a un juicio de valor ajeno sobre su absoluta fragilidad e inocencia. La respuesta suele ser obviamente negativa, y creo además que debe serlo. He aquí la respuesta de José Ramón Recalde, sobre la que volveré: “¿Por qué le iba a perdonar? [...]” (El País 18/09/2010)

In example (236) the quoting journalist already infers that [p] – introduced by Creo que – represents a conclusion, which proves that native speakers also perceive creer as a marker of inference: (236) [...] las tres cajas sumadas nos haría ser muy potentes y muy competitivos”, aseguró, tras destacar que el empresariado vasco se merece una caja de más entidad para hacer frente a retos mayores. “Pero haremos lo que digan los tres presidentes” reiteró. “Creo que la política debiera inmiscluise lo menos posible”, concluyó. (El País 12/08/2010)

The last example shall illustrate that it is even possible to ‘sincerely believe’ that [p], that is, the affective dimension is clearly present: (237) Un gesto para mejorar las relaciones entre Cuba y la UE. “La presencia de Fariñas sería muy apreciada por los eurodiputados y por los ciudadanos europeos. Creo sinceramente que la autorización de su viaje podría tener un impacto positivo no sólo en nuestra cámara, sino en las relaciones entre Cuba y la UE”, escribe el polaco [...] (El Mundo 12/12/2010)

5.2.2 pensar [Think] has a very general meaning, roughly equivalent to “cognize” which, according to the context, is construed as a judgement over available evidence or as a personal opinion. […] the evidential dimension does not seem inherently present in the verb. Think is therefore a purely epistemic verb, which can either be interpreted as “in my personal opinion” or “maybe” (Cappelli 2007: 185).

These words can also be applied to Spanish pensar, even though the verb may convey an inferential reading from contextually provided information as in example (239): (238) “[...] Dentro de un solo vino puedes ver no sólo diferentes sabores, sino también filosofías, lugares y personas. El vino griego proviene del corazón del pueblo heleno”, señala Lazarakis por vía telefónica. “Por eso pienso que quien de verdad ame el vino tiene mucho que descubrir en el vino griego”, añade. (El País 22/07/2010) (239) Estudié periodismo, una carrera de inútiles que para mí fue muy útil. Me puse a trabajar muy pronto, a los dos años, de lo que fuera. ¿Qué sabía hacer yo? Escribir, por eso a veces pienso que con la primaria me bastaba. Ahí me enseñaron a poner sujeto, verbo y predicado. Bueno, hace falta cabeza. La herramienta de la cabeza me la formó una familia mitómana. Pura tradición oral. (El País 13/07/2010)

283 (240) “Todo esto viene de algo muy profundo”, explicó la modelo de 40 años. “Viene de otro tipo de desorden emocional porque no se trata de que si no se hace lo que quiero tiro cosas. Pienso que deriva de un tema de abandono y de que he intentado construir una familia con gente que no lo es. Entonces, cuando siento que traicionan mi confianza, me derrumbo. Veo todo rojo”. (El País 07/05/2010) (241) “[...] Yo nunca he pedido explicaciones a nadie. Llegó un momento en que pensé: ‘Hay que vivir con esto [...]’. Pienso que hice lo correcto y que el equipo también obró correctamente”. Ante la insistencia, reconoce un posible error: “No me he dedicado a ser entrenador [...]” (El País 08/05/2010)

Example (239) contains at the sentence level – beside pienso que – por eso, indicating the reason why the speaker thinks so, which clearly signals the inferential use of pensar. Comparing this use to the one in (238), the use is slightly different due to contextual clues: in example (239) the use is inferential because the writer concludes that after having learned how to write he would have learned sufficiently for his professional life. Considering the use of pensar in connection with por eso in (238) we are dealing with a different case: the speaker simply gives a reason for his believing. ‘The Greek wine comes from the heart of Greece, and that is why I think that persons who love wine must see something special in Greek wine’. This use cannot be called inferential because the second clause containing pienso que does not represent a conclusion drawn out of the preceding clause. Examples (240) and (241) are in so far special cases as the speakers refer to themselves when they say to think that [p]. In text passage (240) the model tries to explain why she behaves in a certain way. The speaker in (241) evaluates the current situation he finds himself in, perhaps now evaluating that he had all done correctly (Pienso que hice lo correcto…). Although evidentiality is non-inherent in the semantics of the verb, it can be used in contexts in which evidence for the epistemic judgement is explicitly provided (cf. example 239). Hence, an inferential reading is possible from contextually provided information as also demonstrated in the two following examples. The speaker in example (242) explains that he thinks that ‘multiculturalism has failed’ (el multiculturalismo ha fracasado) and adds a justification for his thinking starting with porque ‘because’. So the speaker has inferred from the fact that ‘multiculturalism is only practiced in the occident’ (se practica sólo en occidente) that multiculturalism has failed: (242) SirCharles 18.abr.2011 | 20:25 #15 Pienso que el multiculturalismo, ha fracasado. Porque se practica sólo en occidente. Los musulmanes, no lo practican. Por eso, el chocque de civilizaciones. Sólo a ZP se le ocurre una alianza de civilizaciones. Por que no entiende nada. (El Mundo 20/04/2011)

In example (243) the causal relation between the speaker’s thinking and the justification for his thinking is established by the double dot (after razón). The

284 speaker thinks that the hackers who have stolen user data from Sony are right because Sony actually does not sell but leases its products: (243) Ayudas web a países que luchan por su libertad como hicieron con Egipto. Ahora Sony? Hay un dilema que cuestionan los hackers que pienso que tienen mucha razón: Sony te vende productos/aplicaciones que pueden ser cambiados en cualquier momento sin permiso. En otras palabras, es un tema de ‘leasing’ mas que de ser dueño por lo que pagas. (El Mundo 27/04/2011)

In comparison to other verbs such as saber, which is said to inherently involve some sort of evidentiality, the evidential meaning is not encoded by pensar. So after having looked beyond the sentence level, the evidential dimension may be described as contributed to certain uses of pensar by contextually provided information, which confirms that an “awareness of processes of extension beyond the boundaries of the sentence can alert us to structures we might otherwise miss” (Mithun 2008: 113).

5.2.3 Excursus: On other pragmatic functions of pensar In the preceding chapter the epistemic and evidential uses of pienso (que) were analysed. As this chapter aims to illustrate the functional diversity of pienso (que) beyond these two uses, the form-to-function approach is adopted. Furthermore, this short study on the different functions of one and the same construction will have a quantitative side besides its qualitative character. Even though pienso (que) is a frequently used construction, the Spanish speaker does not always ‘think in the same way’ when it is used: sometimes, the speaker realises or has realised the cognitive process itself as in the following examples which represent the ‘canonical’ use of pensar, that is, the act of thinking, the cognitive process itself: (244) Si pienso en un futuro... en cómo seré dentro de 10 años, no veo nada, y eso me angustia como nada. No puedo decir “tendré una casa” o “siempre he querido tener dos hijos”. Mi generación es la generación que ya no sueña. (El País 29/09/2010) (245) Después de conseguir el scudetto, con que si era mentira que tenía pie y medio en el Madrid. “Tengo los dos en el Inter... y sólo pienso en la final de la Champions. Cuando termine, me tomaré un par de días para mí y valoraré qué es lo que me hará más feliz”, dijo a pie de campo en Siena. (El País 26/05/2010) (246) “El trabajo me distrae, no es un sacrificio. Pienso horas en cada proyecto que hago. Ahora estoy haciendo un acuario dentro del agua. Nunca nadie hizo un acuario dentro del agua”, afirmó en relación al Acuario de Buzios [...] (El Mundo 17/12/2010)

285 At other times the speaker uses pienso (que) to modalise – and sometimes even ‘evidentialise’ – a certain proposition and in other instances the construction is again used to fulfil a particular pragmatic function. These are the uses that are of interest for the present excursus-chapter. In order to give a broad overview of the different functions of pienso (que) the data were retrieved from CREA (http://corpus.rae.es/creanet.html), more precisely, from the section debates ‘talk shows’ and entrevistas ‘interviews’ of the oral part126 of peninsular Spanish127. In summary, CREA provides 195 results – 78 in debates (in 22 documents) and 117 in entrevistas (in 40 documents). In analysing the different meanings of one and the same form that are associated with different functions, the semantic access and the pragmatic access are combined (cf. Travis 2005). The oral part was chosen because constructions such as pienso (que) are clearly more found in spoken discourse than in written discourse because it represents a linguistic means to express the speaker’s validative attitude concerning [p] or to evidentially mark [p]. Pienso (que) is furthermore said to be used as a discourse marker or pragmatic marker. Aijmer/Simon-Vandenbergen (2006: 2) differentiate between these two notions as follows: Discourse marker is the term which we use when we want to describe how a particular marker signals coherence relations. Pragmatic markers as we see them are not only associated with discourse and textual functions but are also signals in the communication situation guiding the addressee’s interpretation.

In contrast to Aijmer/Simon-Vandenbergen – and, as will be shown, in contrast to Travis (2005) and Blakemore (2002) – Schiffrin (1987: 328), referring to certain linguistic conditions, defines discourse markers according to the following criteria128: 1. syntactically detachable from the sentence; 2. commonly used in initial position of an utterance;

126 The data from oral discourses were transcribed without having added any information about intonation etc. However, the transcribed data may at least be said to be ‘oral in character’. 127 I am aware of the fact that the utterances analysed here were probably not all pronounced by speakers of peninsular Spanish. There could have equally been a few Latin American native speakers in a talk show, for instance, who thus spoke another Spanish variety. 128 The feature that refers to prosody and phonological reduction of a discourse marker cannot be taken into account in the present study and is thus not included in the criteria list.

286 3. may operate at a local as well as global level of discourse, that is, at different discourse levels; 4. does not have conceptual meaning (cf. also Fraser 1999: 931) or a vague meaning, or is reflexive of the language or the speaker. In the present study – as will be shown later – the feature of being reflexive of the speaker is not considered as a feature of discourse markers but as a feature of cognitive particles. The notion of cognitive particle should be introduced here in order to describe uses of pienso (que) as in the following example: (247) – Hombre, no; pero, pienso, pienso, pienso que ya, ya vamos de eso, camino de la jubilación... ¡Ay! la jubilación del Consejo ¡qué emoción! [...] (Habla Culta: Madrid: M11129)

The notion of cognitive particle is applied to those uses of pienso (que) where it is inserted to fulfil the function of giving the speaker more time to structure or formulate his utterance. In example (247) the first two uses of pienso obviously fulfil such a function. At the moment of speaking, the speaker does not yet clearly know what he wants to say. If a speaker uses pienso (que) to refer to the interlocutor’s attitude, pienso (que) fulfils the function of a pragmatic marker because this use depends on the pragmatic, communicative situation. According to Aijmer/Simon-Vandenbergen, a pragmatic marker has no influence on the propositional content: “if a word or a construction in an utterance does not contribute to the propositional, truth-functional content, then we consider it to be a pragmatic marker” (2006: 2). This definition is also applicable to the use of pienso (que) as a cognitive particle but it is not applicable to the use of this construction that functions as a marker of the speaker’s epistemic/evidential stance (Englebretson 2007: 3): (248) “[...] Nunca he visto a un compañero cometiendo este tipo de acciones y no sé si es porque soy un cándido y no me entero de nada o porque a mi alrededor nunca ha pasado, pero sinceramente pienso que son hechos aislados que se dan en el deporte, aunque cuando se producen tienen una repercusión grande”, ha comentado. “Me parece imposible o casi imposible que haya dopaje en el mundo del fútbol y lo digo de corazón”, ha añadido. (El Mundo 16/12/2010) (249) [Comentario] Contar las noticias a medias ... Es como cuando mi hijo me dijo que un viernes fué al colegio, y no mintió pero tampoco entró. Si, amigo, te creo, pero sinceramente pienso que no eran los lugares para estar el candidato a Presidente del Gobierno, cuando se ha cerrado el espacio aéreo español, mas recogidito en su Hotel y en contacto, hubiera sido lo prudente y sensato 5. (El Mundo 17/12/2010)

129 This example was taken from the Corpus del Español in order to demonstrate the use of pensar as a cognitive particle. See http://www.corpusdelespanol.org/.

287 Examples (248) and (249) represent those uses of pienso que which convey the speaker’s validative attitude concerning [p]: in example (248) son hechos aislados que se dan en el deporte [...] and in example (249) no eran los lugares [...]. The speaker ‘sincerely’ (sinceramente) conveys his attitude. Example (250), in contrast, represents an inferential use of the construction under consideration: (250) Según ha explicado “tienen en sus manos” las propuestas del departamento para evitar el cierre del espacio cultural y la Consejería aguarda su respuesta. “Creo que es una negociación muy importante y muy sensible para todos y, por lo tanto, pienso que nos compete ser discretos en los términos de la negociación”, ha dicho. La consejera se ha mostrado satisfecha porque el anuncio del posible cierre del museo de Hernani haya ‘disparado’ el número de visitas. (El Mundo 15/12/2010)

The speaker utters that ‘he therefore thinks that [p]’. The utterance before es una negociación muy importante y muy sensible para todos – even though it is not a factual claim because of creo que – leads the speaker to the conclusion that [p], which is introduced by por lo tanto pienso que [...]. In contrast to cognitive particles and pragmatic markers, discourse markers are defined by having influence on the propositional content. That is why “it is certainly not the case that speakers can simply slot a [discourse] marker in whenever they might need to organize their thoughts” (Travis 2005: 1). Discourse markers structure the discourse and contribute to discourse coherence (cf. Schiffrin 1987: 326). The feature of coherence is also addressed by Fraser (1999: 931): “With certain exceptions, [discourse markers] signal a relationship between the interpretation of the segment they introduce, S2, and the prior segment, S1”. As mentioned above, according to Schiffrin (1987), discourse markers are also reflexive of the language or the speaker and she analyses the uses of oh, well, and, so, now, because, but, or, then, you know and I mean, while oh, well, then, now, I mean and you know are described as ‘particles’ (without specification) and so, because, and, but and or as ‘connectives’. This obviously illustrates that it should be better dealt with different classes of discourse markers as Schiffrin (1987) does or it should be dealt with different classes at all as Aijmer/Simon-Vandenbergen (2006) do because they distinguish between a pragmatic marker and a discourse marker, without subsuming both below the heading ‘discourse markers’. As it is the aim of this chapter to demonstrate the different uses or functions of pienso (que), it will be distinguished between its uses as discourse marker, pragmatic marker, cognitive particle and marker of the speaker’s epistemic/evidential stance. The latter use of I think was also analysed by Kärk-

288 käinen (2007), who describes I think as one expression of the speaker’s epistemic/evidential stance: […] American English speakers use a limited set of high-frequency markers in everyday speech for their expression of epistemic/evidential stance. It is therefore conceivable that many of these frequent stanced items have specialized into some routine function(s) in the interactional organization of conversation. This is the case with […] I think, which in everyday American English has been shown in first-position turns to routinely frame an upcoming stanced turn or longer opinion sequence (Kärkkäinen 2007: 183; cf. also Kärkkäinen 2003).

In the corpus analysis, the epistemic use of pienso (que) or rather its use to express the speaker’s validative attitude towards the proposition and the inferential use of pienso (que) will be summarised below the heading ‘the speaker’s stance’. This will only be done in order to simplify the quantitative work. Qualitatively, if possible, it should be tried to differentiate between an epistemic use of pienso (que) and an evidential use. The following example of pienso que, for instance, demonstrates an inferential use: (251) Algo así ha pasado con los regímenes “igualitarios” que han querido imponerse en los últimos cien años a lo largo y ancho del mundo. Por eso pienso que la única razón de ser de la política en lo que se refiere a la igualdad está en la necesidad de limar las asperezas de un camino por donde pueda avanzar la libertad. (El País 24/07/2010)

The speaker in example (251) uses pienso que inferentially because he justifies why he thinks what he thinks using por eso (compare to example 250). The speaker in example (252) uses pienso que to express his validative attitude towards [p], whereby the first use of pienso que may be accompanied by an inferential moment because it is used with deber de, which represents one linguistic means to indicate an inference (cf. Dendale 1994). Example (253) could be analysed in similar terms even though poder is used to convey a conclusion of lowered certainty in comparison to pienso que + deber de (cf. Silva-Corvalán 1995 and Tasmowski/Dendale 1994): (252) No pretendo defenestrar a todo el colectivo de los críticos, ni hablar, pero definitivamente hay unos cuantos especímenes que parece que en vez de sangre tengan bilis. De algunos de ellos pienso que deben de ser novatos y que, por eso, son atrevidos y no saben respetar el trabajo y el sacrificio ajenos. De otros pienso que son demasiado veteranos y que han acabado odiando su trabajo por culpa de la rutina y la saturación. De otros no sé ni qué pensar. Y de todos ellos pienso que, como Edward Lewis, no construyen nada con su trabajo. Y eso es algo que no me gustaría que se dijera de mí. (El País 29/09/2010) (253) [...] el piloto mallorquín rubricó su primer triunfo en la categoría reina del motociclismo la temporada pasada tras superar a Rossi en la octava vuelta y dejar

289 atrás al italiano hasta la bandera de cuadros final. “Me gusta y además pienso que puede ser grande para nosotros esta temporada”, asegura Lorenzo. “Es la casa de mi equipo y para mi es siempre un orgullo pilotar en Motegi”, añade el español. (El País 01/10/2010)

Discourse markers are said to structure the discourse and to establish coherence between parts of discourse. Cognitive particles are used – to take up Travis’ words again – whenever speakers might need to organise their thoughts (cf. Travis 2005: 1). It is obvious that the discourse markers entonces, pues, o sea and bueno (analysed by Travis) cannot be inserted if the speaker only wants to gain time to structure his thoughts. Hence, it should be differentiated between discourse markers and cognitive particles. In contrast to a discourse marker, a cognitive particle may be inserted if the speaker would like to have some more time to organise his thoughts and to structure his ‘own discourse’. Consequently, cognitive particles have a vague meaning and are reflexive of the speaker (cf. Schiffrin 1987). The use of cognitive particles is cognitively motivated, while the use of discourse markers is bound to the discourse structure and the use of a pragmatic marker depends on the pragmatic, communicative situation. Pragmatic markers are characterised by the fact that they serve a particular communicative function such as referring to the interlocutor’s stance. A pragmatic marker is thus what Aijmer (2002) describes as a “hedging particle” with reference to a special use of I think: Hedging particles are of special interest since they are an important resource for the realization of politeness strategies, I think can, for instance, be used as a strategy redressing an action threatening the hearer’s negative face such as criticism or advice (Aijmer 2002: 8).

So in those contexts, where pienso (que) is used as a resource for the realisation of politeness strategies, pienso (que) will be analysed as a pragmatic marker because this use highly depends on the communicative situation, that is, on the speaker-hearer constellation.130 In this short study, I will try to differentiate, on the one hand between pienso (que) as a linguistic means that gives voice to the speaker’s epistemic/evidential stance and pienso (que) as a discourse marker, pragmatic marker or cognitive particle. On the other hand, it will be distinguished between the three last mentioned uses. It goes without saying that in this connection, the consideration of the context that surrounds a particular token of pienso (que) is of high importance. This 130 I am aware of the fact that, for all practical purposes, this strict distinction between the different uses is not always possible. Nevertheless, the distinction made here illustrates that pensar is used to serve various functions.

290 is also emphasised by Kärkkäinen (2003: 111; cf. chapter 5), who analyses the different meanings of I think. After having analysed only a few examples from CREA, it has become clear that there also exist ‘transitional uses’ of pienso (que) between the one and the other use or between the one and another one. It seems reasonable to assume uses of pienso (que) where it is impossible to distinguish between its use to express the speaker’s epistemic/evidential attitude towards [p] and its use as a pragmatic marker. The following three examples are comments which were found in a chat forum where readers of the newspaper El Mundo make comments on different newspaper articles and discuss them. In such a context it is sometimes very difficult to distinguish between uses of pienso yo or pienso que to express the speaker’s stance and uses that are pragmatically motivated. On the one hand, the following uses express the speaker’s stance but on the other they refer to other comments made previously: while in example (254) pienso yo is posposed to [p], where it obviously fulfils the function of a pragmatic marker, pienso que in examples (255) and (256) is found in initial position and explicitly refers to algunas personas de este foro (‘some persons of this forum’) or varios comentarios (‘various comments’). Hence, examples such as the following three, where it seems to be difficult to differentiate between pienso yo/pienso que as a stance marker and a pragmatic marker have led to the assumption of ‘transitional uses’: (254) [Comentario] La verdad es que sí tuvo mucha suerte pero las condiciones eran igual para todos, que nadie corría en seco. Algo de habilidad tendrá también [,] pienso yo. (El Mundo 27/10/2010) (255) [Comentario] Al contrario que algunas personas de este foro, pienso que Moratinos ha sido uno de los mejores, si no el mejor ministro de asuntos exteriores de España. (El Mundo 25/10/2010) (256) [Comentario] Soy burgalés, y me siento orgulloso de mi tierra. He leído varios comentarios y algunos me han parecido bastante sensatos; otros no tanto. Factores hay muchos, todos influyen, el clima, la cultura, etc. pero pienso que hay uno que es un aval. En Castilla y León la familia es más tradicional, más conservadora. La familia es el motor de nuestra educación. (El Mundo 09/12/2010)

Nevertheless, if possible, it will be strictly differentiated between pienso (que) as a linguistic device to express the speaker’s attitude, as cognitive particle and as discourse or pragmatic marker. It will be additionally demonstrated that the introduction of the term ‘cognitive particle’ is necessary if dealing with a particular use of pienso (que). To sum up, the construction pienso (que) obviously fulfils various functions, whereby its use as a discourse marker or cognitive particle cannot or can only rarely be found in journalistic discourse because it is not predominantly oral in

291 character. If the journalist inserts direct speech, he mostly quotes only one or two sentences. He thereby possibly ‘deletes’ discourse coherence that was formerly established by the speaker. The use of pienso (que) as a cognitive particle was also not found, maybe due to the following reason: usually, an interviewee deliberates carefully in answering the questions and does not start talking without previously having cautiously thought about his answer. Hence, it is necessary to systematically analyse the different uses of pienso (que) with the help of the corpus CREA, that is, with the help of 78 results that were found in talk shows and 117 results that were found in interviews. So I will work with 195 results of pienso (que) which shall be divided – relying on the former analysis – into the following uses: 1. marker of the cognitive process itself, 2. marker of the speaker’s stance, that is, an inferential or epistemic use, 3. pragmatic marker, 4. discourse marker, 5. the speaker’s stance and pragmatic marker and 6. cognitive particle. After having searched for pienso in debates it became clear that, in the majority of cases, pienso appeared with que, that is, as the construction pienso que. The 78 tokens of pienso resulted in the following distribution of uses/functions: Table 16: The distribution of the different functions of pienso in talk shows TOTAL

78

100%

Marker of the cognitive process Marker of the speaker’s stance Pragmatic marker Discourse marker The speaker’s stance and pragmatic marker Pragmatic marker and cognitive particle Cognitive particle

6 50 7 – 11 1 3

7.69% 64.10% 8.98% – 14.10% 1.28% 3.85%

As the table above demonstrates, it was necessary to introduce another function of pienso (que), namely where it serves the function as a pragmatic marker as well as a cognitive particle: (257) Pero tú me reconocerás que tú has sido muy apoyada probablemente por lo que has dicho, por tus propias creencias, ¿no? . Bueno, yo yo yo pienso que desde el primer momento en que a mí me plantean de la noche a la mañana que mi hija tiene una leucemia linfoide aguda, me quedé tan pegada que dije: “Se muere”. Sin más. (Debate: la eutanasia, 04/06/87, TVE 1)

292 The different forms that refer to the interlocutor (tú, ¿no? etc.) in a talk show about euthanasia indicate, on the one hand, that the use of pienso que is pragmatically motivated, but on the other hand, pienso que seems to fulfil the function of a cognitive particle: the speaker talks about a situation he himself experienced; he speaks about the moment when he was told that his daughter suffers from leukaemia. That is why it does not make (real) sense to introduce [p] by pienso que. The speaker knows exactly how he reacted in that situation and what he said. The use of pienso que does not only fulfil the function of a pragmatic marker but also the function of a cognitive particle, which is emphasised by the triple use of the personal pronoun yo ‘I’. The speaker is highly emotionally involved and nervous so that he also inserts yo yo yo pienso que to gain a little more time in order to be able to structure and organise his thoughts.131 The following examples contain uses of pensar to mark the cognitive process of thinking. Hence, pienso is used with the preposition en: (258) Es decir, por ejemplo, asistimos actualmente, y en esto doy la completamente la razón a Octavio a a episodios, que en cinismo político, absolutamente terrible, pienso en en lo ocurrido con el el Irangate, estas extrañísima relación que va del presidente Reagan, el Estado de Israel, un mil multimillonario saudí que vive en Marbella [...] (Debate: el compromiso de los intelectuales, 18/06/87, TVE 1)

The uses of pienso que in examples (259) and (260) fulfil the function of marking the speaker’s stance. The speaker in example (259) as well as the speaker in example (260) indicates the reason why he does (not) think what he thinks. That is why the uses can be described as conveying an inferential reading. Contextually provided information illustrates that the uses are inferential. The speaker in example (259) ‘relies on real facts’ (basándome en hechos reales) and the speaker in example (260) justifies his ‘non-thinking’ that [p] indicating the reason with the conjunction porque: (259) Yo creo que que ba basándome en hechos reales Sí. pienso que cincuenta mil soldados en do en dos, tres, cuatro horas, no pueden abandonar un país. La cosa se calienta, hay polémica. (Radio, Madrid, 01/03/91, Cadena COPE) (260) Pero pero también es cierto, yo a veces les pregunto, ¿y está dando resultado eso realmente? es decir, estamos en el teatro tal, haciendo, y dicen: “Sí sí sí, da mucho y tal”. A la hora de la verdad yo pienso que no, porque la verdad que cuando una obra no funciona, a lo mejor funciona dos días más o tres días más porque lo han dicho en televisión, pero no más. (Televisión, Madrid, 08/03/91, Antena 3)

131 Another possible interpretation of the triple use of the pronoun is that this use could be interpreted as expressing the speaker’s stance because the speaker is highly emotionally involved. Nevertheless, the interpretation given above is considered to be more likely.

293 At this, it should be emphasised that every use of pienso (que) is – at least a little bit – pragmatically motivated. The use of this construction always depends on the communicative situation and thus on the speaker-hearer convention. But – and this is what distinguishes the use of pienso (que) as a marker of the speaker’s stance from a pragmatic marker – both these components are differently strong. In the two following text passages the use of pensar is clearly predominantly pragmatically motivated because the speaker uses (yo) pienso (yo) “as a strategy redressing an action threatening the hearer’s negative face such as criticism or advice” (Aijmer 2002: 8): (261) Es que nosotros estamos aquí hablando y ya casi todo el tiempo de un aspecto solamente, porque la ley penitenciaria dice hablando de la finalidad de las prisiones dice que un aspecto es la custodia. Y el la cosa material de la custo que lleva la custodia, ¿no? Y otro aspecto es la rehabilitación. Bueno, estamos hablando todo el tiempo de la custodia y estamos olvidando la la segunda parte, pienso yo. (Vida y muerte en las cárceles, 05/02/87, TVE 1) (262) Usted ya me entiende . Bueno, ¿qué le diría a aquellos carniceros que crían mala fama, por ejemplo, descuartizando a su mujer? Yo creo que no se ha dado ningún caso, yo por lo menos pienso que bueno Ha habido crímenes donde se ha . demostrado una gran habilidad Sí para descuartizar Sí, bueno, pero te quiero yo pensaba que decías que si ha sido el que ha hecho el crimen un carnicero. Y yo pienso te digo que no no no sabía nada. (Hablando se entiende la gente, Madrid, 17/01/92, Tele 5)

In example (261) the speaker implicitly criticises his interlocutors mentioning that nobody considers ‘the other side’ or ‘the second part’ (la segunda parte). The posposed pienso yo thus predominantly serves the function as pragmatic marker.132 The use of yo pienso in example (262) could be explained by the words quoted above: “Pragmatic markers […] are also signals in the communication situation guiding the addressee’s interpretation” (Aijmer/Simon-Vandenbergen 2006: 2). Obviously, yo pienso in example (262) serves exactly this pragmatic function. The uses in the following two examples are considered markers of the speaker’s stance and pragmatic markers. In contrast to the (pure) markers of the speaker’s stance whose use is always also pragmatically motivated, the pragmatic component and the stance component are equilibrated in examples (263) and (264): (263) Imagínate, por ejemplo si, bueno, si los medios de información dijeran, cuando hay un atentado, que eso parecía el Parlamento Español, cómo cómo se 132 Clearly, because of the fact of being posposed it may also be regarded as a stance marker. However, in the context above the function as a pragmatic marker is considered to be more likely.

294 pondría, ¿no? El nuestros representantes españoles. Por supuesto. Entonces quiero decir o periodistas, en fin, yo pienso Es que yo creo que son los académicos de la lengua los que los tendrían que arreglar, ¿no? Bueno, pues lo deberían de arreglar, y Hay que buscar otra acepción. (Hablando se entiende la gente, Madrid, 17/01/92, Tele 5) (264) Esta parte debe ser un chiste más o menos Negro. negro. Pero en fin, que no deja de ser una pista, Lo que quiero decir El problema, pienso yo perdón. es que hay que tener en cuenta que cuando yo yo yo tengo aquí un párrafo que yo creo que merece la pena citarlo, y es que cuando el presidente de la Sociedad de Eutanasia de Gran Bretaña visita el hospital donde trabaja la doctora Saunder? [...] (Debate: la eutanasia, 04/06/87, TVE 1)

The use of pienso que in example (263) has – at first glance – a lot in common with the use of yo yo yo pienso que in example (257), here reproduced as example (265): (265) Pero tú me reconocerás que tú has sido muy apoyada probablemente por lo que has dicho, por tus propias creencias, ¿no? . Bueno, yo yo yo pienso que desde el primer momento en que a mí me plantean de la noche a la mañana que mi hija tiene una leucemia linfoide aguda, me quedé tan pegada que dije: “Se muere”. Sin más. (Debate: la eutanasia, 04/06/87, TVE 1)

The use of yo yo yo pienso que in example (257/265) was analysed as serving two functions, that is, as being a pragmatic marker as well as a cognitive particle. In example (263) the utterance yo pienso Es que yo creo que expresses the speaker’s validative attitude concerning [p] (son los académicos de la lengua los que los tendrían que arreglar) as well as it refers to the interlocutor’s stance asking ¿no?, which obviously requires an answer or at least an answer with a nod. In example (257/265) the speaker also requires at least a nod of the interlocutor’s head but he does not use yo yo yo pienso que to express his attitude towards [p] but as a cognitive particle as he talks about the situation he himself experienced. That is why different uses of the construction under consideration were applied to examples (263) and (257/265). In example (264) pienso que is used as a pragmatic marker and as a marker of the speaker’s stance. Pienso que serves a pragmatic function because it is used with the phrase hay que tener en cuenta que ‘it should be considered that’ which obviously refers to the interlocutor(s). The speaker additionally inserts a perdón ‘sorry’ showing that now it is the speaker’s turn to express his thoughts about what the problem is, according to him. In the following two examples pienso clearly functions as a cognitive particle: (266) Bien, esto son muchas frases las que se han oído aquí ya, esta noche, y yo creo que es el momento de pienso, de de llevar el tema a la prisión y no llevar el tema a

295 la teoría general de la sociedad, a los fundamentos de la sociedad, a la estructura de la sociedad, etcétera etcétera. (Vida y muerte en las cárceles, 05/02/87, TVE 1) (267) Habla un funcionario de instituciones penitenciarias. La que podría tener cualquier otro que conoce el tema. No no no. O sea, la Dirección General está en relación pues con con con sindicatos, está negociando, está pactando, está discutiendo con ? para que un funcionario que habla, lógicamente, pienso, en nombre en nombre propio y, por lo tanto, de representatividad Por supuesto, y asumo lo que digo. de representatividad, ninguna. (Vida y muerte en las cárceles, 05/02/87, TVE 1)

Examples (266) and (267) do not contain pienso in connection with the conjunction que which is followed by a subordinate clause. By contrast, pienso syntactically represents an insertion; pienso is interjectional. This use is clearly to be distinguished from a parenthetical use of pienso: Aijmer (2002) explains that the absence of the conjunction that in a similar use of I think resulted in the grammaticalisation of the phrase: “To explain the grammaticalization of I think it is assumed that the distinction between main clause and subordinate clause has first become fuzzy as a result of the deletion of that […]” (Aijmer 2002: 17-18). Thompson/Mulac (1991) also explain that the so-called ‘epistemic parentheticals’ are grammaticalised versions of the main clause and offer the following examples a) I think that exercise is really beneficial. b) I think exercise is really beneficial. c) Exercise is really beneficial, I think (Thompson/Mulac 1991: 313)

to illustrate that the matrix clause and the subordinate clause exchange their functions. Example (c) in particular demonstrates that the information ‘exercise is really beneficial’ is now converted into the main clause, while in example (a) it is the information of the subordinate clause.133 The following example represents a use of pienso which is syntactically similar to example (b): 133 Glikman’s study of the Old French construction ço crei ‘je crois; I believe/think’ highlights the importance of the consideration of intonation in the analysis of parenthetical uses of verbs of cognitive attitude. Glikman (2010: 29) shows that the construction under discussion can “apparaître soit en construction incise, soit comme forme rectrice de proposition subordonnée complétive”, that is, the construction can govern a subordinate clause as well as represent a subordinate clause. In Old French, the construction je crois tu viens ‘I think you come’ is ambiguous, whereby the syntactic ambiguity can be solved if applying prosody as the criterion of analysis (cf. Glikman 2010: 32; cf. also Dehé/Wichmann’s study 2010). Glikman thus studies the same sentence structures as Thompson/Mulac (1991: 313): (a) je crois P

296 (268) Durante casi cuarenta años usted ha estado exiliado, es prácticamente ha sido el decano de los exiliados latinoamericanos. ¿Por qué causas si usted no ha pertenecido nunca a un partido político, por qué causas se exilió? Yo pienso hay seguramente varias hipótesis y generalmente las hipótesis de los que me expulsaron del país serán muy distintas de las mías. (Entre líneas, 16/04/90, TVE 1)

In contrast to a parenthetical use of pienso as in example (268), pienso in examples (266) and (267) is interjectional: it is inserted as a cognitive particle that gives the speaker more time to structure and organise his thoughts. That is why it should be distinguished between a parenthetical use and a use as a cognitive particle. To sum up, the data retrieved from transcribed talk shows have illustrated that the predominant function of pienso que is to mark the speaker’s epistemic/evidential stance. Pienso (que) which serves the function of a discourse marker, in contrast, was not found. I could not find any use where it functions to structure the discourse, that is, where pienso que is used to establish discourse coherence. In talk shows – if it is an organised talk show – usually, only one person speaks. This implies that the other participants then have time to think about what to say next and how to formulate their opinion or their stance towards a certain state of affairs. Furthermore, talk show participants have usually had time to previously acquaint themselves with the subject under discussion. This could explain why pienso que as a (pure) cognitive particle was only found three times. The most interesting fact is, if used to mark the speaker’s stance, the speaker has already realised the cognitive process of thinking; nevertheless, pienso que is always used in the present tense. Just as in talk shows, pienso mainly appears in connection with the conjunction que in interviews. 117 results were retrieved from CREA and also classified (b) je crois que P (c) P, je crois. It is essential to differentiate between the syntactic structures found in (b) and (c): while in construction (b) the conjunction que establishes a relationship of dependency between je crois and [p] so that a ‘typical’ relation between main clause and subordinate clause is represented, in construction (c) [p] is not represented as subordination (cf. the terminological problem addressed in this connection by Thompson 2002: 155; cf. also chapter 3.3). Glikman (2010: 35-37) then examines whether the different syntactic structures correlate with a particular prosody and shows that both construction types are bound to a particular intonation contour. Hence, Glikman’s study shows that the criterion of intonation may represent an “indice supplémentaire” for the syntactic analysis of verbs of cognitive attitude – including their parenthetical use (Glikman 2010: 40). So for an analysis of the parenthetical use of verbs of cognitive attitude, the study of spoken data seems desirable and is considered most appropriate.

297 according to the different uses illustrated above (this time I already include its use as a pragmatic marker and cognitive particle): 1. marker of the cognitive process itself, 2. marker of the speaker’s stance, that is, an inferential or epistemic use, 3. pragmatic marker, 4. discourse marker, 5. the speaker’s stance and pragmatic marker, 6. cognitive particle and 7. pragmatic marker and cognitive particle. The following distribution of the results followed from the results obtained: Table 17: The distribution of the different functions of pienso in interviews TOTAL

117

100%

Marker of the cognitive process Marker of the speaker’s stance Pragmatic marker Discourse marker The speaker’s stance and pragmatic marker Pragmatic marker and cognitive particle Cognitive particle

7 77 – – 19 1 13

5.99% 65.81% – – 16.24% 0.85% 11.11%

Comparing the frequency of use of pienso (que) as a cognitive particle in talk shows with the frequency of use in interviews, it is obvious that pienso (que) as a cognitive particle is four times more frequent in interviews than in talk shows. It is assumed that this is due to the fact that interviewees answer more spontaneously than talk show participants. The following three text passages contain pienso (que) where it functions as a cognitive particle: (269) No quiere decirse que no haya un problema, naturalmente. Lo cual no quiere decir Sí. otro tema, Fernando. Sí. A un a la alternativa de gobierno en España se le podría plantear un pienso, con infinitos temas. Yo de todas formas quería plantearle tres muy concretos. Sí. (Televisión, Madrid, 14/01/92) (270) [...] pues bueno, era un tema muy difícil y esto, solamente por este motivo nosotros a través de la unidad de atención al paciente que está haciendo una labor maravillosa en el hospital que y que se han dado a conocer como tales, porque es lo que decía antes la compañera, un pienso que cualquier la importancia de cualquier servicio de cualquier profesión está en relación directa a lo a lo que se da a conocer y a lo que se hace servir de él mismo para darse a conocer. (Radiografía, Madrid, 13/03/91, RNE, Radio 5) (271) Por encima de todo sitúo a la a la cultura, en primer lugar. Pienso que la cultura es el motor actual del mundo, creo que pienso que con la cultura, en el

298 futuro de la humanidad, se pueden lograr todas las cosas que queramos los hombres, inclusive los bienes materiales, a través de la cultura. (Gente de primera, 07/04/94, TVE 1)

Pienso (que) in examples (269) and (270) does not only seem to be inserted but also to cause the initial part of the utterance to be completely interrupted: after having inserted pienso (que) the speaker continues with a totally different syntax. This emphasises the fact that the speaker uses pienso (que) because, at the moment of speaking, he is still organising his thoughts. The uses in examples (269) and (270) especially demonstrate that it is important to differentiate between a use as a cognitive particle and a use as a discourse marker. In example (271) the use of pienso que seems to be superfluous as the speaker also uses creo que to introduce his information. In the three examples above pienso (que) does not structure the discourse. In contrast, the speaker destructs it as in the following example as well: (272) Sí. La situación que tiene ahora, lo que decíamos antes el problema de la defensa, de las inversiones en la defensa son muy grandes necesita bastante solidaridad. Entonces lo que lo que hay que hacer un poco es, yo pienso, como hicieron los los obreros que hablamos antes de Gijón, que ellos estaban en la situación más crítica, estaban en sus mayores luchas contra la patronal y contra la policía de allá y y a a la vez estuvieron consiguiendo herramienta [...] (Búscate la Vida, 15/04/86, TVE 2)

In example (273) yo pienso que is also used as a cognitive particle such as pienso in example (274) which ‘destroys’ the syntactic structure: (273) Mire usted, cuando yo empecé a decir eso hace cinco años, eso fue un escándalo y ahora pues está aumentando de una manera muy notoria el porcentaje de personas que estamos pensando lo mismo con ligeros matices. No sé, yo pienso que, ¿de dónde le viene la potencia al narcotráfico? De es poder puramente económico. . ¿Dónde hemos de presentarles la batalla? En el campo económico. (Radiografía, Madrid, 05/02/91, RNE, Radio 5) (274) Bueno, pues eso, pornográficas o ¿cómo se llama ahora?, de esta otra manera más elegante, ¿no? Pedro Zabaleta de San Sebastián dice: “Yo soy de los que pienso aquello de ‘haz el amor y no la guerra’, en su caso, lo segundo ya lo tengo claro, pero lo primero ¿cómo lo lleva?” No me entero. (La Luna, 21/11/89, TVE 1)

In example (275) the use of pienso que is inferential. The speaker justifies why he thinks that ‘yes’ explaining that he is every day more in a writing mood (cada día tengo más ganas de escribir): (275) [...] yo diría que estoy como como dijo Aristóteles “en la tercera juventud”, no en la tercera edad, que dicen ahora, sino en la “tercera juventud”. Que puede ser, además, la juventud más creativa. ¿Perdón? Digo que puede ser la juventud más

299 creativa. Yo pienso que sí, porque cada día tengo más ganas de escribir. (Radio, Madrid, 07/10/91) (276) Además los países desarrollados debieran saber cómo les ha ido con los bárbaros en la historia, ¿verdad? Entre los países desarrollados Pero son muy egoístas, señor Alfonsín, ¿no le parece? Así pienso, sí. Así pienso. Sobre todo las políticas Ahora lo suyo globales, . A lo de cada uno, a ganar lo más posible y a los que están en vías de desarrollo se les utiliza y ¡adiós muy buenas!, ¿no? (El martes que viene, 26/02/90, TVE 1)

In example (276) the speaker uses pienso to express his validative attitude towards the information los países desarrollados son muy egoístas ‘the industrial countries are very selfish’; he expresses his personal opinion. The speaker is so convinced that [p] that he doubly reinforces his attitude with the help of Así pienso, sí. Así pienso ‘I think so, yes. I think so’. Consequently, the phrase así pienso serves a reinforcing function. The following example contains a use of pienso yo to indicate the speaker’s stance, simultaneously functioning as a pragmatic marker: (277) [...] ¿es una niña que se crea fantasías, que se le pueda haber ocurrido ir sola a alguna parte, haberse caído en algún sitio? Yo pienso que no, porque a pesar de que es una niña muy extrovertida, muy simpática, sin embargo, yo creo que ella no, no es capaz de fiarse de alguna persona que no conoce, o al menos que haya visto pocas veces, vamos, pienso yo. Y y no es una niña de sueños y de fantasías [...] (Ésta es su casa, Madrid, 13/03/91, TVE 1)

In example (277) the speaker is asked whether the girl who is the subject under discussion ‘makes up stories’ (crear fantasías). The speaker is fairly convinced that this is not the case and explains why he thinks what he thinks indicating the reason (Yo pienso que no, porque [...]). While this use predominantly conveys the speaker’s attitude towards [p], the posposed pienso yo seems to convey the speaker’s attitude but also seems to serve a pragmatic function. On the one hand, the speaker conveys his attitude towards [p] (the girl does not make up stories; the girl does not trust in anybody she does not know) and on the other hand, [...] vamos, pienso yo is clearly pragmatically motivated referring to the interviewer, signalling that the speaker is aware of the fact that the interlocutor may have another opinion. Examples (278) and (279) are especially interesting because the speaker emphasises together with the use of (yo) pienso que – which already expresses ‘something personal’ – that he expresses ‘his personal opinion’. If a speaker emphasises that he ‘personally’ thinks that [p], he does so to interact with his interlocutor (or interviewer); he does so explicitly to create a border between him and his interlocutor. That is why the following uses of (yo) pienso que pre-

300 dominantly serve a pragmatic function, that is, in the following contexts they are termed pragmatic markers: (278) ¿No temen ustedes que pueda ocurrir esto en Cuba, en algún momento próximo? Yo creo que habría que profundizar en las... en tratar de identificar las causas de los fracasos de esos proyectos que se presentaban como el llamado socialismo real, que se decía en aquella época. Yo pienso, es mi opinión personal, que la causa del fracaso del llamado socialismo real era la ausencia de un verdadero socialismo. (Los desayunos de Radio 1, 16/06/94, TVE 2) (279) Qué piensan ustedes cuando escuchan reproches tan serios de organizaciones como Amnistía Internacional que denuncian la existencia de presos políticos etcétera, etcétera. Bueno, yo personalmente pienso que no hay un país que se salve de de de ser un país absolutamente democrático y de ser un país absolutamente libre. Creo que la libertad es una cosa tan ansiada por el por el ser humano y tan poco lograda y que en ninguna parte se alcanzan libertades absolutas y pienso que todo el mundo cuando reclama esas cosas siempre tiene un poco de razón, no solamente en mi país sino en todos los países. (Gente de primera, 07/04/94, TVE 1)

Example (280) contains a very interesting use of yo pienso because on the one hand it could represent a cognitive particle but on the other a pragmatic marker: (280) [...] la boca se convierte en una especie de pequeño laboratorio ambulante, ¿verdad?, en la cual se da una reacción química que está absolutamente equilibrada con el organismo. Un poco, yo pienso, lo que tal vez se producía cuando los viejos europeos mascaban tabaco ? Ya sí, sí puede parecerse. o sea, un poco, es decir, tampoco lo mascaban porque era un poco tener ahí dejarlo ahí. (El martes que viene, 01/05/90, TVE 1)

In order to appropriately analyse this example, it would have been necessary to have some information about the prosodic structure and pitch of yo pienso: if it is softly pronounced without being accentuated, it would be considered a cognitive particle; if yo is accentuated and the whole phrase not pronounced softly, it would be considered a pragmatic marker. In sum, there was no use of pienso (que) identified as a (purely) pragmatic marker possibly because in interviews – between interviewer and interviewee – no discussion as between talk show participants emerges. In interviews the distribution of tasks is unequivocally determined: one person asks, the other person answers. Nevertheless, 19 tokens of pienso (que) were identified as transitional uses, that is, uses that convey the speaker’s stance and that serve a pragmatic function because there are interview situations in which the interviewer only functions as mediator while he interviews more than one person simultaneously. In those cases pienso (que) can partly serve as a pragmatic marker if the speaker’s answer refers to another interviewee. Again, no token of pienso (que)

301 fulfilling the function of a discourse marker was found. This may be due to the fact that the answers given in interviews are not long enough to see the necessity of establishing discourse coherence. There simply is no need to connect one part with another using a discourse marker. Actually it is doubtful that pienso que alone (!) is able to function as a discourse marker, in the sense of establishing discourse coherence. In the following two examples it is used with bueno, es que and entonces – commonly known as discourse markers because they establish discourse coherence and relate the former part with that which is to follow. In those cases pienso que might be applied as a discursive function – but only because it is used with ‘real’ discourse markers. Additionally, at the same time it may function as a marker of the speaker’s stance concerning the information that follows, which is clearly not obvious from the following example because the speaker seems to be very agitated, but from the second one: (281) Entonces, una muerte digna porque claro Morir cada uno como quiera, como elija. Bueno, es que yo pienso que una muerte Yo creo sinceramente que ustedes utilizan un término ambiguo. Eso de la muerte digna, deben de cambiar de nombre, porque yo creo que no es aceptable que tengan ese nombre. (Debate: la eutanasia, 04/06/87, TVE 1) (282) Bueno, yo pienso que sí, pero de hecho no debe ser así, porque cada una tiene su cometido. . Entonces pienso que que, vamos, que se diferencia perfectamente, puesto que el a-te-ese, bueno, pues sí, es la que atiende al enfermo junto con la auxiliar. (Radiografía, Madrid, 13/03/91, RNE, Radio 5)

So, in my view, pienso que alone does not (ever?) function as a discourse marker. In this excursus-chapter it was demonstrated that pienso (que) has different functions. One function is to serve as a cognitive particle. It was necessary to introduce this notion because the already existing terms such as discourse marker or pragmatic marker do not adequately describe the use of pienso (que) in those cases where the speaker inserts it to gain more time to structure his thoughts. Furthermore, it was shown that a clear-cut distinction between one use and the other is not always possible. Pienso (que) sometimes serves two functions – as a pragmatic marker and a marker of the speaker’s stance or as a pragmatic marker and a cognitive particle. Communicative situations are very complex and influenced by many factors – especially by the interaction between speaker and interlocutor. In summary, the following results were obtained, whereby the analysis of the tokens of pienso (que) in talk shows and interviews has revealed that pienso (que) is predominantly used to indicate the speaker’s stance:

302 Table 18: The overall distribution of the different functions of pienso TOTAL

195

100%

Marker of the cognitive process Marker of the speaker’s stance Pragmatic marker Discourse marker The speaker’s stance and pragmatic marker Pragmatic marker and cognitive particle Cognitive particle

13 127 7 – 30 2 16

6.66% 65.13% 3.59% – 15.38% 1.03% 8.21%

Cappelli calls the most frequent use ‘metaphoric’, whereby she also demonstrates a range of different uses between ‘the act of thinking’ and the ‘result of the act of thinking’: In the most “literal” case, think is construed as an action, the act of thinking, and, in the most “metaphoric” case, it is construed as the subjective result of the act of thinking, as an indicator of opinion. In between these two extremes, discourse can constrain many construals of the semantic potential of think […] (Cappelli 2007: 194).

In order to demonstrate the parenthetical use of I think, Thompson/Mulac (1991) only use examples where I think appears in the utterance’s periphery. An analysis of uses of (yo) pienso (yo) which distinguishes between this use as an epistemic parenthetical and as a cognitive particle would be very interesting. I speculate that such an analysis is only possible if working with oral, real spoken – not transcribed – data because the study of the different uses certainly depends on intonation and pitch. The following uses of pienso represent such examples that should be the focus of further analyses: (283) [...] jet lag, del desajuste de los biorritmos, de una aventura aérea que en menos de dos días la ha llevado desde Madrid hasta Tokio (...) Esta mujer, pienso, es un encanto (...) Ana Botella, después de plantar un olivo que seguirá allí -si arraigadurante muchas centurias, recorre a fondo [...] (El País 12/12/2010) (284) Mi postura es que tanto el socialismo como el comunismo es una utopía, es decir, la utopía de establecer un paraíso en la tierra, [...], en el cual el Estado tendría una un protagonismo y el gobierno sería ese ejecutor. y es esa utopía, pienso, de u es un no es sólo una utopía socialista y comunista, es un viejo sueño del hombre el el traer el paraíso a la tierra. (Radio, Madrid, 14/01/92 A)

There is no doubt that pienso (que) serves (too) many different functions.

303

5.2.4 saber In ordinary everyday communication, if a sentence bears no explicit marks of evidential-epistemic information, the hearer is allowed to infer that p is a piece of knowledge: in other words, we can assume that in assertions the cognitive attitude expressed by I know is implicitly stated. This brings about the interesting question of why, on occasion, the speaker feels the need to make this attitude explicit (Cappelli 2007: 156).

The following examples will show why (Spanish) speakers feel the need to make their cognitive attitude explicit. They do so, for instance, when they want to express a very high committed epistemic evaluation that a certain state of affairs is the case, in other words, when they want to emphasise their high commitment (example 285); when they express an inference they really believe in (example 289); or when they are able to even give reasons for their knowing (examples 286, 287, 288): (285) Ya sé que esto no es más que un juego, y que solo el hecho de que esa pelotita puñetera entre o no en la portería separa el desengaño de la gloria. También sé que nuestros graves problemas como país siguen ahí con victoria o sin ella, pero el otro día España era un solo corazón ilusionado empujando a esos chicos que se dejaban la piel en el campo. (El País 12/07/2010) (286) El cónsul de Marruecos, Mohamed Said Douelfakar, alabó la “cada vez mayor cooperación” en materia agrícola. José Antonio Martín, de 52 años, dueño de esta empresa de babel, dijo entender bien a los extranjeros. Quizás porque el también lo fue: “Sé lo que sienten porque en 1980 me enrolé como marinero en un barco de Angola. Allí reuní dinero para comprar mi primera hectárea”. (El País 12/07/2010) (287) “[...] Pero uno se pregunta si entre tanta masa gris no deberían incluir a algún descerebrado o a algún poeta. Los tontos y los poetas tienen a veces intuiciones geniales, lo sé por mi experiencia de tonto. [...]” (El País 13/07/2010) (288) “[...] en una crónica de la sección de Cultura, de que voy a publicar un libro en una colección de literatura infantil en la Editorial Alfaguara. Por experiencias muy cercanas, y por libros muy queridos y admirados por mí, sé lo difícil que es escribir para los niños con dignidad y respeto [...]” (El País 03/05/2010) (289) “[...] los pilotos de motos no están majaretas, y digo eso avanzándome a todos aquellos que ahora aprovecharán una noticia tan fantástica como es el regreso de Valentino [...]. Conociéndole un poquito, sé que lo habrá calculado todo antes de tomar una decisión tan trascendental y arriesgada como esta. [...]” (El País 17/07/2010)

In example (285) the evidential dimension of saber is not confirmed by the context compared to examples (286)-(289), where the contextually provided information underlines that saber is a verb that is not only epistemically used but also evidentially. In example (285), there are no clues in the context that the

304 verb is used to express inference or any other kind of evidence. But although not specified, “[r]eference to the availability of evidence, either objectively reliable or subjectively considered as such, is inherent in the semantic potential of the verb” (Cappelli 2007: 166). But the speaker in (289) uses sé que to express an inference, which is indicated by the introducing phrase Conociéndole un poquito ‘Knowing him a little’, as well as by the futuro sintético following sé que. The speakers of examples (286), (287) and (288) give reasons for their knowing, they justify their knowing: the utterances all contain porque or por (mi) experiencia(s). That means that the speakers somehow infer a certain state of affairs from their own life experience. The same is stated for English know: […] the verb know, in its most common qualificational sense, can be considered as a true marker of evidentiality, and its tendency to occur in contexts in which explicit reference to evidential information is made seems to provide support to this hypothesis (Cappelli 2007: 157). The fact that know points towards the evidential domain is also revealed by its frequent co-occurrence with certain evidential expressions, generally referring to experience (Cappelli 2007: 158).

But Vet (1994), who contrasts the French verbs croire ‘to believe/think’ and savoir ‘to know’ with trouver ‘to find (that)’, explains that savoir is not used to refer to experiences made by the speaker: Les verbes trouver, croire et savoir indiquent l’attitude du locuteur à l’égard de la proposition ce film est intéressant. Je trouve présuppose en outre que le jugement du locuteur est basé sur une expérience directe de la chose “elle-même” (Ducrot, 1980). Cette nuance est absente dans le sens des verbes croire et savoir (Vet 1994: 56).

Anyway, the examples above have shown that the Spanish equivalent of savoir indeed is used to mark the speaker’s direct experience so that it is assumed that savoir may be used in a similar way.134 Generally, “[t]he type of evidence that is

134 Vet (1994) – explaining the function of je sais que – contrasts the following utterances and explains that (a) and (b) are fundamentally different: (a) Pierre est marié. (b) Je sais que Pierre est marié. […] en énonçant [(a)], le locuteur […] suppose que la proposition Pierre est marié ne figure pas dans le domaine de connaissances de son interlocuteur […] et invite celui-ci, par l’acte de parole de l’assertion, à l’introduire dans son domaine de connaissances. […] L’énoncé [(b)] n’a pas la fonction d’inviter l’interlocuteur à introduire la proposition Pierre est marié dans son domaine de connaissances, mais de poser que, contrairement à ce que pense ou semble penser l’interlocuteur, le locuteur a déjà assigné la valeur de vérité “VRAI” à la proposition. Des énoncés

305 said to lead to knowledge ranges from objective sensory evidence to very subjective affective evidence” (Cappelli 2007: 166). So in saber, the two dimensions – the epistemic and the evidential one – are present. Consequently, Cappelli’s table of know can also be applied to the use of its Spanish counterpart. When a speaker uses saber, he implies that he has some reason for his knowledge, and this reason can be and often is specified in context but does not have to be. Table 19: The conceptual dimensions lexicalised by know (cf. Cappelli 2007: 166) EVIDENTIALITY know unspecified but reliable

EPISTEMICITY Likelihood Certainty highest possible highest possible degree degree

The following example contains two instances of sé que. The first use illustrates that “the meaning potential of the verb does not include any reference to specific types of evidence” (Cappelli 2007: 159). There is no contextually provided information that makes obvious whether the speaker’s knowledge derives, for instance, from hearsay or inference, although it is assumed that it most likely derives from personal experience. The second use of sé que, in contrast, clearly indicates knowledge from former experiences because the speaker himself is an author; that is why he should know that [p]: (290) “Allí donde fallo yo como hombre, fallan también mis personajes. Por otro lado, ellos sienten orgullo por las mismas cosas que yo, es decir, por los pormenores cotidianos de la vida”, aseguró el escritor checo Bohumil Hrabal. Sé que sobrarán los que tras leer estas palabras me corrijan, los talleristas que me enseñen aquello que no entiendo, los críticos que se apresuren a explicarme lo inexplicable. Antes de que lo hagan, déjenme decir que sé lo fácil que es diseccionar un personaje, un texto, una situación o incluso una palabra, y también lo inútil que resulta. Así que mejor contéstenme cómo es que Tolstói huyó de su muerte, para ser exactos de su casa instantes antes de su muerte [...] (El País 02/08/2010)

Similar to the second use of sé que in the example above is the following one in example (291). It indicates knowledge from former experience: comme [(b)] semblent donc avoir pour fonction de corriger les présupposés de l’interlocuteur (Vet 1994: 65). Vet (1994: 66) therefore concludes that je sais que realises an act of “meta-communication”. This may truly be the case as the speaker who knows that [p] obviously feels the need to emphasise that he knows that [p]. However, working with real language data shows that the reason for the use of I know that is not (always) a meta-communicative message alone.

306 (291) He corrido un Tour con una costilla rota; sé lo que es tener aquí, en la espalda, un puñal clavado, que por la mañana me tenían que llenar la mano de agujas de acupuntura, todos los nudillos... Sé lo que es sufrir encima de una bicicleta, pero en el Giro pensaba que no. (El País 04/07/2010)

The following two examples illustrate that speakers use sé que if they are absolutely convinced that [p], even though [p] is a future event: (292) [...] muchos de ellos aficionados con solera que ven en su figura el último torero de pellizco y duende. “Me lo tomo con muchas ganas. Hay mucha presión, pero sé que haré lo que siento, como lo siento y como siempre he intentado” explica en alto, convenciéndose a sí mismo. (El País 02/08/2010) (293) Lo ve como una oportunidad de mostrar otra cara, más optimista, del continente: “Mi gran deseo es que en África tengamos la oportunidad de demostrar que podemos trabajar igual que los demás. Y sé que estaremos a la altura. Estoy convencido”. (El País 21/06/2010)

The following instance of sé que is used with que si, by which the speaker intensifies that he knows that [p]: (294) Por eso señala la falta de eficacia de la estructura de seguridad del país como una de las variables que más influyen en el auge del crimen organizado. “El sistema federal mexicano es tan federal que si sé que hay un policía que está con los criminales y no tengo la evidencia judicial, simplemente no puedo removerlo”, indicó. En consecuencia, Calderón manifestó su intención de emprender una reforma de la policía para la creación de [...] (El País 12/08/2010)

Cappelli analyses similar uses of I know as introducing a true premise: I know introduces a true premise (or a declaration of the subject that he/she indeed believes p to be true) that creates a common ground135 with the interlocutor. It is as if the speaker/evaluator was anticipating a possible objection by the hearer and he/she was making clear that the “evidence” for such possible objections is available to him/her as well (Cappelli 2007: 163).

Generally, in saying Sé que [p] and thereby expressing a very high committed epistemic evaluation that [p] is the case or thereby expressing an inference, speakers want “to foreground the reasons that they have to support the truth of the propositional content” (Cappelli 2007: 156), even though the reasons are not always explicitly referred to.

135 The term ‘ground’ comes from Langacker, hence from cognitive science. In “Remarks on the English Grounding System” he explains that “[t]he term ground is used for the speech event, its participants, and its immediate circumstances” (Langacker 2002: 29). Nuyts (2002: 438) adds meaningfully that one could also simply speak about the “communicative situation”.

307

5.2.5 suponer Cappelli found out for the English equivalent of suponer that […] suppose, when used as a verb expressing the cognitive attitude of the evaluator, lexicalizes a relatively low committed epistemic evaluation that a certain state of affairs is the case, reached via an inferential process (Cappelli 2005: 240).

So with suppose the speaker expresses an inference he is not totally sure of. The inferential process is a dynamic one so that the conclusion is tentative and provisional in nature. Hence, Cappelli proposes the following table for suppose: Table 20: The conceptual dimensions lexicalised by suppose (Cappelli 2007: 224) EVIDENTIALITY Inferentiality suppose

encoded

EPISTEMICITY Likelihood Certainty positive degree

very low level of commitment

COGNITION dynamic inferential process, usually presented as still ongoing, resulting in a provisional tentative conclusion.

In order to understand whether these words can also be applied to suponer the following examples will be considered: (295) “[...] A mi no me extraña.... siendo como es, sin duda estará el hombre toda su vida dudando de quienes le rodean, sospechando que una eventual mujer solamente lo querrá por su dinero. Es el precio de la fama supongo. En este contexto, lo mejor que puede hacer es [...]” (El País 11/07/2010) (296) P[regunta]. Han jugado más de 75 partidos juntos, entre el Barça y la selección. Y además, son amigos fuera del campo. ¿Eso ayuda? R[espuesta]. Supongo que sí, porque cuando estamos juntos hablamos mucho de fútbol, nos gusta. Somos como una pareja de hecho. (El País 12/07/2010) (297) Cuando Contador le pasa, imagino que Contador ve que algo raro le pasa a Schleck, tal vez haya visto la salida de cadena, pero supongo que habrá pensado “ok si se le ha salido la cadena, tardará 5 segundos en ponerla”. Contador también tenía pensado atacar al final del puerto. No tiene sentido que ya una vez lanzados los ataques, Contador pare, espere a Shleck para volver [...] (El País 22/07/2010)

Analysing examples (295) and (296) we could also conclude for Spanish suponer what Cappelli states for the use of suppose: As far as the evidential information lexicalized is concerned, suppose indicates an ongoing inferential process. In certain cases, it can be used to indicate a suddenly reached conclusion based on inferential work following from the evaluation of available evidence of various sorts (Cappelli 2007: 224-225).

308 The verb in example (295), for instance, is partly used epistemically, partly used inferentially: the speaker assumes that all VIPs have to pay for being famous, that is, epistemically evaluating the state of affairs. On the other hand, the use of supongo may also indicate an inferential process. The fact that the verb is to be found at the right periphery of the sentence may underlie that it is a suddenly reached conclusion that is postpositioned. According to Cappelli suppose “is a verb that can occupy many different positions along the epistemic scale, as it is very sensitive to contextual variables” (Cappelli 2007: 224). The commitment expressed by the speaker in examples (296) and (297) is higher than that in example (295). The use of porque ‘because’ introducing the reason why the speaker supposes what he supposes (example 296), underlines the fact that suponer is also used when the commitment to the state of affairs is fairly high, and the preceding inferential process is somehow justified. Example (297) is a very special one, as it contains a verb of imagination (imagino que…), a modal adverb (tal vez ‘maybe’) + a subjuntivo form (haya visto), the verb suponer + a futuro compuesto form (habrá pensado), an imperfecto form (tenía pensado) and a justification introduced by No tiene sentido que ‘It does not make sense’. In this context these lexical and grammatical means make clear that the speaker imagines having been in Contador’s situation. Therefore, all his utterances are to be understood as a supposition and his personal opinion, and simultaneously as a relatively high committed epistemic evaluation that the certain state of affairs was the case. However, he distances himself via the use of the subjunctive, the synthetic future and the imperfecto, all indicating incertitude concerning the state of affairs under question (using subjuntivo and imperfecto) or inference (using futuro sintético). Consequently it can be concluded with Cappelli: Suppose usually introduces the sudden conclusion to which an evaluator arrives. Such a conclusion is never too carefully thought out, nor is it considered totally reliable. It is an “unstable” evaluation, a provisional conclusion or an intermediate “stopover” where the evaluator pauses for a while, before starting again his/her stream of thoughts. In this sense, suppose differs from both think and believe, which encode subjective but definite conclusions, which, although probably only temporary, come at the end of a complete evaluative process (Cappelli 2007: 218).

The fact that Spanish suponer is also used to express a provisional tentative conclusion illustrates the following example, where the speaker evaluates the current situation explaining that he now notices that ‘it must have been difficult (no fácil) for his parents’ when he left home: (298) Salí de casa con 16 años... me fui al norte, a la Toscana. P. ¿Cómo es la vida del emigrante? R. Dura y más si eres ciclista y más si eres del sur. He sufrido porque he tenido que aprender a vivir lejos de mi familia. Supongo que para ellos tampoco

309 debe haber sido fácil dejar salir a un chiquillo. Con esa edad necesitas apoyos. Me estoy dando cuenta ahora cuando veo a mi hermano pequeño. Ha seguido mis pasos y también se ha ido a vivir a la Toscana [...] (El País 23/09/2010)

In sum, if a speaker supposes that [p] he must have some reason so that the evidential, more precisely, inferential dimension seems to be inherent in the verb. The speaker’s commitment varies from low (cf. example 295) to fairly high (cf. examples 296 and 297).

5.3 Summarising and concluding thoughts concerning the verbs and adverbs The chapter’s goal was to describe the evidential and epistemic meaning aspects of Spanish verbs of cognitive attitude and of Spanish modal adverbs in particular contexts. The importance of taking the context into account when analysing these linguistic elements was highlighted. The study of naturally occurring data in their context has been proven to help to determine which meaning aspects are encoded by a particular linguistic item and which one(s) may be contributed by the context. It should have become clear that, for instance, aparentemente has certain meaning aspects that are assumed to be encoded by the adverb, namely ‘visual evidence’ and ‘inference’, while the ‘reported’ meaning is considered to be contributed by the context. If used to convey a reported reading, aparentemente usually appears with further references to information sources. This can only be found out if having looked beyond the sentence level and having taken the context into account. Aparentemente can also mean ‘by all appearances’, ‘by all indications’ or (simply) ‘seemingly’. The meanings ‘by all appearances’ and ‘by all indications’ could be described in terms of weak inference, while the meaning ‘seemingly’ expresses an ‘even weaker inference’ that could be described in terms of assumption where the modal component is more prominent than the inferential one. If aparentemente conveys a hearsay reading it usually has scope over a whole sentence or at least over a clause. If the adverb has the meaning ‘visual evidence’, ‘by all appearances/indications’ or ‘seemingly’, the adverb’s scope is usually smaller; it is often used to qualify an adjective only. With reference to the use of the other adverbs, such a relation between meaning and scope is not as easy to determine (see below). Generally speaking, in analysing the use of the adverbs it was on the one hand possible to differentiate between the different types of source of information (inference, visual evidence or hearsay) – although sometimes two readings were possible – and on the other hand between higher and lower levels of com-

310 mitment (with regard to the use of probablemente and posiblemente as well as with reference to the use of seguramente). Concerning the meaning aspects of evidentemente, it is assumed that ‘visual evidence’ represents a meaning aspect that is encoded, while ‘inferential’ evidence’ is a meaning aspect that is contributed by the context. In every single use it is not possible to determine the exact meaning as visual evidence and inference may overlap. This is the case if evidentemente means ‘evident from certain indications’ or ‘evident from the situation I can observe’. Obviamente, in contrast, nearly always seems to express ‘as evidence/knowledge of the world shows’ or ‘as observable evidence shows’ so that obviamente clearly is an evidential adverb. Even if no contextual information about the kind of evidence is provided it is evidential. While evidentemente and obviamente are evidential adverbs, the modal component and the evidential one may meet in the use of seguramente. Generally, seguramente is used if the speaker is not (totally) certain that [p] (or that part of [p]). However, the degree of certainty expressed by the adverb may be ‘affected’ by its surroundings such as verbal mood. For instance, if seguramente is used with the indicative (future), the reading conveyed is inferential, while if used with the subjunctive, the reading is more modal and the speaker’s degree of certainty is definitely lowered as if the adverb were used with the indicative. Hence, in certain contexts seguramente should be translated as ‘certainly’ and in other contexts as ‘surely’ – depending on the degree of certainty that is conveyed. In contrast to all the other adverbs analysed in the present study, supuestamente is notably frequently found in contexts treating crimes or juristic questions. In this context supuestamente was shown to combine the reported reading and the inferential one, albeit not the ‘classical’ inferential reading. In the ‘classical’ inferential use of a certain linguistic expression, the speaker himself is the person who drew the conclusion. In the use of supuestamente the speaker/journalist transmits the inferentially gained information of another person so that the adverb combines both readings. Consequently, the evidential meaning aspect is regarded as inherent in the adverb. Obviamente, evidentemente, seguramente and supuestamente were shown to have scope over whole sentences as well as over parts of a sentence. With reference to the use of these adverbs, the relation between meaning and scope could be the focus of later studies. So the question of whether the scope the adverb has in a certain context influences the adverb’s meaning nuances should be dealt with. Concerning the ‘classical’ modal adverbs, probablemente normally expresses a higher degree of certainty that a certain state of affairs is the case than posiblemente, but both can occur in connection with the subjunctive. If so, then the certainty about the state of affairs is lowered. Consequently, comparing the

311 use/function of posiblemente + indicative and probablemente + subjunctive, the former expresses a relatively high possibility and the latter a lowered probability, which – semantically speaking – seems to be highly similar. The scope of probablemente and posiblemente varies: they may function as a sentence adverb or qualify only parts of a sentence. For both adverbs, the inferential meaning may be contributed by the context, even though they are not evidential adverbs. So while ‘classical’ modal adverbs may convey an inferential reading, an evidential adverb such as aparentemente may be used to convey a modal reading – just as real evidentials may have extensions related to epistemic modality (cf. Aikhenvald 2004; cf. chapter 2.1.1). Verbs of cognitive attitude were also shown to encode certain meaning aspects, while other meaning aspects may be contributed by the context. One example is the use of pensar: in contrast to saber, for instance, which is said to inherently involve some sort of evidentiality, evidentiality is not a meaning aspect that is encoded by pensar. It does not inherently lexicalise any reference to specific types of evidential information. Nevertheless, it can convey an inferential reading due to contextually provided information. So the inferential reading represents a meaning aspect of pensar that is contributed by the context. Concerning the use of the verbs of cognitive attitude, one could conclude, on the one hand, that all verbs lexicalise the cognitive attitude concerning a certain state of affairs. But this conclusion could have been drawn at the beginning of the study already. All verbs of cognitive attitude were shown to be sometimes used to convey an inferential reading, but they differ according to the meaning aspects that are encoded and to those that are contributed by the context. These verbs also vary concerning the speaker’s commitment that a certain state of affairs is/was the case. Creer lexicalises the evidential domain beside the epistemic one, whereby it depends on the context as to whether the inferential or modal meaning is more prominent. The speaker has affective evidence for what he believes. Besides the evidence being affective, it may be more or less intersubjectively comprehensible: creer was found to be used in contexts where the speaker explicitly verbalises the reasons for his believing as well as in contexts that lack evidence which is intersubjectively comprehensible. For instance, if a speaker uses creo que in connection with the synthetic future form será without giving further ‘back-up’ for his belief, the use is regarded as inferential but not compulsorily intersubjectively comprehensible (cf. example 225). With reference to chapter 5.2.2 and 5.2.3, there is no doubt that pienso (que) serves (too) many different functions. Besides being a marker of the speaker’s epistemic/evidential stance, pienso (que) also functions as a cognitive particle and pragmatic marker. It is not in every case possible to distinguish between its

312 different functions. In comparison to other verbs such as saber, which is said to involve some sort of evidentiality, the evidential meaning is not encoded by pensar. However, having a look beyond the sentence level, it becomes obvious that the evidential dimension may be contributed by contextually provided information. Hence, even though pensar – such as think – may be described as “a purely epistemic verb” (Cappelli 2007: 185), the use of pensar reveals that it may convey an inferential reading. The subchapter on saber was, above all, concerned with the question of why (Spanish) speakers feel the need to make their cognitive attitude explicit. It was shown that they do so when they want to express a very high committed epistemic evaluation that a certain state of affairs is the case, that is, emphasising their high commitment; when speakers are able to provide reasons for their knowing (cf. also Cappelli 2007: 156 on know) or when they express an inference of which they are convinced. While saber may be used to express an inference to which the speaker is highly committed, with the use of suponer the speaker expresses an inference he is not totally sure of. The inferential process is a dynamic one, which means that the conclusion is tentative and provisional in nature. Consequently, the evidential, more precisely, inferential meaning aspect is considered encoded by this verb because if a speaker supposes that [p] he must have some reason to do so. The study has furthermore revealed that, with reference to the epistemic-modal meaning aspect, the speaker’s commitment varies from low to fairly high. Even though epistemic modality and evidentiality represent interwoven dimensions in the verbs (except in the case of pensar), in analysing the use of a particular verb in a particular context it was finally decided whether to speak either of an inferential use or of an epistemic use, relying on the meaning aspect that was considered more prominent. Thus Cappelli (2007) was right to develop a table for every single verb of cognitive attitude. Cappelli (2005) explains the fact that in verbs of cognitive attitude epistemic modality and evidentiality cannot be easily separated as follows: […] the difficulty in separating the two interwoven categories must derive from the fact that the two semantic dimensions tend to co-occur and to “evoke” each other because, cognitively, they work in strict contact. Our experience of the world tells us that if one holds that something is possible, one must have some sort of evidence justifying this sort of attitude. Conversely, if one has a certain type of evidence available, one tends to epistemically evaluate all the relative states of affairs accordingly (Cappelli 2005: 229).

As epistemic modality and evidentiality work not only in strict contact but also evidentiality and epistemic modality with subjectivity and deixis, they have re-

313 ceived a superordinate category that encompasses them (cf. chapter 2.2). Verbs of cognitive attitude prove that evidentiality and all the other categories should have a superordinate category imposed on them because these verbs represent linguistic devices in which particular categories overlap. Cappelli treats verbs of cognitive attitude “as means for the expression of subjectivity, of epistemic modality or stance” (Cappelli 2007: 53, cf. also 82). As she analyses the epistemic/evidential use of English verbs of cognitive attitude and as she describes “subjectivity as a deictic for the speaking ego” (Cappelli 2007: 131), she does not only relate subjectivity to deixis but also implicitly evidentiality (and epistemic modality) to subjectivity. Understanding subjectivity as a deictic for the speaking ego makes it possible to arrange this ‘deictic understanding’ of subjectivity with the concept of evidentiality because the speaker must be ‘allowed’ to encode reference to the absence of an external source (cf. Cappelli 2007: 131), i.e. to an internal source – to himself as source for [p] so that the speaker provides “internal evidence” (Cappelli 2007: 132). Cappelli labels internal evidence ‘affective evidence’. In affective evidence “subjective judgement itself is assumed to the role of evidence intended as a base for evaluation” (Cappelli 2007: 133). Such affective evidence is, for instance, expressed by particular uses of creer (cf. chapter 5.2.1). While the use of verbs of cognitive attitude in its non-descriptive qualificational construal is always deictic and subjective, the use of certain adverbs may be additionally polyphonic. The notion of polyphony enters the picture if certain adverbs are used to convey a hearsay reading. When conveying a hearsay reading, the use of aparentemente and supuestamente is polyphonic since another voice is introduced by the current speaker or at least referred to. Hence, one other deictic centre is established (or at least referred to) where the reported information originated. The evidential value that is applied to [p] by the reportive use of aparentemente and supuestamente originates in the current speaker’s origo so that its use is also subjective and at least one other subjective point of view is introduced by the reference to another voice. Both adverbs and verbs of cognitive attitude are regarded as means of speaker’s perspectivisation since they are means which bring to the fore the speaker’s perspective of the narrated event.

314

6 A context-sensitive and functional approach to the use of the Spanish modal verbs poder and deber The English modals have been described from different perspectives and within a whole series of different theoretical frameworks and can probably be considered as one of the best studied language categories overall (de Haan/Hansen 2009: 1).

With their collected volume on modals Haan/Hansen would have had the chance to provide a study of the modals of languages other than English with all its facets and meaning nuances. But unfortunately, they explain: […] we understand modality in a narrow sense as comprising exclusively the meanings ‘necessity’, ‘obligation’, ‘possibility’, ‘permission’, and ‘volition’. […] We shall […] exclude all types of evidential markers. Whereas modal meanings like possibility and necessity refer to the commitment of the speaker to the truth of what he/she is saying, evidentiality refers to the source of evidence the speaker has for his statement […] (de Haan/Hansen 2009: 3).

So this implies that their starting point is the semantic category of modality as such. Adopting a function-to-form approach they look at which modals represent which modal meanings. If they had combined a function-to-form approach with a form-to-function approach, it would have helped the modals of others languages to be as ‘equally well’ studied as the English ones. Cornillie/De Mulder/Van Hecke/Vermandere (2009) treat “Modals in Romance languages”, that is, in French, Italian, Spanish and Romanian, but restricting their work to the analysis of modals expressing ‘necessity’ and ‘obligation’, that is, to the Romance equivalents of English must and its synonyms (cf. Cornillie/De Mulder/Van Hecke/Vermandere 2009: 107). But Cornillie (2007a) studies poder, deber and tener que from a cognitive-functional perspective (analysing their epistemic/evidential use), while Dendale (1994) studies French devoir and Tasmowski/Dendale (1994) study pouvoir in terms of evidentiality. Pietrandrea (2005) is concerned with the Italian equivalents. In the descriptive grammar of Spanish, the evidential nature of the modals is not touched upon. In his article for the Gramática descriptiva “Modo y modalidad. El modo en las subordinadas sustantivas” Ridruejo (1999) distinguishes between four kinds of modality: [...] hay cuatro grupos de conceptos modales: a) modalidades aléticas, integradas por las nociones de necesario, posible, contingente e imposible; b) modalidades epistémicas, constituidas por los conceptos de sabido como cierto, indeciso y sabido como falso; c) modalidades deónticas, formadas por los conceptos de obligatorio,

315 permitido y prohibido; d) y finalmente modalidades existenciales, que incluyen los conceptos de universal, existente y nulo [...] (Ridruejo 1999: 3212).

Spanish poder is said to convey epistemic modality and deontic modality: [...] el auxiliar poder en español se emplea para transmitir tanto modalidad deóntica de permiso o capacidad como modalidad epistémica de incertidumbre: Puede entrar en español significa tanto “tiene la capacidad de entrar” como “quizá entre” (Ridruejo 1999: 3215).

Mendikoetxea (1999: 1713) also merely mentions that the modals poder and deber are used epistemically and deontically. While Ridruejo and Mendikoetxea only apply a deontic and epistemic reading to poder, Cornillie shows that the evidential dimension is “not necessarily absent in poder” (Cornillie 2007a: 186) and that – if used evidentially – it conveys an inferential reading (Cornillie 2007a: 187), whereby it is said usually not to rely on inference (Cornillie 2007a: 254). This is in contrast to deber and tener que which “most often involve inductive inferences and deductive inferences, respectively” (Cornillie 2007a: 254).136 The most frequent use of deber in general, however, is deontic, but the most frequent non-deontic use is the inductive-inferential one (Cornillie 2007a: 199).137 If an instance of deber that conveys an inferential reading is found with hedges like creo que the role of inference is reduced in favour of a more epistemic reading: […] hedges like creo que “I think” or supongo que “I suppose” more often accompany inductive than deductive readings. These hedges […] are in turn an indication of the reduced role of inference itself, in favor of a purely epistemic reading (Cornillie 2007a: 195).

136 Müller (2005) also explains that the epistemic use of poder “does not presuppose an informational source” as “the speaker does not necessarily infer anything from internal or external circumstances” (Müller 2005: 140). Deber, in contrast, is said to indicate that the speaker has/had some pieces of evidence to rely on, but while deber is considered to be bound to a weak information force, tener que is bound to a strong one (cf. Müller 2005). So tener que “signals a high level of verification in the sense that the information available to the speaker provides him with a maximum degree of confidence in the realisation” (Müller 2005: 142). 137 Cornillie investigated 152 instances of deber conveying a non-deontic reading and came to the following distribution (Cornillie 2007a: 199):

deber

Inductive inferential 59,21%

Epistemic 35,53%

Deductive inferential 5,26%

TOTAL 100%

316 Epistemic modality is often enhanced by means of subjective hedges […], which highlight the speaker’s belief state. […] These hedges often make for a purely epistemic reading replacing the explicit reference to an inductive inferential process (Cornillie 2007a: 201).

This is also shown by Coates (1983: 46, 138) for English must whose use is – nevertheless – considered as mostly inferential (Coates 1983: 41).138 Cornillie relates the three prototypical readings of poder, deber and tener que to different degrees of speaker commitment139: With modal verbs, the absence of inference appears to lead to weak speaker commitment to the proposition, the creative reasoning of inductive inferences results in intermediate speaker commitment, while the logical reasoning of deductive inferences results in strong speaker commitment. […] Thus, the modal verbs […] clearly link the evidential qualification with the epistemic modal qualification (Cornillie 2007a: 254).

Cornillie consequently links the absence of inference or kind of inference to weak, intermediate and strong speaker commitment. So the different degrees of speaker commitment represent a logical consequence of the absence/kind of inference. The following example illustrates that, if using tener que, there must have been evidence available to the speaker because he introduces his utterance containing tener que by the clause ‘what has already become clear is that’ (lo que ya ha quedado claro es que): (299) Desde luego, lo que ya ha quedado claro es que muy mal se tienen que sentir los populares con este asunto hasta el punto de que les hace incurrir en auténticas 138 According to de Haan (1999: 8-9), in contrast, English must is an epistemic modal because it is said to have no other reference point than the speaker alone. He only concedes must “evidential nuances” (de Haan 1999: 9). De Haan solely analyses ‘contextless’ sentences, which maybe explains why he terms must an epistemic modal. 139 In their article “Speaker commitment: back to the speaker. Evidence from Spanish alternations” Cornillie/Delbecque (2008) adopt a cognitive approach to the analysis of the modals poder, deber and tener que, providing an alternative to the analysis of the modal verbs in terms of weak, intermediate and strong speaker commitment. In analysing poder, deber and tener que in terms of speaker involvement, they show that […] (pure) speaker involvement decreases from poder to tener que: poder reflects the speaker’s own evaluation of the likelihood, deber is based on the conceptualizer’s reasoning, most often triggered by circumstantial evidence or speaker knowledge, whereas tener que involves deductive logical operations that often rely on external laws […] (Cornillie/Delbecque 2008: 54). This means that weak commitment (poder) correlates with strong speaker involvement, while strong speaker commitment (tener que) correlates with weak involvement (cf. Cornillie/Delbecque 2008: 37).

317 exageraciones a personajes tan poco dado a ello como puede ser el caso de Javier Arenas. (El País 28/04/2010)

So the speaker is certain that [p] because he can obviously rely on some kind of evidence. In sum, all three modals – albeit poder not usually – may be used to express an inference. Dendale (1994) and Tasmowski/Dendale (1994) have also shown an evidential – more precisely, inferential – use for the French equivalents pouvoir and devoir. For Dendale, devoir is ‘so evidential’ that it is termed a “marqueur essentiellement évidentiel” (1994: 37). Dendale explains that every use of devoir underlies an “opération épistémique”, i.e. “une opération de création d’information que l’on considère en général comme une opération d’inférence” (Dendale 1994: 38). Hoye (2008: 165-166) observes that the English equivalent must typically attracts adverbs such as apparently, clearly, evidently, inevitably, obviously, which underlines the inferential meaning of must: In fact, where epistemic modal-adverb collocations involve combinations with must, the role of the adverbial serves to focalize the inferential quality of the modal: when asserting a judgement, with the conviction typically conveyed by must, the speaker creates the expectation that the argument will be supported by appeal to some kind of evidence (Hoye 2008: 166).

While Hoye addresses the importance of the ‘narrower context’, that is, collocations, Maché (2009: 28-29) refers to the ‘wider context’ and explains that modal verbs generally acquire their definite meaning only in context: “Ihre endgültige Bedeutung erlangen die Modalverben jedoch erst durch den konkreten Gesprächszusammenhang, in dem die Äußerung gemacht wird” (Maché 2009: 28). Because of the context-importance in the analysis of modal verbs, SilvaCorvalán’s (1995) study of poder and deber is also context-sensitive. She shows “that the ‘modal verbs’ each have an invariant meaning […], but are interpreted to convey different contextual meanings […] as a consequence of their interaction with other elements in the context” (Silva-Corvalán 1995: 69). This makes her approach to modals monosemantic as she assumes modals to have an encoded meaning, whereby further meanings are contributed by contextually provided information. She convincingly argues that – while modals themselves are understood as being monosemantic because they are assumed to have a basic, core meaning (Silva-Corvalán 1995: 72) – “their context of occurrence may be polysemantic” (Silva-Corvalán 1995: 70), that is, modals appear in a variety of contexts. This, in turn, implies that changing the context of a particular modal may result in another interpretation of the modal’s meaning (cf. Silva-Corvalán 1995: 100). The basic, core meaning is the invariant meaning of the modals,

318 whereby the invariant meaning is considered one type of meaning in language. Altogether, Silva-Corvalán suggests three types of meaning: 1. Invariant meaning: “meaning which underlies, or is present in, all uses of a modal”; 2. Contextualised meaning: “the message which the modal conveys, or the analyst infers that the modal conveys in a specific context”; derives from the interaction with other elements of the context; incorporates “pragmatic factors that may not be context-independent as the [invariant meanings] of grammatical forms (as opposed to lexical forms) appear to be”; 3. Prototypical discourse meaning: the most frequent contextualised meaning which is conveyed by a particular modal in a specific corpus of language data (Silva-Corvalán 1995: 73). So Silva-Corvalán’s notion of invariant meaning roughly corresponds to ‘meaning aspects that are encoded by a certain linguistic item’, while her notion of contextualised meaning represents the notion ‘meaning aspects that are contributed by contextually provided information’. About her third type of meaning I will not make any claim as the present study is not a quantitative one.

6.1 The evidential use of deber Volkmann (2005: 377-381) shows that both constructions deber de + infinitive and deber + infinitive are used to convey the speaker’s validative attitude concerning [p]. According to Volkmann, the use of deber (de) is epistemic if it is used with further expressions of epistemic relativisation (cf. chapter 3.2): Für das Hervortreten der epistemischen Bedeutung lassen sich folgende Kontextfaktoren finden: Die Reihung oder Kombination mit weiteren epistemischen Relativierungen, z.B. der Markierungen der eigenen Schlussfolgerung als Informationsquelle durch Lexeme, die mentale Vorgänge bezeichnen (Volkmann 2005: 382).

While Volkmann (2005) makes no explicit reference to the inferential use of deber (de) – even though she mentions ‘conclusions as information source’ as a contextual factor for an epistemic use – Cornillie (2007a: 181) explains that “a statement with deber is most often based on inductive inference and stands for an intermediate commitment” and that “as far as deber is concerned, it is justified to assume that the mode of knowing and the speaker’s commitment go hand in hand” (Cornillie 2007a: 188-189). This means that, according to Cornillie, evidentiality and epistemic modality overlap in the use of deber. Certainly, the speaker’s commitment depends on the mode of knowing. In the following two examples the speaker expresses an inference. In example (300) the speaker refers to an event in the past and infers that the montage must have been simple for the CIA; in example (301) the speaker is fairly commited that ganar un título

319 (‘winning’) must be marvellous, whereby it is reasonable to assume that the speaker must have some reason to think so. He has surely seen other sportsmen winning a title and that this is something marvellous was certainly obvious. That is why one could even argue that the speaker in example (301) expresses a deduction. Admittedly, there are no premises found in the context. Instead, the inference must be based on previous knowledge: (300) * 27 Luis - 25-08-2010 - 22:11:04h El montaje debe haber sido muy simple para la CIA. Qué acusación más fácil contra un personaje que les resulta incómodo. No pueden denunciarlo por asesinato o por robo porque requerirían pruebas. Basta conseguir una mujer que esté dispuesta a denunciarlo por […] (El País 29/08/2010) (301) “Mi mejor motivación es ganar un título con el Athletic. Debe de ser algo maravilloso. [...] La mejor motivación posible es conseguir ganar con tu club algo así, aunque no hay nada similar a un Campeonato del Mundo, pero debe de ser maravilloso ganar un título con tu club, especialmente el Athletic, donde la consecución de títulos es un tanto más dificultosa que en otros clubes. [...]” (El País 29/08/2010)

The following example also contains an inferential use of deber. Whether the speaker compares two photos, the left photo showing the woman being pregnant and the other photo showing the woman after having lost weight or whether the speaker compares two photos of different women, the inference remains the same: every woman is overweight after having given birth so that the woman of the photo must also have been overweight after pregnancy: (302) #16 marantonia 23.Dic.2010 | 02:10 La foto de la izquierda es cuando estaba embarazada, así que su gordura no era por obesa, simplemente está encinta. Claro que debe haber quedado con sobrepeso después del embarazo y al someterse a dieta los resultados son de lo mejor. (El Mundo 25/12/2010)

Although not explicitly mentioned or referred to, the speaker in the following example must have some reason to infer that the Partido Popular must be frightened so that the reading conveyed by deber is inferential. Example (304) can be analysed in similar terms because the speaker judges – as in example (303) – about the emotional state of a third person/party: (303) [...] García Albiol sea la persona más indicada para conocer la problemática del colectivo. En un tono mucho más duro, el portavoz y viceprimer secretario del Partit dels Socialistes, Miquel Iceta, avisó ayer que “el PP está jugando con fuego y debe de tener miedo de los planteamientos xenófobos de alguna otra formación política” por su forma de plantear el debate sobre la inmigración. (El País 19/09/2010) (304) “El proceso establece un periodo de información interna previsto no solo a partir de la presentación de candidaturas”, concluyen. David Lucas, portavoz de la

320 campaña de Jiménez, señaló que el ex alcalde de Parla “debe estar muy nervioso si hace del envío de cartas el eje central de su campaña”. (El País 19/09/2010)

In the following text passage the journalist himself shows that deber is used to convey a deductive reading. So here, the context unequivocally shows that the use of deber is not always inductive, even though – I admit – another journalist could have chosen the verb inducir ‘to induce’ instead of deducir ‘to deduce’: (305) Y, de paso, quedó claro que aquello por lo que le preguntaban era una cuestión de partido “muy interna”, es decir, privada. De lo que se deduce que, frente a la tendencia de Aguirre a confundir lo público y lo privado, Zapatero debe tener muy claro lo que nos incumbe y lo que no. (El País 12/08/2010)

Example (306) contains a use of deber which is introduced by the verb estimar ‘to estimate’, and if a speaker estimates that [p], he must have some reason for so doing. So the person who estimates (Wanner) must rely on some information in order to be able to estimate that he must have more than half a million: (306) [...] Wanner contesta: “Yo soy creador, no contable”. Tampoco conoce el número de objetos guardados en su almacén de 500 metros cuadrados y adquiridos en los mejores talleres de cristal del Este de Europa a lo largo de décadas, aunque estima que debe tener más de medio millón. (El Mundo 27/12/2010)

The speaker in the following example explains that ‘it must be an error’ (debe ser un error). Deber seems to be used to establish a logical relation between the utterance before and the utterance containing deber. So debe [p] represents the logical consequence of the utterance before: (307) #11 Anónimo 21.Dic.2010 | 14:14 No veo nada en este proyecto que diga que se suprime el canon digital, porque ahora si no van a existir descargas ilegales se debería suprimir dicho canon. Debe ser un error de nuestra excelsa Ministra Sinde Derechos. (El Mundo 23/12/2010)

As the examples above show, the reading conveyed by deber is not always inductive. In analysing the inferential use of French devoir, Dendale/de Mulder (1996) also differentiate between different kinds of inferences that are expressed by the modal under discussion. Providing the following example (a) (A voir ses yeux rouges) Marie a pleuré (Dendale/de Mulder 1996: 306),

they explain that the type of inference involved may be deductive as well as abductive: Il y a entre ces deux formes d’inférence – abductive et déductive – au moins deux différences essentielles. La première se situe au niveau de la majeure mise en jeu. Dans la représentation abductive de [a], la majeure reflète une relation de cause à effet (Guentchéva 1991, p. 49), une relation causale (avoir pleuré  avoir des yeux rouges). Dans l’interprétation déductive, le lien d’implication logique entre q et p reflète une relation empirique d’effet à la cause, c’est-à-dire une relation d’explication

321 d’une situation, d’un état (avoir des yeux rouges  avoir pleuré). La seconde différence concerne la validité de la conclusion: conclusion nécessairement vraie pour la déduction, conclusion seulement probable, plausible pour l’abduction […] (Dendale/de Mulder 1996: 307-308).

The deductive character of devoir is confirmed by Haßler (1998): Tatsächlich weist […] devoir immer darauf hin, daß ein Sachverhalt aus Prämissen abgeleitet und bewertet wurde, während seine modalen Werte nicht mit der gleichen Klarheit bestimmt werden können. (Haßler 1998: 176)

The fact that deber may also be used to express a deductive inference, shows, for instance, example (305), as the introductory phrase De lo que se deduce que already illustrates. The conclusion expressed is thus necessarily true. However, as the data show, deber is used to express different kinds of inferences. In the following example the speaker infers that [p] ‘must be the case because of that announcement’ (Debe ser por aquel anuncio). So the inference expressed is on the one hand deductive but on the other, the conclusion drawn by the speaker must not necessarily be true: (308) Pues ese cacharro en manos del típico dueño de todo terreno que abusa del prójimo en ciudad y en carretera (de todos es sabido que los TTs salen muy poco del asfalto) tiene que ser mazo de peligroso... Cuidao.... 2. #2 Anónimo 27.Dic.2010 | 21:35 Debe ser por aquel anuncio de no sé cual todoterreno que decía: Hemos perdido al número uno!!! De ahí la necesidad de crear un todoterreno blindado. Se venderá como churros, pero el día de de la diabetes. (El Mundo 29/12/2010)

It is doubtful that it is possible to determine which type of inference is expressed exactly in every single use of deber. In the following two examples deber (de) is obviously used inferentially, whereby the context does not provide any information about the reason(s) why the speaker concludes what he concludes, which could help to determine the type of inference: (309) [...] alguno decia por ahi tienen estos mindundis si hubieran estado en manos de franco pinochet videla donde estarian en algun lugar del atlantico pacifico indico etc etc etc que mas vamos a hablar viva fidel * 367 Gustavo - 13-07-2010 18:17:43h Eso debe haber sido un juego de políticas de España, los españoles nunca ha sido tan buenos para ser tan generosos [...] (El País 15/07/2010) (310) Y hablando de reyes, ¿por qué no llama a Juan Carlos a declara? El sitio lleva su nombre, ¿no es sospechosos? O mejor todavía, Mari Sarah Palin, algo debe de saber ella también, y ¿quién mejor que ella para echar una mano a Pio Moa en reinterpretar la historia de España? Para eso están los Tribunales, ¿no? (El País 15/07/2010)

In example (311) the inferential use of deber (de) is bound to the fulfilment of certain conditions marked by si ‘if’ so that the following structure emerges: ‘si X, then debe (de) ser [p]’. So if the conditions can be interpreted as premises,

322 the inferences drawn by the speaker may be – for the speaker himself at least – necessarily true. But the speaker may be commited to different degrees that [p] must be the case (cf. the contrary line of argumentation of van der Auwera/Plungian 1998). At this, I subscribe to Haßler’s view that deber’s modal values cannot (always) be determined exactly (cf. Haßler 1998: 176): (311) Si esa cantidad de 18000€ es cierta, deben de ser muy pocos los que se acerquen a ella Yo no conozco a ninguno y si existe debe ser en unos pocos puestos de los mas altos, lo que probablemente es comparable a un sueldo de jefe en una multinacional del centro de Europa. (El Mundo 23/12/2010)

Even though the modal values are even more difficult to determine than the evidential values of deber (de), Cornillie (2007a: 181) has shown that deber – besides conveying mostly an inductive-inferential reading – stands for an intermediate commitment. This is logical insofar as low commitment would be expressed by poder, for instance, while very high commitment is almost never expressed by deber as there almost always remains ‘a rest of uncertainty’ since an inference can almost never be presented as being necessarily/absolutely true. In the introductory part of the present study Pietrandrea’s (2005) findings on the reportive use of Italian dovrebbe (Spanish debería) are summarised. Pietrandrea concedes that the use of dovrebbe to mark reported information is less due to the semantic potential of the modal but more to the conditional morpheme – that is, to the grammatical form – itself. Borrowing Spanish examples from Olbertz (1998), Squartini explains that deber + infinitive occurs as a reportive marker when combined with verb forms expressing a lower degree of assertiveness, as shown by the counterfactual interpretation of the Indicative Imperfect [a] and Conditional [b] (Squartini 2004: 887). (a) Debía haber llegado a casa hace una hora (Olbertz 1998: 395; in Squartini 2004: 887). (b) No deberías estar en Madrid hasta dentro de quince minutos por lo menos (Olbertz 1998: 395, 506; in Squartini 2004: 887).

Example (a) is reminiscent of Reyes’ (1996: 12) example Mañana daba una conferencia María, where the imperfect is also said to indicate “información […] de segunda mano” (Reyes 1996: 12; cf. also the example given by Volkmann 2005: 234).140 So Reyes (1996) and Volkmann (2005) apply a modal/evidential value to the grammatical form of the imperfect. With reference to Squartini’s words (2004: 887), however, it is questionable as to whether the modal verb itself in (b) can clearly be applied to a reportive reading. I agree with 140 Gómez Torrego also refers to the modal/evidential use of the Spanish imperfect by providing the following sample sentence: “No dijiste que el partido era mañana” (Gómez Torrego 2002: 149; my emphasis).

323 Pietrandrea (2005: 87) and would like to put emphasis on the fact that – generally speaking – the reportive reading is clearly ‘more due’ to the conditional morpheme than to the modal’s semantics. In the following example, the introductory phrase Para González-Pons ‘For González-Pons’ as well as the quoted fragment following debería give proof of debería being an instance of the journalistic conditional: (312) Para González-Pons, España es menos fuerte con Zapatero como presidente y hay más desconfianza en la economía, por lo que debería “ponerse a trabajar” [...] (El Mundo 28/12/2010)

The example illustrates that debería could be replaced by podría or tendría que. Considering the context, debería seems to be the most appropriate modal. Nevertheless, the replacement by podría or tendría que shows that the reportive reading is bound to the conditional morpheme and not to the modal itself. Squartini (2004) also mentions the reportive reading of devoir, which should not be confused with the reportive reading of the conditional form. So the following use of devoir (example taken from Kronning 2001: 76) is also analysed as conveying a reportive reading because Squartini claims that French “devoir + infinitive has both inferential and reportive functions” (Squartini 2004: 875876): (c) D’après les prévisions météo, le temps doit s’améliorer demain.

Squartini (2004: 876) considers this example reportive, “provided that the speaker bases his/her utterance on previous knowledge, reporting that some given state of affairs was planned or supposed to happen”. “[T]he speaker is only reporting what is expected to happen, attributing it to an external source” (Squartini 2004: 877). So the origo the inference originated in is not the speaker. But in my view, the external source is not to be found in doit but in the phrase D’après les prévisions météo ‘According to the weather report’. If I say (d) According to the weather report, it may rain tomorrow, (e) According to the weather report, it is going to rain tomorrow, (f) According to the weather report, it possibly rains tomorrow, or (g) According to the weather report, it rains tomorrow, is then the use of may, going to, possibly and Ø reportive as well? In example (g), in particular, the reportive value would be applied to the indicative present tense of to rain. Clearly, the source for [p] is not the current speaker; the information source of [p] is the weather forecast but the reportive reading is bound to the introductory phrase. The same holds for the following example:

324 (313) Según el PP, la vicepresidenta debe informar sobre los efectos para el coste de la vida de esta decisión tomada “en plena crisis económica”, recuerdan. (El Mundo 31/12/2010)

The speaker obviously reports on a statement uttered by the party PP, thus attributing [p] to another deictic centre. There is nevertheless no reason to assume that debe itself is used to convey a reportive reading. The original speakers could have also said that the vice-president might/could inform about the effects. Would then puede in Según el PP, la vicepresidenta puede informar sobre... also be described as being reportive? However, what Kronning’s weather example differentiates from the following ones is that the current speaker actually reports on an inference141 made by someone else (the weatherman). In the following examples [p] is also marked by deber and the transmitted information is marked as being reported as phrases such as En su opinión ‘In his opinion’ show: (314) En su opinión, el Consell debe plantearse introducir capital privado en las empresas “públicas, parapúblicas o semipúblicas” que no sean rentables y “pierdan y pierdan dinero”. (El Mundo 22/12/2010) (315) Según esta Directiva, cuando los edificios son construidos, vendidos o alquilados, debe ponerse a disposición del posible comprador o inquilino, un certificado de eficiencia energética. (El Mundo 25/11/2010) (316) Según el presidente del comité de empresa, Astilleros debe contemplar la diversificación industrial de la zona para recolocar allí al excedente laboral que no pueda acogerse a las prejubilaciones. (El País 29/08/2010)

While the use in the weather example is originally inferential, the uses in the examples above are deontic and the information is – such as the weather example – represented as being reported. This raises the question of whether those uses of deber are also said to convey a reportive reading.142 In the inferential use 141 Squartini (2004), in contrast, does not translate French devoir by English inferential must. Sample sentence (c) is translated as “According to forecasts, the weather will get better […] tomorrow” (Squartini 2004: 875; my emphasis). Perhaps he translated the French verb this way in order to emphasise the reportive reading and to differentiate it from the inferential reading. However, this is not translated correctly and the example represents an inference which is reported on. Furthermore, in the context of weather reports, it is highly likely that the use of deber is inferential. I will be concerned with weather examples in the chapter about the synthetic future. 142 It should be emphasised that Squartini (2004: 890) only claims a reportive reading for French devoir – neither for Spanish deber (in the present tense) + infinitive nor for Italian dovere (in the present tense) + infinitive. Even though the reportive reading is only stated for devoir and exemplified by the weather example (c), which should not be termed ‘reportive’, I made the first test with English examples and the second with Spanish examples because I kept to the structure (containing a phrase that indicates re-

325 as well as in the deontic use both the inference ‘debe [p]’ and the order ‘debe [p]’ originate in a source that is external to the current speaker (cf. Boye’s (2005) concept of force-dynamic potential). In both cases the current speaker reports on the fact that debe [p]. But it is not the modal itself that conveys a reportive reading. One might ask now why there is a reportive reading applied to the conditional, if the journalistic conditional is also often used with phrases such as Según X ‘According to X’ (cf. chapter 7.1). A reportive reading is applied because the conditional alone can function as a quotative marker. The following example contains such a use: (317) Miró ha confirmado lo que en su día avanzó Consuelo Císcar, directora del IVAM, de que se está buscando financiación privada para que colabore en el proyecto de ampliación. Pero en una primera fase, subrayó, lo importante es tener el espacio. Esa sería su prioridad porque en lo que respecta a la ampliación del museo, la consejera reconoce que la actual situación económica es “muy complicada”. Tras lo cual se quejó de que el Gobierno central no ayuda como debiera. (El País 11/08/2010)

There is no further phrase at the sentence level which represents the information source for [p]. Only the context verifies that the verbal form is an instance of the journalistic conditional. Cornillie (2007a: 199) can also not verify the reportive reading of deber; his data did not admit a reportive reading of deber so he calls for further investigation. From the corpus data he worked with “it appears that deber hardly allows the speaker to transmit someone else’s inference” (Cornillie 2007a: 204). I subscribe to this view. As also referred to in the introductory part of the present study, Italian dovrebbe was shown to condense “a process of conditioned inference” so that “it can be described as an inferential evidential that marks tentative conclusions” (Pietrandrea 2005: 86). Spanish debería is also used by speakers who want to ‘tinge [p] inferentially’. But besides the fact that the conditional form of deber is used to convey a reportive and inferential reading, debería is also used to express the speaker’s subjective opinion without being explicitly represented as indicating an inference. The context where speakers utter ‘debería (de) [p]’ to express their opinion concerning [p] is highly subjective so that the following examples come from chat forums, or debería (de) is part of direct quoted speech: ported information) in which Squartini claims to have found a reportive reading of deber. I only want to put emphasis on the fact that devoir – that is, the modal itself – can also not be said to be used to convey a reportive reading.

326 (318) [...] en caso de que se produzcan cancelaciones y retrasos masivos como los más recientes en el aeropuerto londinense de Heathrow, anunció el ministro de Transportes, Philip Hammond, al ‘Sunday Times’ este domingo. “Debería existir una multa en caso de suspensiones de servicio”, sostuvo Hammond. Para tranquilidad de los viajeros, deberían exigirse mayores responsabilidades, explicó. (El Mundo 29/12/2010) (319) “No estoy planteando una subida de impuestos, simplemente digo, como reflexión, que España, en relación con los servicios públicos que tiene, y en relación a las infraestructuras que tiene, y que debería seguir desarrollando, cuenta con una base impositiva muy baja en relación con la media de los países europeos”, ha indicado. (El País 17/08/2010) (320) Comentarios 1. #1 Anónimo 28.Nov.2010 | 12:50 El dinero público se debería de usar para otras cosas que para que los jovenes beban y consuman drogas [...] (El Mundo 29/11/2010) (321) 144 alf - 02-07-2010 - 18:01:31h tienen razon los que piesan que sobran funcionarios. el estado debería privatizar todos sus servicios. hacer un gran eres en los servicios públicos. y contratar al personal según gestión de empresa privada. todos saldriamos beneficiados y todos PAGARIAMOS CULAQUIER SERVICIOS [...] (El País 04/07/2010) (322) “No basta con descargar costes sobre la Seguridad Social, deben aportar los ingresos correspondientes para compensar el sobrecoste del sistema. Esa debería ser la línea de futuro”, ha precisado. (El Mundo 22/12/2010) (323) #1114 Anónimo 20.Dic.2010 | 14:28 Simplemente ridículo, tanto como los que les molesta la educación sexual (nuevamente por sus creencias religiosas). La religión (todas) debería quedarse fuera de las aulas. (El Mundo 22/12/2010) (324) El PP culpa a Zapatero de crear un clima de crispación “El PSOE, que lleva años cosechando derrotas por goleada en Murcia, debería elegir mejor a sus candidatos”, ha advertido antes de insistir en que no es “un buen comienzo” para García Retegui acudir a esa concentración que no contaba con autorización previa. (El Mundo 25/12/2010) (325) #109 (javier), lo de disponer de una plaza para toda la vida, aun si no se ocupa, es mentira. Lo de que no les despiden si no trabajan, como tú dices, hay excepciones, de acuerdo. Se debería regular la actividad de los empleados publicos y vincularla con el desempeño. (El País 04/07/2010) (326) ¿Censura? ¿Nuevo régimen? ¿Te parece medianamente normal que exista semejante libro? ¿Y que haya gente que lo compre? Ya puestos, ¿por qué no publican una guía del violador o del maltratador? Y digo más: yo creo que este “escritor”, el tal Greaves, debería hacérselo mirar. (El Mundo 13/11/2010) (327) Pero aunque “me tiren piedras”, debo decir que algunos primero opinais y luego conocéis algo. Incluido el periodista que firma el artículo. Lo lógico debería ser al revés. (El Mundo 28/10/2010)

327 (328) #1 Anónimo 26.Oct.2010 | 14:00 No solo se ven afectados estos dos imputados... se ven afectados TODOS LOS IMPUTADOS. La justicia debería resolver más rápido. Una vez que han declarado debe pensarse si siguen o no imputados. (El Mundo 28/10/2010) (329) Me parece que el justamente celebrado e impresionante festejo reivindicativo del Orgullo Gay madrileño, este año centrado en la transexualidad, debería plantearse en los siguientes y hacer ostensibles, preferiblemente con carruajes, a los hombres y mujeres homosexuales de tantísimos países musulmanes en los que se persigue, a veces hasta la muerte, no ya el ser visiblemente gay [...] (El País 04/07/2010)

In the examples above, the conditional form of deber (de) is clearly used to express the speaker’s subjective opinion about a certain state of affairs in a polite way. All speakers express that – in their view – a third entity they are talking about ‘should do X’ or that ‘X should be this or that way’. So deontic modality also plays a role here. The speakers may have inferred from certain facts or previous knowledge that [p] should be the case; the speakers may have reasons to consider why ‘X should be done’ or why ‘X should be this or that way’. At the moment of transmitting [p], being marked by debería, “the speaker cannot verify the truth value of [p]” (Pietrandrea 2005: 75) so that the conclusion expressed is only tentative and perhaps more appropriately dealt with in terms of assumption (cf. Boye 2006: 66). The following examples contain, additionally to debería, a causal conjunction such as porque ‘because’ and pues ‘because/since’ or a causal construction such as de ese modo ‘that way’ or ya que ‘since/because’. Even though debería appears in equally subjective contexts as in the examples above, the use of the conditional form in the examples below should be described in terms of intersubjectivity as well. The causal conjunctions/constructions provide an explanation or justification why [p] should be the case.143 That is to say, they are used to convince the interlocutor or other chat forum participants why the subject under discussion ‘should do X’ or why ‘X should be this or that way’: (330) #2 Anónimo 23.Dic.2010 | 15:31 Ya que no hay forma de que CAMPS Presidente deje el cargo, al menos lo debería dejar Camps Vice- económico, porque las cuentas son un desastre , se mire por donde se mire. El nivel de la mayoría de los consellers es ÍNFIMO, Sobran políticos y faltan técnicos PROFESIONALES [...] (El Mundo 25/12/2010) (331) Evidentemente, y aunque estoy a favor del Sahara, estoy en contra de una intervención del ejército español, aunque creo que el gobierno de España debería

143 Cf. also Blühdorn/Reichmann’s remarks (2010: 20) on “inferential evidence-conclusion relations” that are established by Portuguese porque.

328 cambiar radicalmente su postura ya que ha roto la postura que ha venido trayendo España tradicionalmente sobre el Sahara y eso es peligroso. (El Mundo 13/11/2010) (332) #68 Anónimo 23.Dic.2010 | 22:09 Quien lo haga contribuye a la delincuencia facilitándoles el lavado de capitales así como se expone a que le investiguen por disponer de un capital que no gana. Nadie debería efectuar esas ventas pues legaliza a ladrones, traficantes, políticos corruptos, etc... (El Mundo 25/12/2010) (333) #31 Anónimo 28.Nov.2010 | 22:57 Ha sido una vergüenza lo que ha hecho el PSC en Cataluña. Ha sido más nacionalista que un partido radical. Impresentable que representantes del PSOE apoyen a esta gente. El PSC debería contratar a Laporta xq [porque] representa sus valores para esta comunidad autónoma. (El Mundo 29/11/2010) (334) Anónimo 07.Dic.2010 | 05:52 #5 Se debería votar según número de habitantes. De ese modo las moscas cojoneras como Colombia no molestarían a los gigantes como China e India ni a paises con casi 30 veces más habitantes como USA u 8 veces más habitantes como Alemania. (El Mundo 09/12/2010) (335) Sheen está añoz luz por delante de Gibson y Cruise, que siempre me han parecido más bien mediocres. Pero debería poner un poquito de orden en su vida porque se está perdiendo la infancia de sus hijas que imagino que a estas alturas ya estarán traumatizadas por los espectáculos que organiza su padre. (El Mundo 28/10/2010)

Example (336) contains an implicit explanation why [p] should be the case because the speaker explains that ‘if [p] then the priests would be happier’: (336) #58 Anónimo 13.Nov.2010 | 03:09 #33 La Iglesia debería plantearse eliminar el celibato para los curas que tienen que reprimir sus necesidades sexuales; y digo yo...¿no serían más felices con su mujer a la que amar, con sus hijos a los que querer y predicando la palabra de Dios con sus feligreses? (El Mundo 15/11/2010)

The form-to-function approach and the function-to-form approach are therefore combined in the present study because when analysing the use of debe the deontic uses had to be separated from those that express an inference, an assumption or the speaker’s subjective opinion. So from a semasiological perspective, debe is also used with a deontic meaning, for instance. Example (337) contains a ‘classical’ deontic use of deber. Athens must pay a lot: (337) El alto precio que debe pagar Atenas por sus letras ha atraído a los inversores con menor aversión al riesgo, ya que la demanda ha sido más que satisfactoria. (El País 15/07/2010)

The use of debe appears in a subordinate clause and represents an already known fact: it is known that Athens must pay a high price. Hence, this use is clearly deontic as it expresses what Athens must do. The following four uses of deber are deontic because they also express that ‘X must do Y’ but, in contrast

329 to the use in example (337), they are also to a certain extent subjective and/or inferential: (338) No obstante, quiero decir que la reforma fiscal debe ser mucho más profunda que eso, debe recuperar impuestos, como patrimonio y sucesiones, y la capacidad recaudatoria del de sociedades. (El País 29/09/2010)

In example (338) the speaker explains how the tax reform must/should be. The use of debe in the following example is similar: (339) “Pedir nuestras disculpas es algo más que un acto vergonzoso. Al contrario, si alguien tiene que pedir perdón es Turquía porque fueron ellos los que empezaron esta escalada. Quién debe pedir perdón es Turquía por apoyar el terrorismo, el IHH (grupo turco que organizo la flotilla), Hamas y Hizbulá. Nosotros no pediremos perdón”. (El Mundo 28/12/2010)

In example (339) the speaker expresses his opinion. According to him, ‘who must ask for forgiveness is Turkey’ (Quién debe pedir perdón es Turquía). Thus, he judges that it is Turkey – and not the other side – that must beg pardon and that is why the deontic and a subjective use overlap. If it is furthermore assumed that the speaker may have reasons for thinking that Turkey must ask to be forgiven, it could be argued that this use is accompanied by an inferential moment. However, no contextually provided information gives hints about an inferential use since it is ‘mostly’ deontic but also subjective. The presence of the subjective meaning component is underlined by the information structure, i.e. by topicalisation: the speaker does not simply say Turquía debe pedir perdón but emphasises that it should be Turkey (and not the other state). The following two uses could be analysed in similar terms. In example (340) the opinion of 70% of interviewees is represented – this use is not reportive – and in example (341) the speaker utters what, in his view, must be done by the Partido Popular so that the uses – besides being mostly deontic – are also regarded as subjective: (340) [...] y otro 70% sostienen que si un español y un inmigrante legal solicitan el mismo puesto de trabajo se debe seleccionar al más cualificado sin tener en cuenta su lugar de origen, frente al 22,2% que opina lo contrario. (El Mundo 28/12/2010) (341) Si hay algo peor que no saber solucionar el problema del paro, es usarlo de forma tan torpe y nauseabunda para soltar un bulo electorero. ¿Nos lo creemos? El PP debe registrar una moción de censura al gobierno de forma inmediata. (El Mundo 28/12/2010)

The four uses of deber above (examples 338-341) could explain why Boye (2005) is able to represent the deontic reading of must via the same scheme as the inferential reading of the modal. The deontic and the inferential reading of deber may – in certain contexts – overlap because, to speak in Boye’s terms, the

330 order as well as the inference originates in one and the same source (cf. Boye 2005: 61, 65-66).

6.2 The evidential use of poder Dendale/Tasmowski (1994: 55) describe pouvoir – just like devoir – as an evidential marker because utterances containing pouvoir “sont l’indice d’un processus mental qui consiste à créer des prémisses, à en inférer des conclusions et à évaluer ces conclusions pour n’en retenir que partie”. Cornillie (2007a: 181), in contrast, explains that “an epistemic judgment with poder usually does not involve inferences and corresponds to a weak speaker commitment”. However, Cornillie (2007a: 186) demonstrates that poder may be used to convey an evidential reading. Even though “the inferential dimension is not crucial to the epistemic reading of poder”, this modal can indeed be used inferentially (Cornillie 2007a: 187). With regard to contextual factors that qualify the use of a modal, Volkmann (2005), explains: “Schlussfolgerungen, wie sie durch die epistemische Relativierung ausgedrückt werden, beruhen auf vorher erworbenem Wissen, durch den Bezug zu vergleichbaren Situationen” (Volkmann 2005: 410). Such an example is provided by Cornillie. He illustrates that the use of poder may be based on an inference from the preceding context and provides the following example: … por ejemplo una amiga se la pasa todo el santo día en la casa... después cuando ya están establecidos en el matrimonio etcétera ¿no? Eso creo que puede traer unos problemas hasta más graves que los anteriores ¿no?. Bueno... (Habla culta: Caracas: M16; in Cornillie 2007a: 186).

The preceding context represents the speaker’s knowledge, which he draws the conclusion from: the described situation may carry problems (cf. Cornillie 2007a: 186). Thus, insofar as the findings by Dendale/Tasmowski, Cornillie and Volkmann are considered, the use of French pouvoir and Spanish poder seem to differ slightly. While the former describe pouvoir as an evidential marker, the latter explain that poder usally does not involve inferences, but may be used to express inferences (based on previously acquired knowledge). Besides the fact that poder may be used to express inferences based on the speaker’s knowledge, the modal conveys a relatively weak speaker commitment as it is used to express a possibility among various – at least two; [p] may be the case or not or rather ‘may be this or that’: (342) Y, a tenor de las declaraciones de Schäuble, son dudas bien fundadas. “Ni la UE ni el Gobierno alemán han tomado todavía una decisión, lo que significa que la

331 respuesta puede ser positiva o negativa”, advierte Schäuble en un artículo publicado por el diario dominical Bild am Sonntag. (El País 28/04/2010)

The phrase ‘the answer may be positive or negative’ (la respuesta puede ser positiva o negativa) perfectly illustrates that poder is used to indicate that [p] may be this case or that case: the answer may be positive as well as negative. Even if it is only spoken about ‘being positive’ or ‘being negative’, the hearer would know that the opposite may equally be possible. So poder is used to express one possibility among two possibilities or even more. The same applies to example (343). ‘The situation may worsen’ or may not (la situación puede agravarse): (343) La situación es de alto riesgo, según explicó el comandante nacional de Protección Civil, Gil Martins, que alerta que la situación puede agravarse. (El País 16/08/2010)

Indeed, poder is used to express that it is probable that [p], whereby it is also possible that [p] is not the case. In the following example, the utterances ‘It is a factor that may give you points’ (Ese es un factor que […] puede darte puntos), ‘and this may cause anxiety in Madrid’ (y eso en Madrid puede crear ansiedad) as well as ‘And this may increase the risk of exploitation’ (Y eso puede aumentar el riesgo de explotación) show that puede does no more than indicate one possibility – among others – that [p]: (344) Yo ya sabía que ciertos jugadores estaban en peligro de ser traspasados, pero hemos tenido la suerte de conservar a algunos veteranos que saben llevar el peso del vestuario. Ese es un factor que, durante la temporada, puede darte puntos. Y nosotros vamos a necesitar todos los puntos posibles. (El País 01/09/2010) (345) Hoy el Barça va sumando títulos y está jugando el mejor fútbol del mundo, y eso en Madrid puede crear ansiedad. (El País 01/09/2010) (346) La causa es “la persecución de muchos municipios, con multas por ejercer en la vía pública”. Y eso puede aumentar el riesgo de explotación. (El País 06/09/2010)

All the following utterances containing puede also express that the state of affairs described may happen/may be the case (or may not). The speakers may have reasons to think so and they may rely on experiences gained formerly or they may rely on knowledge acquired previously (cf. Volkmann 2005: 410; cf. Cornillie 2007a: 186), whereby they do not verbalise what exactly they rely on: (347) Y a la zona euro en su conjunto, e incluso más allá. “No se trata de un problema de España, ni de la periferia de Europa: puede convertirse en una crisis que afecte a casi todo el mundo desarrollado, que comparte ese agujero fiscal”, explica el economista José Luis Alzola. (El País 01/05/2010)

332 (348) El año del diluvio es una distopía satírica, tan ingeniosa como inquietante. Lo que te cuento te divertirá, parece decir la autora, pero presta atención, pues puede suceder pronto o quizá ha sucedido ya. Y al ver a Atwood, con su melena rizada blanca, sus brillantes ojos verdes y sus pómulos marcados, es imposible no pensar en una pitonisa. (El País 16/08/2010) (349) “[...] Con la nueva norma, la compañía eléctrica que distribuye la electricidad producida no tiene que dar su visto bueno. Bebe del sistema alemán, en el que la instalación de paneles solares en los hogares es casi una notificación. La reforma puede tener un efecto socializante y conseguir muchas instalaciones pequeñas”, resume Anta. (El País 28/04/2010) (350) Pérez destaca que el decreto abre la puerta a lo que en unos años puede ser un nuevo mercado: la generación de electricidad solar fotovoltaica en casa sin prima para autoconsumo. [...] La fotovoltaica ha puesto sus esperanzas en la “paridad con la red”. Aunque sigue siendo cara, su precio cae a una velocidad de vértigo y en 2014 puede rondar el precio que paga el consumidor por la electricidad. (El País 28/04/2010) (351) [...] pero Alemania, que los debe liderar, no está convencida de ello y su instrumentación ha de ser aprobada por el Parlamento en medio de unas importantes elecciones en el Estado de Renania Westfalia, lo que puede retrasarla. (El País 28/04/2010)

All the text passages above do not provide further contextual information about why the speakers assume that ‘puede [p]’ and neither do the following examples. Poder is always used in the same way: it usually expresses the speaker’s personal judgement about [p] which puede ser ‘may be’, puede reducirse ‘may reduce’ or puede cruzarse ‘may come across’ etc. In every case it is equally possible that [p] may not be the case: (352) La siguiente fase de la crisis puede ser el estancamiento sin empleo Llegan Europa y EE UU a ese G-20 con distintas coyunturas y diferentes actitudes: la UE, con un crecimiento anémico y reinventando sus reglas del juego; EE UU, saliendo de la recesión con más fuerza [...] (El País 26/05/2010) (353) Como Adrián, más de 5.300 personas esperan un órgano en España. Son 56.000 en toda Europa. Ahora, con la futura entrada en vigor de la directiva europea de trasplantes de órganos -que se aprobó el jueves pasado-, el tiempo de dilación puede reducirse. (El País 26/05/2010) (354) No debería tener demasiados problemas Paraguay para alcanzar los octavos de final, pero en su camino puede cruzarse Eslovaquia, una selección de mínimo rango que aspira a dar la campanada y atrapar la segunda plaza. (El País 26/05/2010) (355) Un hijo de esta edil logró en 2008 una plaza como electricista en calidad de personal laboral, después de haber sido beneficiado con varios contratos temporales directos por el Consistorio, y puede tratarse de uno de los casos de irregularidades investigados por la juez. (El País 28/04/2010)

333 (356) *17 manolo el de valencia - 24-05-2010 - 22:02:06h La iglesia, esa piedra en el zapato que tiene la ideologia comunista, puede ser la mejor medidora para este y otros conflictos netre cubanos, al menos los cubanos de Cuba,puesto que los de Miami no lo aprueban en su tozudez y anti-todo-acercamiento. (El País 26/05/2010)

Actually, it is reasonable to assume that speakers always rely on certain knowledge; otherwise they would not utter that puede [p]. But if they were more committed, they would use deber instead of poder. In using poder the speaker indicates that “the inferred conclusion is not the only possible conclusion […], but one of the possible conclusions” (Pietrandrea 2005: 90). One might argue that the use of puede in example (357) is in no way different from the uses in the text passages above. But there seems to be a difference in whether [p], marked by puede, is intersubjectively comprehensible or whether [p], marked by puede, is represented as a personal opinion in subjective contexts. Example (356), for instance, represents a text passage that was found in a chat forum. In other words, while the use of puede in example (356) seems to be highly subjective, the use in example (357) seems to be more intersubjective as the fact that ‘every kind of contaminant may represent a danger for public health’ (Cualquier efluvio contaminante puede representar un peligro para la salud pública) is commonly known and thus intersubjectively comprehensible: (357) El humo de una chimenea, el vapor de una planta de compostaje o el simple olor a fritanga que emana el extractor de un restaurante. Cualquier efluvio contaminante puede representar un peligro para la salud pública y, por eso, la nueva ordenanza de Medio Ambiente que aprobará en verano el Ayuntamiento de Barcelona pretende poner coto a este tipo de emisiones. (El País 26/05/2010)

It is not only the speaker who is committed to [p]; it is reasonable to assume that he shares his commitment with many other people so that the use of puede is undoubtedly intersubjective (cf. Nuyts 2001b: 65, who provides similar examples from Dutch and German). Here, “[t]he effect of intersubjectivity […] is […] a contextual effect, due to our general knowledge” (Nuyts 2001b: 65). As already indicated above, Cornillie (2007a: 187) explains that even though “the inferential dimension is not crucial to the epistemic reading of poder”, the modal can be inferential. The following examples undoubtedly contain evidential uses of poder. Contextually provided information leads to the assumption that the use in the following examples is inferential: (358) [...] es, sin duda, su extraordinaria pregnancia, la inmediatez de su percepción, la manera en que logran producir ese “impacto directo” del que habla Ami Barak en la presentación de esta muestra. Se trata, por cierto, de un impacto que a primera vista puede resultar enfáticamente inocuo, más bien amable, incluso seductor. (El País 16/08/2010)

334 (359) Una simple división muestra que el coste de la captura de cada animal se aproxima a los dos euros. “Puede parecer excesivo, pero es el coste de un proceso que es complejo y que se tiene que hacer de acuerdo con una normativa”, argumentan fuentes de la agencia. (El País 01/05/2010)

The speakers in example (358) and (359) use poder inferentially. Both speakers judge how a certain state of affairs ‘may be seen at first sight’ or that it ‘may seem excessive’, that is, they infer from some given facts how it may be seen by others. In the following example it is not only the journalist who already concludes that the quoted person expresses a conclusion but it is also the context that verifies that poder is used inferentially: (360) “Pero el problema es más general, va más allá de Europa, y puede durar años, como ya sucedió con la crisis asiática de los noventa, que fue noqueando a varios países sucesivamente”, concluye. (El País 01/05/2010)

The speaker infers that Europe’s crisis may take a long time ‘as happened with Asia’s crisis in the nineties’ (como ya sucedió con la crisis asiática de los noventa). So here, contextually provided information illustrates that poder is used to express an inference that is based on knowledge acquired previously. The use of poder in example (361) is very similar: ‘the crisis has demonstrated that’ (La crisis ha evidenciado que) [p] ‘may be’ (puede ser) ‘because’ (porque)…: (361) Y hacia ahí estamos focalizando muchos de los programas que se están poniendo en marcha, precisamente, para tener más técnicos en grado medio de FP. La crisis ha evidenciado que puede ser una oportunidad para todos esos jóvenes que abandonaron el sistema educativo hace unos años porque no necesitaban ningún requisito previo para trabajar y sin embargo son conscientes de que eso ahora es más complicado. (El País 01/05/2010)

In example (362) the proposition that is marked inferentially by poder depends on the fulfilment of a certain condition (indicated by si ‘if’) and ‘if [p] may be the case’ the speaker knows why: he introduces his explanation by the causal conjunction porque ‘because’. Thus, the use of puede is unequivocally evidential: (362) “[...] y hubo críticas respecto a que el dinero no se estaba gastando bien”, explica Jordi Vaquer, director del centro de investigación sobre política internacional CIDOB. “La crisis ha frenado ese crecimiento y, si el recorte es a corto plazo, puede ser positivo, porque sus responsables tienen tiempo de esta forma de adaptarse a estos ritmos de crecimiento tan rápidos y ser más eficientes”. (El País 26/05/2010)

In example (363) puede also conveys an inferential reading because the journalist indicates why ‘the quoted judicial sources consider that the escort reduction “may be very elevated”’ (las citadas fuentes judiciales consideran que

335 la reducción de escoltas “puede ser muy elevada”). Por ello means ‘hence’ or ‘therefore’ so that it establishes a causal relationship between the previous utterance and the following. Por ello indicates that the previous utterance causes the inference: (363) [...] del Superior, de sus salas y de las tres Audiencias provinciales, así como la fiscal superior de la comunidad autónoma y los fiscales jefe de los tres territorios. Por ello, las citadas fuentes judiciales consideran que la reducción de escoltas “puede ser muy elevada”. (El País 06/09/2010)

To sum up so far, examples where the use of poder is clearly inferential are not easily found. Even though I did not work quantitatively, it was conspicuously difficult to find text passages such as examples (358) or (360). The following examples – such as example (362) – also contain propositions marked by poder which depend on the fulfilment of certain conditions: (364) Grecia prosigue su vuelo en espiral hacia una especie de agujero negro, un punto de no retorno que puede acabar en el impago de su deuda si la ayuda internacional no llega de inmediato. Y arrastra en ese vuelo a otros países. (El País 01/05/2010)

‘If Greece does not immediately get international help, it may result in nonpayment of its debts’ (…puede acabar en el impago de su deuda si la ayuda internacional no llega de inmediato). So in this example and in the following ones ‘puede [p]’ depends on the fulfilment of a certain condition: (365) Esperanza Aguirre (PP), partidaria de que el velo no entre en clase, y el ministro de Educación, Ángel Gabilondo (PSOE) que pide que prevalezca el derecho a la educación. Consideran que si el asunto del velo se convierte en un arma electoral “puede suponer un adiós a la libertad religiosa”. (El País 28/04/2010) (366) Otros, de tendencia moderada, dan la bienvenida a la medida del Gobierno de “utilizar” a la Iglesia para resolver “entre cubanos” la crisis en torno a los presos. Fariñas cree que si la Iglesia es “imparcial la cosa puede funcionar”, sobre todo si su mediación con el Gobierno avanza y logra resultados más ambiciosos. (El País 26/05/2010)

Before having a closer look at the use of the conditional form podría, the construction puede (ser) que + complement clause will be further examined. Puede (ser) que is always used with subjunctive mood, whereby [p], marked by puede (ser) que, represents – from a syntactic perspective – the subordinate clause and puede (ser) que the matrix clause. However, Olbertz (1998: 149-150) describes puede que as located in between a verbal construction and a periphrastic adverb and explains that poder is not a complement taking predicate. Rodríguez Espiñeira/López Meirama (2008) also argue that in the construction puede que + clause, “the clause is not a subordinate one, dependent on puede, but a clause

336 preceded by a ‘grammaticalized segment’ which behaves as an adverb” and which is equivalent to quizá(s) or tal vez ‘maybe/perhaps’ (Rodríguez Espiñeira/López Meirama 2008: 294), even though it “does not acquire the autonomy of modal adverbs such as quizá(s)” (Rodríguez Espiñeira/López Meirama 2008: 310).144 So according to them, puede que “functions as an extra-propositional operator, like a marker of modality whose domain extends over a proposition (the content of a clause)” (Rodríguez Espiñeira/López Meirama (2008: 297): (367) No vieron venir nada de lo que se avecinaba. No hicieron bien su trabajo. No advirtieron que la banca estaba vendiendo crecepelo. Incluso puede que contribuyeran a vender humo. (El País 01/05/2010) (368) Pereda: “Puede que sancionemos para mostrar la voluntad de cumplir la ley”. Los hosteleros han encargado informes jurídicos para defenderse de las multas. (El País 01/05/2010) (369) Celina Pereda, explicó ayer a este periódico que los encargados del proceso sancionador serán funcionarios públicos que actuarán como inspectores de trabajo, aunque aún queda tiempo para perfilar el proceso. “Ya hablaremos de cómo se hará esto. Puede que tengamos que sancionar para demostrar la voluntad de cumplimiento de la ley, pero lo que queremos es avanzar hacia el espacio de convivencia”, apuntó. (El País 01/05/2010) (370) Ciudadanos que quiere demostrar que Madrid es una ciudad transitable en bici siempre. “Queremos animar a sacar las bicis todos los días, no sólo los fines de 144 So the terminological problem that is discussed by Thompson (2002) and Mithun (2009) concerning the notions of complement taking predicate and complement clause is also addressed by Olbertz (1998) and Rodríguez Espiñeira/López Meirama (2008) for the structure puede que. Thompson (2002) argues against the terms ‘complement’ and ‘subordinate’ with regard to the clause “where the action the speaker is accomplishing is being done by that clause” (Thompson 2002: 155) and which is syntactically subordinated to an epistemic/evidential/evaluative phrase (such as creo que). So the problem she addresses is primarily terminological in nature. As it is the matrix clause (the complement-taking-predicate phrase) which is usually seen as central and not the complement clause, the structure consisting of CTP + subordinate clause is indeed not appropriately analysed in terms of ‘main clause’ + ‘complement clause’. That is why Thompson (2002: 155) argues that this structure “may be better understood as a combination of an epistemic/evidential phrase together with a declarative or interrogative clause”. If this declarative clause is analysed as being (only) a complement, it does terminologically not convey that it carries new or main information (cf. Mithun 2009: 63). This should also be discussed for puede (ser) que and its complement clause which does not fulfil the function of only being a complement. From a syntactic perspective, however, [p] (epistemically/evidentially marked by puede (ser) que) does indeed represent a complement clause. Hence, a syntactic approach and a functional approach to one and the same phenomenon clash here.

337 semana. Puede que el Ayuntamiento algún día haga carriles bici, que algún día tengamos nuestro espacio vial propio, pero mientras tanto no nos vamos a quedar con los brazos cruzados”, asegura Antonio Fabregat, uno de los cuatro organizadores del evento. (El País 01/05/2010) (371) El fin de flash como herramienta web y el triunfo de los estándares puede que llegue a ocurrir, pero para eso queda un buen puñado de años. Está claro que la apuesta está en HTML5, algo que a mi parecer es de lo más acertado que se pueda hacer para el futuro de la web y un gran principio para la web 3.0 [...] (El País 01/05/2010) (372) El caso del Estatut catalán en el Constitucional es el paradigma de un cul-desac insalvable: la retransmisión de la partida ya dura demasiado tiempo y puede que los jugadores hayan muerto por falta de oxígeno, como apuntaba Manuel Vicent el otro día. Tanta duración hace del espectador un descreído: a estas alturas, salga lo que salga, alguien estará en absoluto desacuerdo [...] (El País 01/05/2010) (373) “Bolt es un extraterrestre; Powell y Tyson Gay son superhombres”, analiza Carraz; “pero el rendimiento de sus fibras musculares no depende del color de la piel. Yo pienso que la genética no tiene nada que ver con el color de la piel”. Puede que no, pero las pruebas de velocidad han estado siempre dominadas por atletas de color y el récord del mundo, desde que se estableció el cronometraje electrónico, allá en 1968, también. (El País 21/07/2010)

Rodríguez Espiñeira/López Meirama (2008: 293) have shown that puede que ‘it may be that’ is the more grammaticalised variant of puede ser que ‘it may be that’. Although the grammaticalisation process is not complete, the development of puede que “entails formal reduction, semantic specialization and an increase in subjectivity” (Rodríguez Espiñeira/López Meirama 2008: 293). The semantic specialisation should be described insofar as “puede que retains one meaning, the epistemic subjective one, out of the numerous meanings possessed by the original form (puede ser que)” (Rodríguez Espiñeira/López Meirama 2008: 297). Rodríguez Espiñeira/López Meirama (2008: 302) have found out that “puede ser que and puede que share certain contexts, those expressing subjective epistemic possibility”. They understand epistemic possibility as […] a neutral or intermediate stage in the epistemic scale going from absolute certainty to improbability, and hence expresses a weak commitment on the part of the speaker to the real state of affairs (Rodríguez Espiñeira/López Meirama 2008: 300).

Even though both constructions are said to share the contexts of subjective epistemic possibility, puede que becomes the specialised item that expresses subjective epistemic possibility. It favours the subjective value, while puede ser que “conveys notions of objective (or intersubjective) epistemic modality” (Rodríguez Espiñeira/López Meirama 2008: 302):

338 (374) El primer ministro de Italia, Silvio Berlusconi, busca comprador para Villa Certosa, la lujosa propiedad que posee en la isla de Cerdeña. Puede ser que necesite hacer caja tras firmar un divorcio multimillonario de Veronica Lario, que cada mes recibe 300.000 euros. O quizá es que la mansión, su piscina y sus céspedes le traen malos recuerdos. (El País 01/08/2010) (375) Nick Clegg roba la imagen de cambio y de juventud a los conservadores pero es el síntoma de la decadencia laborista No se sabe todavía hasta dónde llegará. Puede ser que al final, tras la jornada electoral del día 6 de mayo, quede en poco y no consiga el parlamento colgado , sin mayoría de gobierno suficiente y con el obligado recurso a la tercera fuerza que poseerá la llave del Gobierno. Puede ser que las cosas lleguen a ser más graves todavía: que el partido con más escaños quede desautorizado por un mal resultado en votos y porcentaje que le coloque detrás de los liberal-demócratas. (El País 28/04/2010) (376) Por lo tanto, deben preguntarse si van a poder soportarlo. Por otra parte, puede ser que exista una verdad personal concreta y que eso sea lo que el escritor desea revelar por encima de todo, y eso crea un conflicto insoportable que le hace bloquearse. (El País 06/09/2010)

In summary, for puede ser que the subjective value “is contextually inferable, whereas in the case of its derived form puede que it is an inherent value” (Rodríguez Espiñeira/López Meirama 2008: 307). Hence, puede ser que is described as weakly subjective and puede que as strongly subjective. Puede que is favoured if the speaker wants to express doubt or weak commitment with regard to [p] (Rodríguez Espiñeira/López Meirama 2008: 307). In Italian, the conditional form of potere is analysed as being more epistemic than the indicative form which is considered more deontic because potrebbe is said to condense the condition “‘if one considers things from the speaker’s epistemic domain’” (Pietrandrea 2005: 80). The following examples illustrate that podría is often used in connection with other forms of poder such as the construction puede ser que (example 377) and the indicative form puede (example 379): (377) También se ha constatado que los obesos tienen un peor semen , pero aún no está clara la relación causa-efecto. Puede ser que el exceso de peso incida sobre los niveles de testosterona y, consecuentemente, sobre la salud reproductiva. Pero también podría ser al revés, de forma que los bajos niveles de la hormona masculina incidirían en la ganancia de peso. (El Mundo 08/12/2010) (378) Al ser las plantas más altas, tardaron tiempo en encontrar la tubería de 25 atmósferas que necesita ese tramo del edificio para impulsar el agua, ya que sólo se usa en edificios muy altos y casi nadie tiene en stock. Y la presión no se puede aumentar en las conducciones auxiliares de 16 atmósferas “porque podrían reventar toda la instalación”, aseguran algunos vecinos, que también debieron pagar del dinero de la comunidad la reparación de la conexión entre [...] (El País 16/08/2010)

339 (379) La UE se dispone a poner en marcha el futuro Servicio Europeo de Acción Exterior, un cuerpo diplomático común con 7.000 empleados. Y, según Jordi Vaquer, puede ser una oportunidad para aprovechar esta nueva infraestructura diplomática y reducir la presencia de diplomáticos españoles en algunas misiones. “Se podría simplificar la red de embajadas. No es que se esté malgastando en ellas, porque la plantilla española es de las más reducidas para el tamaño del país. Pero puede haber una posibilidad de recorte”, afirma. (El País 26/05/2010)

The examples above seem to confirm that podría “expresses the existence of a possibility in the speaker’s epistemic domain”, whereby “[s]uch an epistemic domain cannot be accessed directly through the simple use of [poder]” (Pietrandrea 2005: 80). With the use of podría the speakers in the examples above express a personal, subjective judgement. The last two uses of podría could be described as being inferential. Further information provided by the context leads to the assumption that podría in example (380) and (381) is used inferentially. In example (380), the phrase no cabe duda ‘beyond doubt’ and por tanto ‘so/therefore’ are the contextually provided elements that underline its inferential use: (380) Es posible que determinadas leyes impidan a la gente circular encapuchadas, aunque en esta tierra se acostumbra a hacerlo durante la Semana Santa, pero no cabe duda, como bien sabía Esquilache, que el ocultamiento del rostro puede ser fuente de delitos, y por tanto podría perseguirse. (El País 26/05/2010) (381) “El Barça es un equipo que podría jugar en la NBA, sin ninguna duda”, opina Bryant Jackson se toma unos segundos antes de evaluar la posibilidad de que los Lakers alcancen su tercer título consecutivo esta temporada. (El País 10/10/2010)

Volkmann (2005: 299) describes the prepositional phrase sin duda as an inferential marker. The context in example (381) furthermore shows that the speaker obviously has sufficient knowledge gained previously in order to be able to infer that ‘Barcelona is a team which might play in the NBA’ (El Barça es un equipo que podría jugar en la NBA), whereby this is ‘beyond doubt’ (sin ninguna duda). Incidentally, this text passage perfectly illustrates that prepositional phrases like ‘beyond doubt’ or adverbs such as ‘undoubtedly’ are used if there remains ‘a leftover of doubt’. This ‘leftover of doubt’ is also expressed by poder in its conditional form. In the following, example (381) will briefly be compared with Diewald’s example and her analysis of the use of the German modal dürfen (cf. chapter 2.1.2.1): In den Wintermonaten Januar/Februar allerdings dürfte die Kurve – wie immer in der kalten Jahreszeit – einmal noch weiter nach oben gehen. Æ The graph probably goes further up.

340 The modal verb dürfen is used to utter a factuality judgement, that is, the use is deictic (cf. Diewald 1999: 16) and the verb behaves more like a modal auxiliary than a main verb (cf. Diewald 1999: 46). A factuality judgement always originates in the speaker’s origo and such an evaluation represents the speaker’s subjective judgement of an event. In deictic uses the presence of an origo as reference point is indispensable. Podría in example (381), for instance, behaves like dürfte in Diewald’s example. Here, the speaker also utters a factuality judgement, namely that ‘Barcelona is a team which might play in the NBA’ (El Barça es un equipo que podría jugar en la NBA). However, the use of podría is inferential in nature as it is used with sin ninguna duda ‘beyond doubt’. As the form-to-function approach and the function-to-form approach are combined in the present study, the non-epistemic/non-evidential uses of puede had to be separated from those that express the speaker’s personal judgement or an inference. So from a semasiological perspective, puede is also used with a deontic meaning, for instance. According to Pietrandrea (2005: 80), the Italian equivalent of puede, può, is generally more deontic and far less epistemic than the corresponding conditional form. The following text passages provide examples of poder in its deontic use. Such uses were excluded from the context-sensitive analysis of the modal after having had a quick look at the examples which GlossaNet had displayed: (382) Habrá una gran ‘yembada’, en la que el público puede llevar tambores Una novedad en este décimo aniversario de una cita que suele congregar numeroso público: una gran yembada. El público está invitado a traer sus tambores para participar en esta propuesta en la que se mezclarán [...] (El País 06/09/2010)

While in example (382) puede expresses permission, meaning ‘to be allowed to’, the following use of puede should be translated as ‘can’ because it expresses that the possibility of surfing is given and that consequently one can do so: (383) En los 250 kilómetros que suma la costa vasca se puede practicar el surf en 38 playas a lo largo de prácticamente todo el año. (El País 06/09/2010)

The following example could be analysed in similar terms: the possibility of doing so is given so that one can do so: (384) ¿Escribir es algo que uno hace a solas, o necesita a otros que le ayuden? Uno puede tener conversaciones útiles pero repetitivas consigo mismo, y puede obtener placer sexual por su cuenta, aunque tal vez sería alarmante que asegurase que ha hecho el amor consigo mismo. Se supone, en general, que la conversación y el sexo son más productivos e impredecibles con otros. (El País 06/09/2010)

In the following examples puede conveys ability; the subject is able to do so: (385) Los accionistas de la compañía de Mugardos sólo pueden ejercer el derecho de tanteo sobre participaciones que sean igual o menores a las que detentan, es decir, la

341 Xunta puede adquirir hasta un 10% y los Tojeiro, hasta un 18%. (El País 06/09/2010) (386) Vivió todavía ocho días más, aunque para sí mismo ya era un muerto. El condenado. El muerto vivo. Cuánto tiempo puede sobrevivir, inmóvil, el pez en el hielo. (El País 06/09/2010)

In the examples above the entity that is allowed or able to do X or that can do X is part of the proposition so that the modal is used non-deictically, whereby epistemic/evidential uses are deictic in nature (cf. Diewald 1999). The last example containing poder or rather the construction no puede ser que ‘it cannot be true that’ (or ‘it can’t be that’) represents an emphatic use of the modal. The speaker is obviously outraged so that the use is neither epistemic/inferential nor deontic etc. Instead, poder is part of a construction that expresses the speaker’s subjective or rather emotive opinion about [p]: (387) María se explica: “No puede ser que los clientes que él nos roba encima usen nuestros lavabos, que están más limpios”. (El País 17/08/2010)

This construction could be described as being equivalent to ‘it is unbelievable that’ so that it is used to express the speaker’s personal, affective judgement.

6.3 Summarising and concluding thoughts concerning the modals It was shown that deber as well as poder may be used to express – besides the speaker’s epistemic commitment – the speaker’s inference. In those cases, the use is deictic and subjective: the origo where the inference originates is beyond the sentence level. Deber was shown to express different kinds of conclusions, whereby it is not always possible to determine the exact kind of conclusion the speaker has drawn. For Italian deve Pietrandrea (2005) explains: […] the reason why [deve] expresses a strong degree judgment is to be sought in the fact that this form indicates that the speaker has objective knowledge, from which (s)he can infer the certainty of the conclusion introduced by [deve] (Pietrandrea 2005: 80).

In this connection I dissociate myself from the notion of objective knowledge. In line with Nuyts (2001a, 2001b) I prefer the notion ‘intersubjective’ instead of ‘objective’. Hence, it should rather be spoken about ‘intersubjectively comprehensible knowledge’ with regard to the use of deber (cf. also chapter 2.1.3). This knowledge the speaker bases his conclusion on is not always mentioned explicitly or referred to in the context. It is nevertheless reasonable to assume that the speaker, by using deber to utter ‘debe [p]’, relies on knowledge gained previ-

342 ously so that he also expresses an intermediate commitment at least (cf. Cornillie 2007a: 186), whereby it is not always possible to determine exactly deber’s modal values (cf. Haßler 1998: 176). Text passages containing an inferential use of poder were not easily found. Even though not having worked quantitatively, it was strikingly difficult to find uses of poder that could be unambiguously described as being inferential. If poder conveys an inferential reading, it is a meaning aspect that is contributed by contextually provided information. It is comparable to the use of the modal adverbs probablemente and posiblemente insofar as they may also express an evidential meaning, but if so, it is a meaning aspect that is contributed by the context. Although evidentiality is considered a meaning aspect that is encoded by deber, whereby in the use of poder evidentiality is regarded as a meaning aspect that is contributed by the context, I subscribe to Dendale/Tasmowski’s view (1994) that the Spanish counterparts of French devoir and pouvoir may also be used as evidential expressions. The main difference between the inferential use of deber and the inferential use of poder is that the inference expressed by the former often relies on certain premises. A further subgoal with respect to the modals was to compare Pietrandrea’s findings concerning the conditional form of Italian dovere with the Spanish conditional form debería. In comparing the indicative form and the conditional form of dovere (deve and dovrebbe) Pietrandrea explains that the former is used to convey unconditioned necessity and the latter conditioned necessity (cf. Pietrandrea 2005: 74-76). In the present study it was shown that debería often appears in highly subjective contexts where speakers express their subjective opinion about a certain state of affairs in a polite way. Speakers use debería to express that a third entity they are talking about ‘should do X’ or that ‘X should be this or that way’. If the conditional form of deber is used to express a conclusion, it is only a tentative one because the inferential process moves from uncertain premises (cf. Pietrandrea 2005: 88) and the truth value of [p] cannot be verified by the speaker (cf. Pietrandrea 2005: 75). The notion of intersubjectivity comes in if debería is accompanied by causal conjunctions by which the speaker’s subjective opinion is justified or explained further so that [p] becomes intersubjectively comprehensible. With regard to debe, the reportive reading could not be verified (cf. Squartini 2004 and Cornillie 2007a) and with reference to debería it can be said that if it conveys a reportive reading, it is due to the conditional morpheme and not due to the semantics of the modal as such. If debería conveys a reportive reading, the notion of polyphony comes in: the current speaker establishes another deictic centre where the quoted words originated and thus refers to another voice.

343 Podría was shown to express a personal, subjective judgement about a possibility in the speaker’s epistemic domain (cf. Pietrandrea 2005: 80), whereby, in certain contexts, podría may also be described as being used inferentially. In those cases its inferential reading depends on contextually provided information, as shown above. Furthermore, the construction puede (ser) que [p] has been analysed. In this construction, [p] is syntactically subordinated to puede (ser) que; [p] is – from a grammatical point of view – a complement to puede (ser) que. Hence, it was proposed to reconsider the notion of complement clause or subordinate clause for the structure puede (ser) que (cf. Thompson’s (2002) and Mithun’s (2009) line of argumentation). From a functional perspective, puede (ser) que behaves like a modal in its deictic use. Puede (ser) que also expresses a factuality judgement that originates in the speaker’s origo. No matter how a certain state of affairs is evaluated, the speaker is the centre of perspectivisation (cf. Diewald 1999: 211). Both modals poder and deber – provided that they are used to convey an inferential or epistemic meaning – are regarded as means of speaker’s perspectivisation since they are means that bring to the fore the speaker’s perspective of the narrated event.

344

7 A context-sensitive and functional approach to the use of the conditional form and the synthetic future While verbs of cognitive attitude or modal adverbs represent lexical means to express evidentiality, the Spanish condicional, futuro sintético and imperfecto145 are to be seen as grammatical forms that may convey evidential meanings. So in this chapter verbs in the conditional and future form will be dealt with. Giacalone Ramat/Topadze (2007: 7-8) observe that in some languages, evidential meanings “have developed as secondary meanings out of tenses and moods of the verbal system”. Such evidential meanings have developed out of the Spanish condicional and the futuro sintético. After considering the secondary meanings of the conditional, the evidential meanings that developed out of the future tense will be looked at. In the New Reference Grammar of Modern Spanish (2004), Butt/Benjamin list various types of conditions like remote and unfulfilled conditions and other ways of expressing conditions in Spanish (Butt/Benjamin 2004: 361-367). Text passage (388), for instance, represents an example containing a ‘typical’ sentence structure for a remote condition, that is, the use of the conditional is grammatically motivated: (388) Si los sevillanos eliminasen al Oporto, se enfrentarían en octavos al ganador del PAOK Salónica (GRE)-CSKA Moscú (RUS). (El Mundo 19/12/2010)

But how could the use of the conditional in (389) and (390) be explained? (389) Según explicó el propio Perelló en varias ocasiones, el Duque de Palma realizaría en el equipo de vela la labor de representar el área social y cultural con la colaboración de la fundación Cristóbal Gabarrón. Mientras que el ex periodista de TVE sería el director de comunicación del equipo. (El Mundo 23/12/2010) (390) [...] y cinco o más goles (4,96 por uno) en 17 duelos. Quizás por ello, sería lógico apostar por alguna de estas opciones: a 1,76 y 2,28 euros por uno se paga que haya más o menos de 2 goles [...] (El Mundo 01/12/2010)

Sería in (389) seems to have a quotative value (supported by the initial phrase ‘According to Perelló himself […]’), whereas in (390) it is used inferentially: the tentative conclusion is introduced by Quizás por ello ‘Maybe because of 145 In the present study, the Spanish imperfect will not be considered in detail because my colleague Verónica Böhm is currently working on her dissertation about aspectuality in Spanish and the imperfecto in particular. See also Böhm (2012) and Böhm/Hennemann (2012).

345 that’. Haßler (2010d: 101) has also shown that in French the “conditionnel can express reported speech, but also a conclusion”. In his study of evidentiality in Romance languages, Squartini (2001) – testing whether Willett’s (1988) or Frawley’s (1992) model is more appropriate for representing the internal structure of evidentiality in Romance languages (cf. chapter 2.1.2.2) – analyses the use of the conditional in Spanish. He shows that the condicional is used to express inference and is possibly used to mark quotation (cf. Squartini 2001: 321): “[...] the intrusion of the reportive value seems to be a new disrupting factor introducing a non-uniform feature, possibly producing a new development in the Spanish evidential system” (Squartini 2001: 322). Wachtmeister Bermúdez (2004), in contrast, is sure: “[...] el valor evidencial del condicional es doble. En el uso periodístico el condicional señala evidencia indirecta transmitida o mediada. [...] El otro valor [...] expresa evidencia indirecta inferida [...]” (2004: 7; cf. also Squartini 2004: 890). Therefore, I want to contribute to the study of the Spanish conditional in journalistic contexts (even though this particular use of the conditional cannot be said to be bound to journalistic contexts only). It will be shown that the condicional is not only used to express inference but also to mark – especially in news – (foreign) text import (notion adopted but translated from Thurmair 2006). Instances where the use of the conditional is grammatically motivated as in indirect quoted speech are excluded from the analysis, as the focus will be on the inferential use as well as on the reportive use of the condicional that has already produced a new development in the evidential system of Spanish. Therefore, the grammatically motivated use of the conditional, as in the following examples of indirect quoted speech (speech-introducing verbpast tense + conditional), is not thought to be the subject under discussion: (391) El propio Marín explicó recientemente [...] que para Sebastián sería importante que ocupe un español. (El Mundo 02/12/2010) (392) El 56% declaró que Guillermo sería mejor rey, frente a un 12% que se decantó por el príncipe de Gales. (El Mundo 29/11/2010) (393) Las primeras pistas sobre la dimensión británica de las revelaciones las había dado el domingo el columnista Simon Hoggart , que insinuó en el programa político matutino de la BBC que la filtración sería difícil por igual para conservadores y laboristas. (El Mundo 01/12/2010) (394) A pesar de no especificar si se trataría de conversaciones directas o indirectas, el ministro de Exteriores qatarí, jeque Hamad bin Yasem al Zani, jefe de ese comité, dijo en su discurso durante la reunión que sería un diálogo indirecto. (El Mundo 18/12/2010)

346 The three text passages that follow are interesting in so far as they contain the structure of speech-introducing verbpast tense + conditional + directly quoted fragment: (395) [...] con el entonces embajador español en Cuba, Manuel Cacho, adelanta la gestión que realizará Moratinos ante Clinton. El diplomático español le transmitió a su colega estadounidense que sólo esa especie de “teléfono rojo” entre La Habana y Washington, sería la única vía autorizada por el gobierno cubano “para efectuar pasos de envergadura hacia la conciliación con Estados Unidos”. (El Mundo 19/12/2010) (396) Tristeza por la muerte del cantaor Enrique Morente: - Ángeles GonzálezSinde, ministra de Cultura. Horas antes de su fallecimiento, cuando entró en coma cerebral, señaló que la de Morente sería una “pérdida inmensa para el país”. (El Mundo 16/12/2010) (397) Su compañero de partido Mike Huckabee señaló que “todo lo que no sea la pena de muerte” sería un “castigo demasiado suave” para el activista australiano de 39 años. (El Mundo 08/12/2010)

The structure of speech-introducing verbpast tense + conditional + directly quoted fragment could have caused the development that the conditional alone is nowadays used to mark foreign text import in journalistic discourse. If the speech-introducing verb is lost, it is still obvious – because of the quoted fragments – that the quoted words are attributed to another voice. The quotation marks are ‘allowed to leave’ if the foreign text import does not represent direct quoted speech. Hence, the conditional alone may remain to mark reported information. The development of the use of the journalistic conditional in Spanish could represent an interesting task for a diachronic study. Still in 2001, Squartini described the use of the conditional to indicate reported speech as ‘possibly producing a new development within the evidential system’. In a later study, in contrast, Squartini (2004: 890) shows that the conditional may undoubtedly function as a reportive marker. Here, I will merely be concerned with the use of the conditional from a synchronic point of view. What will also not be considered – despite the grammatically motivated use of the conditional due to indirect quoted speech – is the grammatically motivated use of the condicional in conditional sentences like in the following examples: (398) “Imaginad lo distinto que sería el mundo si WikiLeaks hubiera existido hace diez años”, escribe el cineasta. (El Mundo 16/12/2010) (399) Si las bonificaciones se financiaran con cargo al patrimonio de la Seguridad Social su transferencia sí sería una ruptura de la caja única y su transferencia podría ser inconstitucional. (El País 30/09/2010)

347 (400) ¿Cree que sería mejor como nación si careciese de periódicos, pero contase con un infinito número de blogs? (El País 29/09/2010) (401) [...] y si, por casualidad, termina recibiendo ese castigo que no existe, la culpa sería de aquellos que la han defendido demasiado estrepitosamente [...] (El País 29/09/2010)

Giacalone Ramat/Topadze (2007) and Squartini (2001) recognise the evidential values of the Italian conditional. Giacalone Ramat/Topadze explain: As is well known, the Italian (and, more generally, Romance) conditional has a number of uses which are older and more frequent, such as future in the past, unreality, counterfactuality, and the expression of an attenuated wish. Thus, the conditional originally had a temporal value of future in the past, then developed a number of modal values, which convey different degrees of factuality of the situation and of speaker’s commitment […] (Giacalone Ramat/Topadze 2007: 21).

But nowadays, in Modern Italian, the conditional is said to be restricted to the reportive function, that is, the Italian conditional is denied to have an inferential value (cf. Giacalone Ramat/Topadze 2007: 21-22). “The conditional dovrebbe [‘should’] on the other hand, marks tentative conclusions from uncertain premises and is an inferential evidential” (Giacalone Ramat/Topadze 2007: 23). Potrebbe ‘could’ is also said to convey inferential readings (Giacalone Ramat/Topadze 2007: 22). In the present study, the analysis of the Spanish conditional is limited to the forms sería and debería because my aim is to analyse a grammatical form and not a lexical one. Therefore the study is concerned with two lexical items in the grammatical form to be analysed. So GlossaNet was set up to search for deber and ser in the conditional form. Pietrandrea (2005) also observes that evidentiality can be marked by mood: This is so in the case of the Italian or French “reportive” conditional and the German subjunctive, which are employed to indicate that the speaker brings what has been said by someone else as evidence for the truth of his/her proposition (Pietrandrea 2005: 18).

Thurmair (2006), who analyses the use of the German subjunctive, explains that the change of mood from indicative to subjunctive may represent a sufficient signal to mark foreign text import: Moduswechsel vom Indikativ zum Konjunktiv kann nun als hinreichendes Signal anzeigen, dass in einem Text ein anderes Textstück eingebettet ist, das einer anderen Redeinstanz zuzuordnen ist, dass also ein Text importiert wird […] (Thurmair 2006: 73).

348 It will be illustrated that this use of the Spanish conditional is especially prominent in newspaper writing, whereby the condicional then is (usually) accompanied by a reference to an information source (see also example 389): (402) [...] la población pueden desembocar en disturbios que dinamiten el régimen y hagan saltar por los aires los férreos controles que atenazan a los 24 millones de habitantes. Para el ex ministro de Exteriores surcoreano Yoon Young-kwan, lo más peligroso sería que los militares, envalentonados por tener armas nucleares, “traten de distraer al pueblo con acciones audaces y posiblemente destructivas en el extranjero”. (El País 01/10/2010) (403) La especie más utilizada es la Pícea abias, más conocida como abeto rojo, una planta propia de zonas del norte de Europa, que fue introducida en la península ibérica en la zona de los Pirineos y en algunas áreas de la sierra de Guadarrama. No sería un problema plantar más abetos en estas zonas, cuenta Sánchez, pero esta medida tiene un límite de espacio a corto plazo, ya que si sólo entre los madrileños se pusiera de moda comprar un árbol [...] (El Mundo 26/12/2010)

Instances where the conditional is used to mark inference will be analysed as well (see also example 390): (404) Estos objetos poseen un enorme valor sentimental y profesional para la víctima”, dijo Rose en un comunicado. “Pero aunque son extremadamente valiosos, sería muy difícil venderlos porque son tan inusuales y distintivos que serían fácilmente reconocidos como propiedad robada, agregó. “Es posible que el instrumento sea ofrecido en venta en el comercio de antigüedades. [...]” (El Mundo 09/12/2010)

The fact that sería is used inferentially is underpinned by the causal conjunction porque (cf. also Blühdorn/Reichmann 2010): the speaker makes explicit why he supposes that ‘it would be difficult to sell them’, that is, why he has come to the conclusion that ‘it would be [p]’. Guentchéva (1994: 17-18) demonstrates for the French conditionnel journalistique that in certain contexts it is impossible to distinguish between an inferential use and a quotative one. The following instance of sería could be said to convey an inferential reading as well as a reported one: (405) Desde los Mundiales de Atletismo indoor, a la Fórmula 1, pasando por un renovado Open de tenis y el Global Champion Tour de hípica. La capacidad organizativa de la ciudad sería uno de los puntos fuertes de una hipotética candidatura valenciana que tendría que medirse, por ahora, a capitales de la talla de Tokio, Estambul y Roma. (El Mundo 29/12/2010)

As there is no further information provided, it is not clear whether the journalist uses the conditional form to indicate that he transmits reported information (possibly from the mayor of the city or an organising commitee) or to indicate that the information is based on his own inference.

349 It seems as if the same can be stated in certain contexts for the synthetic future. In the following example it is unclear whether the proposition containing será indicates an inference of the journalist or whether it is used by the journalist to attribute the words to the formerly quoted source (Jalili): (406) Jalili, durante su conferencia de prensa, repetía con frecuencia que no entendía por qué ‘lady Ashton’ no estaba a su lado respondiendo a los reporteros. La próxima reunión será quizás la ocasión de limar las asperezas entre ambos. (El Mundo 09/12/2010)

The use of the conditional to indicate reported information which is especially prominent in newspaper writing has already been intensively studied in French (cf. Dendale 1993, Haßler forthcoming, Kronning 2002, Sullet-Nylander 2006): the particular use of the French conditionnel is therefore also known as the conditionnel journalistique. Dendale, Haßler, Kronning and Sullet-Nylander, among others, also address its evidential nature. In “Le conditionnel de l’information incertaine: marqueur modal ou marqueur évidentiel?” Dendale (1993) is concerned with a semantic analysis of the epistemic conditional, whereby he is guided by the hypothesis that the conditionnel épistémique “a toujours la valeur évidentielle d’emprunt” (Dendale 1993: 167). The French conditional is said to semantically encode three properties: (a) Expression du caractère d’incertitude de l’information au conditionnel. (b) Indication de la reprise ou de l’emprunt d’une information à autrui. (c) Expression de la non-prise en charge par le locuteur de ce qui est affirmé (Dendale 1993: 165; emphasis in the original).

Dendale adds the important fact that it is sometimes context-dependent which semantic value is favoured: “Parfois le type de construction ou le contexte dans lequel le conditionnel est inséré fait ressortir l’une ou l’autre de ces trois valeurs” (Dendale 1993: 167). But at the end of his paper, answering the question of the paper’s title, he labels the conditional as an evidential marker (cf. following quote), the other semantic (epistemic) values being context-dependent or a logical consequence of the primary semantic value (cf. second quote): Le conditionnel épistémique est avant tout un marqueur évidentiel parce que sa valeur de base – c’est-à-dire la valeur qui est toujours présente, qui n’est pas soumise à des variations et qui en plus détermine ou explique les autres valeurs – est la valeur évidentielle d’emprunt (Dendale 1993: 175). Une information empruntée est par définition une information qui n’est pas créée par le locuteur lui-même, qui ne provient pas de lui, ce qui a pour conséquence que cette information peut parfaitement être incertaine pour lui (Dendale 1993: 174; emphasis in the original).

350 Consequently, Sullet-Nylander’s (2006: 128) critique of Dendale’s analysis of the example Six appareils argentins auraient été abattus (l’Express; Dendale 1993: 165) is not reasonable: the information transmitted in the example sentences is described as “incertaine ou hypothétique” (Dendale 1993: 167), at which Sullet-Nylander answers: En annoncant [sic] la nouvelle avec un [conditionnel journalistique], le journal transmet en premier lieu le fait que l’information est ‘empruntée’ et qu’il n’en est pas le premier asserteur, et donc qu’il ne la prend pas pleinement en charge. Selon nous, il n’y a pas à proprement parler d’incertitude de la part du locuteur (SulletNylander 2006: 128).

What Dendale states is the simple fact that reported information can be regarded as uncertain by the current speaker; “cette information peut parfaitement être incertaine pour [le locuteur]”. Even though it seems as if Sullet-Nylander and Dendale talk at cross purposes, Sullet-Nylander is right in what he states. Anyway, the most important point that Dendale makes in his paper is that the conditional morpheme is an evidential marker.146 In this regard, SulletNylander obviously agrees with Dendale’s line of argumentation as he also concludes: “la valeur sémantique et contextuelle la plus prégnante est bien celle d’emprunt de l’information à autrui” (Sullet-Nylander 2006: 135), that is, the journalistic conditional is one form of textual polyphony (cf. Sullet-Nylander 2006: 136; cf. also Korzen/Nølke 2001).147 Slightly in contrast to Sullet-Nylander (2006) but in the same line of argumentation with Kronning (2002) and to a certain extent with Dendale (1993) as he applies a modal value to the conditionnel, Haßler (forthcoming: 10148) attri146 That is why it is not regarded as comprehensible that Dendale (2001: 9) lists the ‘conditionnel d’emprunt/journalistique’ below the modal types of the conditional if it is said to be an evidential marker. 147 Sullet-Nylander treats one and the same linguistic phenomenon in terms of different categories. He deals in both papers “Titres de presse et polyphonie” (2002) and “Paratexte, contexte et intertexte dans Le Monde (2005): conditionnels journalistiques et discours rapportés” (2006) with the journalistic conditional or other evidential/polyphonic expressions but in one paper (2002) it is dealt with in terms of polyphony and in the other one (2006) in terms of evidentiality. Hence, polyphony and evidentiality must have something in common. More precisely, the polyphonic approach and the evidential approach to the analysis of reported information are related because both approaches may describe the different points of view which the speaker/journalist introduces as different deictic centres the original and then reported statement originated in (cf., for instance, Haßler 2010b). 148 The page numbers from Haßler (forthcoming) are taken from the word document with the title “L’expression de la modalité épistémique et de l’évidentialité dans les langues romanes”, which was made available to me.

351 butes two semantic properties to the journalistic conditional: “la non-prise en charge” – hence a modal meaning – and an evidential meaning. As she assumes, not only for the conditional, a ‘semiotic coincidence’, the conditionnel belongs, according to Haßler (forthcoming: 12), to the “marqueurs mixtes de la modalité épistémique et de l’évidentialité”. To my mind, the journalistic conditional’s primary meaning is to indicate foreign text import. The speaker, or rather journalist, may be fairly committed, may not be committed or may take a position in between regarding the transmitted information. But it is assumed that he is never totally committed and that is why the conditional’s meaning cannot only be evidential but even has also to be modal: this even holds for uses of the journalistic conditional if it is used with the quotative construction Según X ‘According to X’, that is, if a source of information is explicitly provided so that the information cannot be referred to as (totally) uncertain: (407) Según el diario ‘De Volkskrant’, que cita a fuentes de Holanda, Kenia y Somalia, uno de los doce detenidos sería familiar del comandante de Al Shabab Mohammed Garmashago, quien está vinculado con la red terrorista Al Qaida. (El Mundo 29/12/2010)

So there must be a reason why the conditional, instead of the indicative, is used149: the journalist himself does not want to be held responsible for the transmitted information that is evidentially marked by the conditional, since he may not be fairly committed. This would also explain why the quotative construction según X + indicative is not as often to be found in journalistic discourse as the construction según X + conditional. The former construction is indeed only used if the journalist assumes responsibility. He, in turn, does so, if he regards the source of the transmitted information as reliable. The following text passages represent examples that contain the structure según X + indicative present to transmit reported information:

149 I would like to thank my dear ‘Spanish sister’ Elena Pancorbo Guerrero for discussing with me the differences between the examples Según Anja, esto sería así and Según Anja, esto es así, who confirmed the hypothesis with the following answer: Se pueden utilizar los dos, pero con condicional suena mejor, porque con el indicativo suena como que estás muy muy segura de lo que estas diciendo y que no hay posibilidad de k sea de otra manera, sin embargo con el condicional das la posibilidad de que no eres perfecta y que podrías no tener razón en algún momento por cualquier motivo. Utiliza sólo el indicativo si estás 100% segura de lo que dices (Elena Pancorbo Guerrero, 21.03.2011).

352 (408) Según Ocampo, Gadafi tiene la responsabilidad “de hecho” sobre los actos de represión de las fuerzas libias contra manifestantes pacíficos ocurrida en febrero en su país. (El Mundo 05/03/2011) (409) En cambio, en Norteamérica la recuperación de las cuentas de resultados se debe más, según IATA, a la disciplina de las aerolíneas de esa región en la gestión de la oferta: están logrando rendimientos (yield) mucho mayores que en otras zonas del mundo. (El País 30/09/2010) (410) Según ha explicado a Efe el letrado y experto en Extranjería Carlos Juan González, en casos como éste, la legislación española establece el derecho de sangre para conceder la nacionalidad al niño, por lo que el bebé sería en principio nigeriano, como sus padres, siempre que el Consulado de Nigeria lo reconozca como tal. (El Mundo 15/12/2010) (411) Según detalló la empresa en un comunicado, la solicitud se debe a “la nula actividad comercial que tiene actualmente la empresa, unida a que la generación de liquidez por la venta de activos y cobro de deudas [...]” (El País 30/09/2010) (412) Según Lee, Corea del Norte pretende convertirse en 2012 en un país “fuerte y próspero”, y por ello, se debe “desmantelar su programa nuclear el próximo año”. (El Mundo 31/12/2010)

Reported information may even be indicated by the (indicative) future if the journalist regards the transmitted information as certain enough to assume partial responsibility; it is partial because the words are, in any case, attributed to another source: (413) Según fuentes de la Comisión Ejecutiva de CCOO, la capilla ardiente quedará instalada en la calle Lope de Vega, número 40, de Madrid, entre las 10.00 horas del viernes y las 11.30 horas del sábado. Media hora después el féretro será trasladado a la Puerta de Alcalá, donde se celebrará un acto de despedida a las 12.30. El cuerpo será enterrado a las 14.00 horas en el Cementerio Civil. (El Mundo 31/10/2010) (414) Este es el objetivo que persigue la aerolínea Spanair. Para lograrlo, algunos de sus miembros cambiarán y se reducirá el número de vocales, según fuentes del sector, probablemente de 10 a siete. Los cambios llegarán previsiblemente en septiembre. (El País 01/08/2010)

The following example underlines that the indicative (here: future) is only used – instead of the conditional – if the journalist indeed has evidence (‘see the guide for the liturgical acts in Barcelona and Santiago’) for the information he transmits: (415) [en] el Obradoiro de Santiago sólo se oirá una lectura y una petición en la lengua de Rosalía y de Castelao [Vea la guía de actos litúrgicos en Barcelona y en Santiago]. No sólo la celebración que va a tener lugar en la ‘Casa del Niño Dios’ de Barcelona será en catalán, sino que, además, la eucaristía principal, la que el

353 Papa va a presidir en el templo de Gaudí, también se centrará en la lengua de Cataluña. Más en concreto, el catalán estará presente desde el inicio de la celebración [...] (El Mundo 31/10/2010)

Gévaudan (2009) – even though focussing on selon/según as a construction with polyphonic function without making any reference to the verbal form – also illustrates that selon/según is used with the indicative (future) as well as with the conditional form: Selon nos informations, la chancellerie a défini des seuils d’activité en deçà desquels les tribunaux seront fermés. (Le Monde, 20.09.2007, p.1) La reforma, según el plan del Gobierno, debería estar terminada para la primera mitad de 2008. (El País, 20.09.2007, p.3; Gévaudan 2009: 117; my emphasis)

Hence, by implication, the meaning of the journalistic conditional cannot only be evidential, even though it may be its primary meaning (cf. also Dendale 1992). The following example is especially interesting with regard to the form of quotation and thus with respect to the point of discussion whether the conditional is used to mark uncertain information (cf. Dendale 1993): (416) En este sentido, según Soto, sería conveniente “reflexionar sobre el impacto del pantalán, recuperando así la unidad de la lámina de agua de la dársena”. (El Mundo 15/12/2010)

As Soto’s words are introduced by a conditional form, the conditional cannot be termed a morpheme always to indicate uncertain information because a large part of Soto’s speech is quoted directly. Soto’s words must have been available to the journalist in oral or written form in order to quote them directly. Thus, the transmitted information cannot be uncertain. The journalist himself may be uncertain concerning the truth of the information (that is what is addressed by Dendale 1993: 174) but the information as such cannot be termed uncertain. For Kronning (2002: 563), as well as for Haßler (forthcoming), the conditionnel journalistique “est un marqueur grammatical mixte, médiatif [évidentiel] et modal” because it indicates, on the one hand, reported information and, on the other, “non-prise en charge”. Kronning has also shown that the prepositional syntagm Selon X (the French equivalent to Spanish Según X) belongs to the most frequent ones to indicate the source if used with the conditional (cf. Kronning 2002: 565; cf. also Dendale/Coltier 2004). Besides conveying a modal meaning, the evidential function that is fulfilled by the journalistic conditional is to indicate reported information. It seems therefore necessary to deal briefly with quotation in general. With regard to quotation, Bühler’s notion of “Deixis am Phantasma” (Bühler 1999: 123-126) is applicable to quotation as not the current origo but

354 another origo with another/remembered deictic field represents the basis for the relation (cf. also Diewald 1999: 171). In the case of directly quoted speech the deictic system of the quoted speaker remains by which the quoting speaker uses a ‘foreign’ point of origin to transmit [p]: La peculiaridad del [estilo directo] es mantener intacto ese sistema [deíctico], lo que da lugar a la situación, sin duda curiosa, de que un hablante pueda decir “yo” sin referirse a sí mismo (Reyes 2002: 14). El hablante que cita en [estilo directo] se ha apropiado de un sistema de referencias ajeno, valiéndose de una licencia explotada sobre todo en el lenguaje de la conversación [...] (Reyes 2002: 15).

According to Reyes, direct quotation has one big advantage for the current speaker with regard to his own attitude because it is said not to be conveyed if quoting directly: [...] en [estilo directo] podemos “repetir” expresiones referenciales que no asumimos. Si alguien llama a otro “imbécil”, podemos citarlo en [estilo directo], repitiendo la palabra, sin arriesgar ninguna opinión nuestra [...]. Pero en el [estilo indirecto] podemos elegir la expresión; si decimos otra vez “imbécil” [...] la responsabilidad de la calificación pasa a ser nuestra (Reyes 2002: 20).

There is one little problem with Reyes’ remarks: she does not make any reference to verb mood if quoting indirectly. A speaker may quote indirectly “imbécil” using the indicative as well as the subjunctive. Imagine the following situation: A says to B Anja es una imbécil ‘Anja is a fool’, then B wants to tell C about A’s utterance. He now has two possibilities in doing so: A me ha dicho que Anja es una imbécil ‘A has told me that Anja is a fool’ and A me ha dicho que Anja sea una imbécil ‘A has told me that Anja be a fool’. Thus, the (indirectly) quoting speaker indeed can repeat A’s words without assuming any responsibility or rather without subscribing to A’s view. Besides, the quoting speaker does not convey his attitude towards [p]. In the Gramática Descriptiva, Maldonado González (1999) also speaks about a certain room for interpretation with respect to the correct interpretation of an indirect quotation and the current speaker’s attitude towards [p], whereby I wonder (again) why no reference is made to verb mood: En la cita directa, dado su carácter de contexto opaco, el contenido es siempre responsabilidad del hablante de la situación de enunciación original. En la cita indirecta, en cambio, el contenido puede presentar ambigüedad entre una lectura opaca o transparente150 según las expresiones que en él aparezcan sean interpretadas

150 The notions of lectura opaca (‘opaque reading’) and lectura transparente (‘transparent reading’) go along with the notions of lectura de dicto und lectura de re:

355 por el oyente como responsabilidad del hablante de la situación de enunciación reproducida o del hablante de la situación de enunciación reproductora. Dicha responsabilidad, en la transparencia deíctica, depende del punto de anclado; en las transparencias ilocutiva y atributiva, en cambio, se asocian con el hablante (Maldonado González 1999: 3592).

Maldonado González (1999) generally distinguishes between “discurso reproducido” and “discurso referido”. While the latter only refers to another utterance, the former is characterised by the literal reproduction of another utterance: “reproduce, por tanto, cuál fue el enunciado original, quiénes fueron el hablante y el destinatario del mensaje, y dónde y cuándo tuvo lugar esa producción original del enunciado” (Maldonado González 1999: 3556). If a quoting speaker reproduces the words of another speaker, he makes use of a speech-accompanying verb to indicate the quoted speech (cf. Maldonado González 1999: 3557). In doing so, it is interesting which speech-accompanying verb is chosen because the current speaker may (sub)consciously influence the interlocutor by his choice: “Todos los verba dicendi [...], excepto decir, [...] imponen [...] una cierta lectura al destinatario” (Maldonado González 1999: 3559; cf. also chapter 2.1.4). So by the speech-accompanying verb that is chosen the current speaker may convey his own attitude towards the quoted words. Decir is the only absolutely neutral verbum dicendi by which the speaker can quote in a neutral way. Speech-accompanying verbs are generally used by journalists to indicate reported speech (unless an interviewed person also quotes another person), regardless of whether they accompany a direct (418 and 419) or indirect (417) form of quotation: (417) [...] según la versión de los sindicatos presentes en Barajas. Estos señalan a las nueve de la mañana que Ryanair e Easyjet han intentado de madrugada (5.30 en el caso de Ryanair) cubrir viajes que no entraban en los servicios mínimos. Se lo impidieron, dicen las centrales mayoritarias, UGT y CC OO. Dos horas después, Aena informa de que entre las doce de la noche y las nueve de la mañana habían operado 32 vuelos, cuatro de servicios mínimos. (El País 01/10/2010) El [estilo indirecto], donde el hablante reformula textos, suele tener una lectura de re, lo que significa que las expresiones referenciales se interpretan dando prioridad a su contenido, a su referencia al mundo, sin atender, al menos de manera explícita, al modo en que fueron enunciadas originalmente. La cita directa, en cambio, exige una lectura atributiva, llamada lectura de dicto, según la cual se atiende a la referencia al mundo pero también a la codificación lingüística misma, que, en estas construcciones citativas, debe coincidir con la original. En la lectura de dicto, la responsabilidad de la expresión (y con ella del punto de vista, valoración, etc.) se atribuye al hablante citado (Reyes 2002: 20).

356 (418) [...] reuniones entre la Agencia de Seguridad Nacional (encargada del espionaje), el Departamento de Justicia y el FBI, que en estos momentos debaten la forma en que se puede reformar la normativa. “Se trata de intercepciones legales y autorizadas”, explicó ayer la abogada general del FBI, Valerie Caproni, en un correo electrónico. (El País 01/10/2010) (419) “Puedo morir tranquilo”, afirma Desde entonces, sigue sin aterrizar. “Estar allí arriba... era mi sueño desde los doce años. Desde guaje doy la vida por ellos, son la mejor banda del mundo”, explicó días después desde su Puerto de Vega natal, un pequeño pueblo marinero a 40 kilómetros de Galicia. “Mira tío, puede ser una frikada , pero voy a hacer un cartel a ver si pica”, le dijo a su cuñado Miguel en la furgoneta de camino al concierto [...] (El País 21/07/2010)

In comparing evidential expressions that mark inference and evidential expressions that indicate quotation Reyes (1996) makes the following distinction, explaining that quotation is generally bound to lower speaker commitment than inference: [...] cuando el evidencial indica una inferencia, la afirmación se restringe aunque los adverbios utilizados parezcan, contrariamente, reforzarla (es el caso de evidentemente, sin duda, etc.). Y cuando el evidencial es “citativo”, la afirmación suele debilitarse todavía más, porque el hablante no ha tenido ninguna experiencia de lo que afirma: se lo han dicho otros, y así lo indica con expresiones como dicen o con ciertas formas verbales (Reyes 1996: 31).

Such ‘globalising’ statements should generally be avoided: if a speaker quotes what a person he really trusts in has told him his commitment to [p] is certainly stronger than if he reproduces words of a person he never trusts in because this person is commonly known as a storyteller.151 The speaker’s commitment to an inference certainly varies according to the evidence he can rely on. Cornillie (2007a) also criticises the relation between the kind of evidentiality and speaker commitment which is often established in literature: The idea that speaker commitment to the propositional content is directly linked to the evidential type is less straightforward than assumed in the literature (cf. Palmer 1986: 54), where the hearsay information is generally considered as involving low speaker commitment, in contrast with the strong speaker commitment of inference (Cornillie 2007a: 41). […] the inferential statement does not prove to stand always for a strong speaker commitment. Moreover, the hearsay readings cannot automatically be considered as unreliable (Cornillie 2007a: 42).

151 Cornillie/Delbecque (2008: 39) also explain that “the value of reported knowledge is […] associated with the authority attributed to the source of information, so that speaker commitment can differ from one reported voice to another”.

357 Hence, speaker commitment and evidential type do not have a relationship of dependency (cf. also Fitneva 2001: 404). This study, however, will concentrate on a specific kind of indicating reported speech in journalistic contexts, namely on the reportive use of the Spanish condicional. In those cases, the conditional form is used to introduce a further voice so that it is one form of quotation. Reyes (2002: 8) says about quotations in general: Mediante la cita, un hablante atribuye a otro ciertas palabras: ya sea las palabras exactas, ya sea su contenido, ya sea una mezcla variable de ambas cosas. Estas palabras quedan, así, atribuidas: citar es siempre atribuir intencionalmente (Reyes 2002: 8).

By means of a quotation the speaker can distance himself from an enunciation or he can cultivate his “imagen del hablante, especialmente su imagen de persona digna de confianza, en el caso de los evidenciales” (Reyes 1996: 60). Consequently, “citas con función probatoria o ‘evidencial’” (Reyes 1996: 10) seem to be a stylistic device favoured by journalists. This is not surprising as in manuals on how to write journalistically it is often recommended to use quotes as they authenticate the news (cf. the manual by Cappon 2005). The conditional form, in (slight) contrast, is not primarily used to authenticate the news but to fulfil the evidential and modal function indicated above. If used temporarily, the future describes a state of affairs that is located in the future with reference to the here and now, that is, with reference to the time of speaking. If it is used as “futuro de probabilidad” (Volkmann 2005: 287), the speaker expresses his epistemic/evidential attitude concerning the information he transmits. The fact that the synthetic future fulfils – despite its temporal function – also the modal/evidential one is easy to understand because “[f]uturity is never a purely temporal concept; it necessarily includes an element of prediction or some related notion” (Lyons 1977: 677).152 The future is always unknown so that uncertainty is somehow inherent in the future:

152 Wachtmeister Bermúdez (2005) also assigns the future morpheme a modal meaning and an evidential one: […] el significado básico del morfema de futuro sería una referencia a la potencialidad del evento al que se asocia. Este significado tiene una vertiente modal (referida a la potencialidad de que un estado de cosas sea verdadero en el mundo dado un estado actual del mundo) y otra evidencial (inferencia a partir de los datos de los que se dispone), claramente relacionadas. […] la posibilidad de señalar eventos aún no ocurridos es un resultado de este significado básico (Wachtmeister Bermúdez 2005: 121).

358 […] reference to the future, in contrast with reference to the past or the present, is generally, if not always, tinged either with uncertainty or, alternatively, with expectancy and anticipation. Such attitudes are traditionally regarded as modal […] (Lyons 1995: 319). What is conventionally used as a future tense […] is rarely, if ever, used solely for making statements or predictions, or posing or asking factual questions, about the future. It is also used in a wider or narrower range of non-factive utterances, involving suppositions, inference, wish, intention and desire (Lyons 1977: 816).

The ‘nature of uncertainty’ which seems to surround the future tense is also addressed by Weinrich (1994). In addition, he terms the future ‘a form of expectation’ (cf. Lyons 1995: 319): Es [das Futur Tempus] hat die Aufgabe, relativ zur Aktzeit Information vorwegzunehmen […]. Es handelt sich […], wenn eine Information relativ zur Aktzeit vorweggenommen wird, zwangsläufig um eine unsichere Information. Sie ist ja noch nicht durch den Akt bestätigt, und es ist niemals ganz sicher, ob die Bestätigung wirklich eintreffen wird. Die Vorausschau dieses Tempus ist also notwendig eine Form der Erwartung (Weinrich 1994: 58).

The following example may illustrate that one can only ‘estimate’ a future event and that the synthetic future represents an appropriate means to express estimation: (420) Los expertos estiman que la demanda crecerá un 50% en 10 años. (El Mundo 23/12/2010)

While in examples (421) and (422) the future ‘simply’ describes events that are situated in the future, examples (423) and (424) additionally contain previsiblemente ‘presumably’. These examples could be described as representing the intermediate stage between ‘pure’ reference to a future event and the non-temporal use of the future. Previsiblemente + future shows that the event referred to is tinged with uncertainty, or more precisely, with ‘no-total-certainty’ so that it is neither purely temporal nor purely non-temporal: (421) El controvertido acuerdo, muy criticado por la oposición de centro izquierda, ha sido validado por una mayoría de 90 diputados de 179 en el Parlamento, y próximamente será presentado como un Proyecto de Ley. (El Mundo 19/11/2010) (422) Ríos ha desvelado que está preparando una nueva iniciativa cinematográfica, el Cinemart, que será un festival de cine de autor. Este nuevo proyecto se celebrará la primavera que viene pero será presentada el próximo diciembre en los Cines Verdi de Barcelona. (El Mundo 19/11/2010) (423) “El desarrollo del vehículo eléctrico requiere la elaboración de un programa de formación ocupacional de calidad específico destinado a un número limitado de profesionales que previsiblemente será el objeto real de los nuevos desarrollo”. En estos términos se expresa la Estrategia regional del vehículo eléctrico que fue

359 presentada recientemente por el vicepresidente segundo de la Junta, Tomás Villanueva. (El Mundo 19/11/2010) (424) Una ambulancia del Servicio de Atención Médica Urgente (SAMU) lo llevó al hospital comarcal de la Marina Baixa, en Villajoysa, donde ingresó con pronóstico grave, y previsiblemente será trasladado de nuevo en las próximas horas al General de Alicante. (El Mundo 19/11/2010)

Indeed, Bybee et al. (1994) even explain that the main function of the future is more a modal than a temporal one: […] the central functions in future grams are intention and prediction. It follows from this that future is less a temporal category and more a category resembling agent-oriented and epistemic modality, with important temporal implications (Bybee et al. 1994: 280).

De Haan simply states: “epistemologically, the speaker cannot know for certain that the event will occur” (de Haan 2006: 49), by which he establishes the connection between the (never purely) temporal use of the future and the ‘future of probability’. In other languages, speakers have the choice between different future morphemes by which they simultaneously express their degree of certainty concerning the state of affairs they are talking about: Bybee et al. (1994: 247248) speak about “pairs of future grams which, in addition to expressing prediction, bear an indication of how convinced the speaker is that the event will come about”. In other words, the speaker can choose between morphemes alongside a certainty scale that goes from “future possibility” up to “future necessity” (cf. Bybee et al. 1994: 248). Unlike Squartini (2001) and Wachtmeister Bermúdez (2004), Pietrandrea (2005, 2007) denies an evidential reading for the synthetic future in Italian. In her view, “the epistemic future has a genuine epistemic meaning, without any evidential colouring” (Pietrandrea 2007: 40): The future […] may be in some cases engendered by the existence of a source of evidence, but it never fulfils the function of explicitly encoding it. To put it in another way, the future does not serve the purpose of marking information as inferential, it does not have therefore ‘source of information’ as its core meaning, as a true evidential must have […] (Pietrandrea 2007: 42).

Pietrandrea (2007) claims to be able to prove the fact that the future is genuinely epistemic in nature as it “can be employed in purely conjectural contexts; i.e. in contexts lacking any evidential source” (Pietrandrea 2007: 41) and wants to illustrate this with the help of the following example: Saranno le otto e mezza, immagino ‘It must be eight thirty, I guess’.

360 She then explains: “It is clear that these contexts cannot be considered as evidential by definition: the speaker simply expresses his opinion, without referring to any evidential source” (Pietrandrea 2007: 41). But is it not obvious that the speaker expresses this utterance because he must have had some kind of evidence to infer that it must be eight thirty? The parenthetically used immagino ‘I guess’ (or: ‘I imagine’) rather seems to confirm that the speaker transmits a tentative conclusion. Whether the speaker relies on his biological clock or whether he relies on the altitude of the sun – there must be some kind of evidence he relies on (cf. Squartini 2008). Otherwise he would not have said that it must be eight thirty. So the speaker obviously bases his reasoning process on ‘something’, let it be world knowledge (cf. Squartini 2008). That is why even Pietrandrea (2007: 41-42) admits – referring to Squartini (2008) – that “the genuine epistemic nature of the future can be questioned by the fact that the epistemic future […] is grammatical in generic inference contexts […]”. Anyway, it seems to be difficult and certainly not appropriate to analyse examples that are offered without context. If having a look at context-embedded instances of the synthetic future in Spanish, it becomes obvious that it may represent a linguistic means to express inference (even if no clear evidence is provided – such as in example (425) – to infer that [p]): (425) El medio no es el culpable. Lo es el mal uso. Y lo que hará sobrevivir a los medios tradicionales serán las virtudes de toda la vida a las que tendrán que sumar una gran formación de sus trabajadores. Será imprescindible sumar análisis en pocas horas al simple anuncio de la noticia. Y un análisis que aporte algo al lector de prensa que tiene un nivel de conocimiento bastante elevado y al que le dará bastante igual enterarse un par de horas antes [...] (El País 12/08/2010) (426) “Nos conocemos muy bien. No hará falta ningún informe”, dijo Bryant. “Será una eliminatoria muy dura. Deberemos mejorar nuestra concentración para no tener que correr sin sentido como hicimos en algunas fases del partido ante Oklahoma, aunque ellos fueron muy tenaces”, afirmó Phil Jackson, el técnico. (El País 04/05/2010)

In example (426) the use of será is not obviously inferential although it is highly likely because the speaker certainly knows the teams which are in the same group as his team. This may lead him to conclude that the preliminaries must be/will be hard. However, this contextual information is not linguistically provided (such as in the example Saranno le otto e mezza, immagino). The use of deberemos, in contrast, is clearly inferential. The speaker says: ‘we will have to improve our concentration so that we do not have to run without purpose as we did in a few phases of the match against Oklahoma’ (Deberemos mejorar nuestra concentración para no tener que correr sin sentido como hicimos en algunas fases del partido ante Oklahoma). The fact that ‘deberemos [p]’ clearly repre-

361 sents a conclusion drawn out of formerly made experiences during the match against Oklahoma. The following example could be analysed in similar terms: it is highly probable that Ferrer does not say unfoundedly that he ‘must be/play very aggressive(ly)’. He certainly knows how Nadal plays and infers from this fact that – against him – he should be an aggressive player: (427) “Estoy jugando muy bien esta temporada. Contra Nadal deberé ser muy agresivo”, cerró Ferrer, implacable en su labor de desgaste. (El País 04/05/2010)

As already indicated (cf. chapter 2.1.4), Gévaudan (2009: 118) analyses text passages that contain a certain use of a synthetic future form as an expression of implicit polyphony – in terms of evidentiality: as an expression of reported evidence. In Squartini’s (2001: 321) view, in contrast, the Spanish and French synthetic future form is exlusively used to express inference. As indicated above and in chapter 2.1.2.2, I – in line with Gévaudan – could find text passages containing the synthetic future form used with phrases marking quoted speech. That is why I will have a closer and comparing look at the quotative use of the future and the quotative use of the conditional, e.g. on utterances containing según X + future and según X + conditional. But how could it be explained that the synthetic future in Spanish rather fulfils modal/evidential functions – or more precisely, inferential functions (cf. Squartini 2001; Wachtmeister Bermúdez 2004) instead of the temporal function? The temporal function is mostly expressed by the periphrastic future form (ir a + infinitive). Instead, the synthetic future has become unreliable and redundant in its temporal value. That is why the form could be re-functionalised for modal/evidential uses (cf. Haßler 1997: 42; cf. also Haßler 1998: 176). The same process is said to be especially applicable to the imperfecto that – due to the re-functionalisation process – has taken on modal/evidential values (cf. Haßler 1997: 42; cf. also Haßler 2001). However, Dessì Schmid (2010: 47) explains that the modal use of the imperfect already showed up in Medieval Castilian. By all means, [en] español [...] existen determinadas formas verbales que pueden expresar, por un lado, contenidos temporales en que interviene la anterioridad y, por otro, contenidos modales relacionados con la irrealidad o el alejamiento en general (Rojo/Veiga 1999: 2895).

That is where the modal/evidential use of the Spanish imperfect belongs. Reyes (1996) highlights the evidential, more precisely, the quotative use of the imperfecto explaining it with the aid of the example Mañana daba una conferencia María: El uso, aprentemente anómalo, del imperfecto daba para hacer una referencia al futuro se debe a la intención del hablante de señalar que su información es de

362 segunda mano: se puede agregar fácilmente una expresión citativa como según me dijeron. [...] El enunciado [...] ejemplifica un caso de cita en el que se trata de deslindar responsabilidades: el hablante no asume enteramente lo que dice, y para eso indica (por ejemplo, usando el imperfecto de ese modo peculiar) que su conocimiento procede del testimonio de otro (Reyes 1996: 12).

The same use of the imperfecto is addressed by the Manual of the RAE: El llamado imperfecto citativo o de cita permite al hablante eludir la responsabilidad directa por sus palabras y presentarlas como información emitida por otros, con lo que se logra, de nuevo, evitar la rudeza que se asocia con el presente. Así pues, la oración ¿Tú jugabas al fútbol, no es cierto? significa aproximadamente [...] ‘¿Es cierta la información (conocida, oída, etc.) según la cual tú juegas al fútbol?’ (Manual 2010: 445).

So while the synthetic future acquired an epistemic/inferential value, the imperfect acquired – according to Reyes (1996) and Haßler (2011b) – a quotative one. In so doing, the imperfecto has partly lost its prototypical temporal reference, namely describing a past action in progress (cf. Haßler 1997: 41; cf. Haßler 2011b: 7). In contrast to Reyes (1996) and Haßler (2011b), Wachtmeister Bermúdez (2005) highlights another evidential use of the Spanish imperfect: he explains that the imperfecto is employed if the speaker wants to signal that the interlocutor also has access to the information source. He distinguishes between “acceso privativo [a la fuente de la información]”, and “acceso compartido [a la fuente de la información]” (Wachtmeister Bermúdez 2005: 16). The former implies that only the speaker has access to the information source, whereas the latter implies that the speaker as well as the interlocutor has access to the source of information, which may be expressed through the use of the Spanish imperfect. The most important fact to mention is that Wachtmeister Bermúdez himself notes that the opposition ‘private access’ vs. ‘shared access’ does not deal with the access to the kind of information source but only with the question whether both speaker and hearer have access to it or the speaker only: Esta oposición entre acceso privativo o acceso compartido a la información es evidencial en el sentido de que se relaciona con la fuente de la información y el acceso de los participantes a ella: tanto el hablante como el oyente lo tienen en [(a)] [...], sólo el hablante en [(b)] [...]. Sin embargo, este valor no tiene cabida en el sistema de Willett (1988), dado que ni el imperfecto en [(a)] ni el presente en [(b)] [...] establecen el modo en el que se accede a la información (directa o indirectamente, etc.) sino que sólo especifican quiénes (no) tienen acceso a ella (Wachtmeister Bermúdez 2005: 16). (a) ... y simplificando nos da entonces 1,5 por la raíz cuadrada de 2, y la raíz cuadrada de 2 era 1,4142, así que el resultado final...

363 (b) ... y simplificando nos da entonces 1,5 por la raíz cuadrada de 2, y la raíz cuadrada de 2 es 1,4142, así que el resultado final... (Wachtmeister Bermúdez 2005: 15).

Then Wachtmeister Bermúdez explains that in example (a) the speaker, by using the imperfect era, signals the interlocutor that he also has/should have access to the information source, that is, to the interim result. The speaker reminds the interlocutor. In example (b), in contrast, by using the indicative es, the speaker informs the interlocutor presenting the interim result as new and unknown information (cf. Wachtmeister Bermúdez 2005: 16). Dixon (2003: 178) also observed that evidentials, through conventionalised uses, may develop special meanings in connection with the second or third person. The quotative suffix in Jarawara, for instance, “can be used to remind someone of their [sic] own words” (Aikhenvald 2004: 234): ti-fimiha-mone, ti-na. ‘You were hungry, you said’ (cf. Aikhenvald 2004: 234; my emphasis).

In my view, one could also deal with this use of the imperfect in terms of intersubjectivity: by using the imperfect, the speaker expresses that he (assumes to) share(s) the access