Rethinking Presuppositions: From Natural Ontology to Lexicon [1 ed.] 1527539563, 9781527539563

Citation preview

Rethinking Presuppositions

Rethinking Presuppositions: From Natural Ontology to Lexicon

By

Marco Fasciola

Cambridge Scholars Publishing

Rethinking Presuppositions: From Natural Ontology to Lexicon By Marco Fasciolo This book first published 2019 Cambridge Scholars Publishing Lady StephensonLibrary, Newcastle upon Tyne, NE6 2PA, UK British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Copyright© 2019 by Marco Fasciolo

All rights for this book reserved. No part ofthis book may be reproduced, stored in a retrieval system, or transmitted, in any fonn or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior pennission ofthe copyright owner. ISBN (10): 1-5275-3956-3 ISBN (13): 978-1-5275-3956-3

TABLE OF CONTENTS

Acknowledgements ................................................................................... ix 1 . Introduction 1 . 1 Which are the presuppositions? 1.2 Structure of the book

............................................................................................

......................................................

.......................................................................

Part I: Contingent

vs

1 1 3

Ground presuppositions

2. A standpoint for presuppositions 2.1 From free will to stopping smoking 2.2 A functional notion of presupposition 2.3 What are ground presuppositions? 2.4 Consequences for the treatment of presuppositions

9 9 11 12 14

............................................................

................................................

...........................................

................................................

......................

3. The indeterminacy of presuppositions 3. 1 Presupposition as an illocutionary ac!.. ......................................... 3.2 Presupposition as a propositional attitude 3.3 The indeterminacy of presuppositions 3.4 Dissolving indeterminacy

17 17 21 24 26

4. Presuppositions beyond cooperation 4.1 Presupposing vs. Presupposed conten!.. 4.2 Beyond felicity conditions 4.3 The limits of cooperation 4.4 A consistent development of a 'Gricean' approach

27 27 28 29 30

..................................................

.....................................

..........................................

.............................................................

....................................................

........................................

............................................................

..............................................................

......................

5. Prototypical presuppositions 5. 1 Survival under negation 5.2 The inaccessibility of presuppositions 5.3 Why are there contingent presuppositions? 5.4 Why have scholars focused on contingent presuppositions?

.................................................................

................................................................

..........................................

..................................

........

31 31 33 38 39

vi

Table of Contents

6. A hierarchy of presuppositions 41 6.1 First criterion: to forget, to teach, to learn . . . ................................. 42 6.2 Second criterion: the limits of felicity conditions ......................... 45 6.3 Third criterion: tautologies ............................................................ 47 6.4 Contingent vs. Ground presuppositions ........................................ 50 .............................................................

7. A model for presuppositions: the looking-glass ................................... 7.1 The paradox of the looking-glass .................................................. 7.2 Negative interpretation 7.3 Positive interpretation ..................................................................

...................................................................

53 53 54 55

Part 11: Issues on presuppositions

8. Posited and presupposed conten!.. 8 . 1 Some difficulties in the distinction 8.2 First option: rejecting the distinction 8.3 Second option: reinterpreting the distinction 8.4 Lexical presuppositions

........................................................

...............................................

............................................

................................

.................................................................

9. Presupposition and truth conditions 9.1 Truth conditions vs. Ground presuppositions 9.2 Presuppositions and necessary conditions ..................................... 9.3 Presuppositions and necessity 9.4 Presuppositions and truth ..............................................................

......................................................

................................

.......................................................

59 59 60 61 63 65 65 67 68 73

10. Presupposition and inference .............................................................. 75 10.1 Inference vs. Ground presuppositions ......................................... 76 10.2 Presupposition vs. Access to a presupposition 77 10.3 Presupposition and accommodation 80 10.4 Resurrection for presuppositions 81 10.5 Classification of presuppositions ................................................ 82 ............................

............................................

.................................................

1 1 . Presupposition and anaphor ................................................................ 1 1 . 1 Presupposition as anaphor. 1 1 .2 Anaphor vs. Ground presuppositions 1 1 .3 Anaphor vs. Presupposition: separate phenomena 1 1 .4 Presupposition of identification ..........................................................

..........................................

......................

..................................................

85 86 87 88 91

Rethinking Presuppositions: From Natural Ontology to Lexicon

vii

Part Ill: Towards a study of ground presuppositions

12. How to study an extra-linguistic phenomenon linguistically? ............ 97 12.1 Exhibited predication .................................................................. 97 12.2 A difficulty: experience is a priori consistent ............................. 99 12.3 Sentence meaning as a model of experience's consistency ...... 100 12.4 Conceptual conflicts as access to ground presuppositions ........ 101 13. Lexicon and ontology ....................................................................... 13.1 The polysemy objection ............................................................ 13.2 The specificity objection ........................................................... 13.3 The freedom of the lexicon .......................................................

10S 106 108 110

14. Use of a predicate ............................................................................. 14.1 Predicative use .......................................................................... 14.2 Hyper-classes vs. Object-classes ............................................... 14.3 Relation and classification ........................................................ 14.4 Philosophical lexicography .......................................................

1 13 1 14 119 122 124

IS. Sketches of a philosophical lexicography ......................................... IS. 1 Is a human embryo a person? ................................................... IS.2 Are humans kinds of animals? .................................................. IS.3 Are vegetables living beings? ................................................... IS.4 Do vegetables have the experience of life and death? .............. I S. S Are vegetables sentient beings? ................................................ IS.6 Do vegetables have a body? ...................................................... IS.7 Do vegetables have a personal identity? ................................... IS.8 Do vegetables communicate with us? ....................................... IS.9 Do animals share experiences with human beings? .................. IS. 1 0 Which practices do animals share with human beings? .......... I S . 1 1 Do animals have a will? ......................................................... IS.12 Do animals have a personality? .............................................. IS.13 Synthesis .................................................................................

127 128 129 130 131 133 134 13S 137 138 140 142 143 144

16. Conclusions and perspectives ........................................................... 147 16.1 Looking backwards ................................................................... 147 16.2 Looking forwards ...................................................................... 149 References

..............................................................................................

IS7

ACKNOWLEDGEMENTS

The principal idea ofthis book comes from my These de Doctorat es Lettres defended at University of Geneva on the 12 th October 2009. It develops Chapter VIII of Michele Prandi's monography The Building Block of

Meaning (Amsterdam-Philadelphia: John Benjamins, 2004). During my PhD, I had two complementary mentors: Professor Emilio Manzotti and Professor Michele Prandi. The former was like a microscope: he taught me the need for fine graded linguistic analysis. The latter was like a telescope: he taught me the need to unify apparently distant phenomena. So, in a way, I was in a Pascalian situation: in between the immensely small and the immensely large. 'Whether the result is theoretical schizophrenia or not, I leave the reader to decide. Be that as it may, this work would have not been possible without them, and without the friendship and the teaching of Professors Gaston Gross and Georges Kleiber, who supported my research long after the completion of the PhD. This book is dedicated to all of them. I would like to give further and special thanks to Professor Andrea Bonomi, whose acute observations helped me to improve my arguments, and to Wendy Booth (EditWise), John Humbley and John Ingamells who revised my English.

CHAPTER 1 INTRODUCTION

1.1 Which are the presuppositions? This book is about presuppositions. Presupposition is an idea that is either loved or hated. Some authors, like G. Frege, P. F. StrawsOll, I. L. Austin, P. & C. Kiparsky, Ch. Fillmore, O. Ducrot and R. Stalnaker, have been attracted by it and have attempted to highlight the distinction between posited and presupposed content. Some others, like B. Russell, D. Wilsoll, L. Karttunen and S. Peters, have been suspicious of it and have argued either that that distinction is artificial

or

that there is not a unitary

phenomenon to place under the label presupposition. So, maybe, presupposition is a topic that reveals something about the personality of the people who are studying it. Be that as it may, in this debate there is a question that is never asked, namely: Which are the presuppositions? Of course, the specialised literature offers examples of presupposed contents: that someone beats his wife (He stopped beating his wife), Ibat someone has children (His children are bald), that someone cheated (She

discovered that he cheated on her), and so on. However, it would be pointless to make a list of these presuppositions. The reason is that the previous contents, in themselves, are not presuppositions at all, but ordinary contingent facts. Any contingent fact can become a presupposition through some linguistic structure or manipulation.

So, the

specialised

literature has

not

focused on

presuppositions, but rather on the means to make presuppositions: the so­ called "triggers". This, however, amounts to behaving like an art critic who is more interested in brushes and chisels than paintings and statues. In

2

Chapter 1

other words, the very debate about presuppositions turns out to be built on an epistemological paradox: presupposition is not envisaged as a positive notion. Indeed, all authors-those sympathetic to the intuition of presupposition and those who are not-have focused on contents working as consistency conditions of some utterance

Of,

at best, as felicity conditions of some

speech act. The tacit assumption is that this is the only kind of presupposition. The hypothesis defended here, on the contrary, is that besides contents which enjoy the status of presupposition for the time of an utterance or a speech act (and then come back to being ordinary facts), there are contents that enjoy the status of presupposition forever, and hence are the presuppositions par excellence. I call the fmmer contingent presuppositions and the latter ground presuppositions. Ground presuppositions constitute the natural ontology that P. F. Strawson describes as: [ ] a massive central core of hlllllan thinking which has no history or none recorded in histories of thought; there are categories and concepts which, in their most flUlClamental character, change not at all. Obviously these are not the specialities of the most refined thinking. They are the commonplace of the least refined thinking; and they are yet the indispensable core of the conceptual equipment of the most sophisticated human beings. (Strawson 1959:10) . . .

In my opinion, the distinction between contingent and ground presuppositions is crucial in order to understand both the development of the classic debate and the phenomenon of presupposition itself. The opposition between contingent and ground presuppositions highlights the fact that the former have two sides. On the one hand, if one looks at a contingent presupposition when it works as a precondition for the consistency of some other utterance, one has the intuition that it is placed below the threshold of assertion, negation or question. On the other hand, if one looks at the content of a contingent presupposition in itself, that is, beyond its working as a presupposition, one may ascertain that it can be plainly asserted, negated or questioned. Now, the controversy about presuppositions can be seen as a pendulum swinging between these two alternatives. The pivot of this pendulum is that both the founduig fathers

Introduction

3

of the debate and their critics considered only contingent presuppositions. Hence, both of them are right in their 0\Vll respect, and the result is that either the debate continues indefmitely or it is abandoned. In the light of the opposition between contingent and ground presuppositions, moreover, the whole phenomenon turns out to be manifold, without giving up its unity. This phenomenon describes an ascending curve, ranging from fleeting discursive presuppositions, providing the ground for some lines of a dialogue, to stable ontological ones, providing the ground for our whole form of life. All the points on this curve share the same function of presupposition. The difference lies in the stability of this function. At the top of the curve, ground presuppositions behave necessarily

as

presuppositions, that is, in se,

independently from any other action and without the need for any trigger or special attitude. At the bottom of the curve, discursive presuppositions behave as presuppositions contingently, that is, with regard to some discursive action and thanks to some trigger or attitude, without which they stop behaving as presuppositions. In such a framework, the importance of both kinds of presupposition is emphasised. On the one hand, ground presuppositions acquire a central place because they turn out to be the prototypical presuppositions. In other words, discursive presuppositions are presuppositions insofar as they sometimes behave with regard to

some utterances

like

ground

presuppositions always behave with regard to our whole life. On the other hand, discursive presuppositions become interesting precisely insofar

as

they are not prototypical presuppositions, that is, because they highlight a variety of linguistic strategies in order to make presupposition content that, in itself, is not a presupposition.

1.2 Structure of the book This book consists of three parts. The first part (Chapters Two to Seven) puts forward the distinction between contingent and ground presuppositions.

4

Chapter 1 Chapter Two introduces this idea through a functional conception of

presupposition, while others defend it from two complementary points of VieW. Chapters Three and Four, on the one hand, discuss some influential accounts. The distinction between contingent and ground presuppositions is firstly defended as the solution to an intrinsic indetenninacy affecting Ducrot's

and

Stalnaker's

insights

and, secondly,

as the natural

development of the cooperative principle. Chapters Five and Six, on the other hand, focus on the very notion of presupposition. The fmmer argues for the distinction between contingent and ground presuppositions on the basis of the core intuitions which reveal the phenomenon itself (such as the "survival under negation"). The second sketches a hierarchy, ranging from contingent presuppositions to ground ones, by means of three substantive criteria. Chapter Seven ends the first part by offering a model which will orientate further discussion. The second part of the book (Chapters Eight to Eleven) discusses some classical issues in the light of the distinction between contingent and ground presuppositions, namely: posited vs. presupposed content (Chapter Eight); presupposition vs. truth conditions (Chapter Nine); presupposition vs. inference (Chapter Ten); and presupposition vs. anaphor (Chapter Eleven). The leitmotiv of this part is the separation between two souls of contingent presuppositions. On the one hand, they work as presuppositions in respect of some utterance presupposing them. On the other hand, unlike ground ones, in themselves they are contingent facts like any other. The problems raised by the aforementioned issues stem from a lack of distinguishing between these two aspects. The third part (Chapters Twelve to Fifteen) is devoted to exploring and

Introduction

5

exemplifying the study of ground presuppositions. Chapters Twelve and Thirteen answer the question: Is it possible to study ground presuppositions? The former suggests that a ground presupposition can be seen as a category put on an entity through a practice, and that this category can be described by examining sentences' consistency conditions. The latter discusses Ryle's and Sommers' proposals concerning the study of ontological categories embedded in natural language, and focuses on the relationship between lexicon and ontology. Chapter Fourteen answers the question: How should we study ground

presuppositions? It puts forward an operative notion of use of a predicate in order to elucidate ground presuppositions through an extensive and systematic

exploration

of lexicon.

Hence, the study

of ground

presuppositions takes the form of a "philosophical lexicography": a natural development of Strawson's descriptive metaphysics. Chapter Fifteen offers some examples of such a philosophical lexicography. The result is a set of general propositions in the style of Moore's commonplaces. These propositions are instances of ground presuppositions; they can be consistently listed and they ground the consistency of both our practices

as

regards vegetables, animals or human

beings, and a network of linguistic phenomena far beyond the utterances usually considered in the study of presuppositions. Finally, Chapter Sixteen sums up the distinction between contingent and ground presuppositions and pleads for a genuine linguistic turn in philosophy. This turn consists of envisaging conceptual analysis as the redaction of a "philosophical dictionary": a dictionary of presuppositions providing the ground for the consistency of natural lexicons. The title of this book reproduces the logical order of presuppositions:

from ontological, prototypical ones to ephemeral, discursive ones. The sequence of chapters, instead, is intended to lead the reader from the latter to the former.

PART I CONTINGENT VS. GROUND PRESUPPOSITIONS

CHAPTER 2 A STANDPOINT FOR PRESUPPOSITIONS

In the Introduction, I pictured the phenomenon of presuppositions as a curve ranging from ground presuppositions to contingent ones. This implies two things: a) that there is a continuum between natural ontology-the "central core of human concepts which has no history" (Strawson 1959:10)-and the predicate of a specific historical language; and b) that there is one consistent notion of presupposition capable of encompassing these extremes. This chapter is devoted to arguing for such a continuum and to putting forward such a notion of presupposition.

2.1 From free will to stopping smoking Let us consider a classic moral philosophy problem, free will. Free will does not coincide with the whole concept of freedom, but only with a part of it. Indeed, as regards freedom, two notions can be distinguished (cf. Prandi 2004:330-333): an empirical notion and an ideal one. Free will coincides with the latter. We use an empirical notion of freedom when we say, for instance, that there is little or no religious freedom, or when we say that human beings are not free because they are 'programmed' by their genetic code. In the first case, we can fight to achieve a higher degree of freedom; in the second, we must admit that human freedom is almost zero. Using the empirical notion of freedom means to measure our freedom with regard to external forces (socio-political, biological, physical laws) or to internal forces (our passions and desires). Using the ideal notion of freedom, on the other hand, simply means to state that human beings have free will because they are human beings, that is, the very concept of a human being is unthinkable without free will. Of course, this notion of freedom is a

10

Chapter 2

tautology, but a tautology we carmot live without. The ideal and the empirical notions of freedom distinguished above are not contradictory: the fmmer is the condition with which we can measure the latter. Imagine a man chained to a chair. Surely, he cannot move and we can coherently say (1): ( l a) That man is notfree to stand up. (lb) They prevented him/ram walking away. Now, imagine a statue of a man chained to a chair. Surely, the statue cannot move, but nobody would say (2): (2a) ?That statue is notfree to stand up. (2b) ?They prevented that statue from walking away. The difference between the person and the statue is not empirical, because in both cases something carmot happen according to a physical cause. The difference is ideal: for the human being, it makes sense to raise the question about his freedom; for the statue, by contrast, this question does not arise. The man has zero degree of empirical freedom because he is ideally free; the statue, by contrast, does not have zero degree of empirical freedom because it is not ideally free. In this sense, empirical freedom-a lot, a few, none of it-presupposes ideal freedom. The conclusion is clear. On the one hand, it makes sense to negate, affitm and inquire into someone's empirical freedom because we take for granted that he is ideally free. On tbe other hand, it is meaningless to negate, affirm or inquire into a statue's empirical freedom because we do not take for granted tbat it is ideally free. Ideal freedom is tbe consistency condition to be empirically free or enslaved. Now, consider the question (3): (3) Did George stop smoking? The problem of the empirical freedom of a statue does not arise, just as the question (3) does not arise if George never smoked. The problem of

A standpoint for presuppositions

11

the empirical freedom of a man does arise, just as question (3) arises if George did smoke. The presupposition which makes it coherent to raise (3) (namely, that George smoked) is a contingent fact devoid of any philosophical interest. The presupposition which makes it coherent to measure human freedom (namely, that human beings have free will) casts light on our moral ontology. Both of them, however, are presuppositions. So, after all, there is a continuum starting from free will and ending in predicates like to stop doing something. This continuum is grounded not in their content, but in their function of consistency conditions.

2.2 A functional notion of presupposition Prandi (2004:233-234) distinguishes between a functional notion of presupposition and an absolute one. As regards the absolute notion, the relevant question is: Is this idea a presupposition or not? As regards the functional notion, the relevant question is: In relation to which practice does this idea work as a presupposition? According to this second question, presupposition is a function, the function of a consistency condition carried out by an idea towards a practice. The idea that George smoked works as a presupposition in relation to asking: Did George stop smoking? If that idea is false, this question is infelicitous, and this is all that happens. If someone casts doubt on that presupposition, he rejects the speaker's question and breaks a contingent communicative exchange. Instead, if someone casts doubt on the idea that people enjoy free will, he rejects our whole human way of life and nobody, not even a sceptic, is ready to live coherently with this refusal. Maybe a cynic would do so, but living in a barrel for the sake of being coherent with a philosophy is actually a practical confutation of that philosophy. Be that as it may, the differences between these presuppositions do not lie in their working as such, but rather in the extension of the practice in relation to which they work. I name contingent presuppositions those ideas whose refusal only implies the breaking down of a contingent practice. I name ground presuppositions those ideas whose refusal would cause our

12

Chapter 2

whole way of life to break down. A functional notion of presupposition implies a direct relation between the generality of the grounded practice and the duration of the presupposition involved. If the practice is a contingent one, then its presupposition is a contingent fact: the smaller the practice, the less the presupposition will work as a presupposition. If the practice is a general one, then its presupposition is a stable and long-lasting idea: the larger the practice, the longer the presupposition will work as a presupposition. Therefore, contingent and ground presuppositions identify two poles. In between these two poles, there is a hierarchy of presuppositions. One can pass from a piece of gossip introduced as presupposition in the conversation (namely, the fact that a friend of ours stopped smoking), to the felicity condition presupposed by the act to sell something (namely, the fact that one must O\Vll what one sells), to the condition presupposed by our membership of a social or linguistic community (for instance, the notion of private property or the laws of syntax), to the horizon of consistency conditions presupposed by our belonging to the same fmm of life (for instance, the idea tbat stones do not feel pain and do not enjoy free will). Ground presuppositions constitute tbis horizon: they identify tbe upper limit of a hierarchy of presuppositions. This is the reason why Prandi (2017) names them ultimate presuppositions.

2.3 What are ground presuppositions? Ground presuppositions are what Moore (1925) calls commonplaces; what Carnap (1959) refers to as universal words; what Wittgenstein (1969) calls

certitudes and pictures as the river bed of our form of life; what Searle (1983) describes as Background and opposes to the Net, or what Collingwood (1998) explicitly calls absolute presuppositions. Ground presuppositions are the basic ontological distinctions between humans, animals, vegetables, things, places, times and so on, organised in a net. They can be described by general propositions such as:

Humans have a body, feel pain, feel indignation ..

A standpoint for presuppositions

13

Animals have a body, feel pain, but do notfeel indignation . . Vegetables do not have a body, do notfeel pain, but have diseases . . Ground presuppositions manifest themselves directly in superordinate or general nouns (Rosch 1976, Mahlberg 2005) and, indirectly, in selection restrictions (prandi 2004, 2017). Ground presuppositions are possibility conditions of experience. If, during a walk in the woods, a branch hits me, one can appropriately say that something has happened to me. But if someone slaps my face, it is misleading to say, simply, that something happened to me; in fact, I was a victim of aggression. If we cut the bark of a pine tree, one can appropriately say that we see resin coming out, but if we slap someone, it is misleading to say, simply, that we see him giving out ocular secretions; in fact, we see him crying. If I see someone pinching my wallet from my bag, what I see is not the fact that he has free will, but the fact that he is stealing my wallet. Free will is the condition to which I can see that I

am

being subjected to a theft and not simply losing my wallet. If I see someone crying, what I see is not the fact that he is a human being, but the fact that he is sad. The idea of being human is the condition to which I can interpret his tears as an expression of sadness and not simply ocular secretions. In this sense, ground presuppositions are not facts we fmm experiences from, but rather possibility conditions to which we can experience facts. We do not make experience of a thing such as free will, but we experience robberies, murders, petty and noble wishes, good or evil actions, loyal or deceitful people and so on, which presuppose it. We do not experience a thing such as the soul, but we experience expressions of joy or sorrow, courage or cowardice and so on, which presuppose it. Ground presuppositions are possibility conditions of knowledge. If we see a friend of ours crying, we can ask: (4)

Why is she crying? Is she crying because she has been fired or because her boyfriend left her?

These utterances concern things about which we can coherently ask questions, be wrong, doubt and look for evidence; they concern facts, that

14

Chapter 2

is, objects of possible knowledge. The condition to which we can coherently make those utterances is that our friend is indeed a human being. For this very reason, however, this idea is not an object of knowledge. In fact, facing a friend crying, imagine stating: (Sa) 71 think that she is crying because she is a human being. (Sb) 7Why is she a human being? Unless we suspend our common sense, these utterances sound odd. We can doubt whether a friend of ours is happy and try to find out why, because we presuppose that she is a human being. But, in everyday life, it would be odd both to doubt whether she is a human being or a thing, and to look for evidence of it. Of course, if forced by some philosophical question, we may say: She is a human being because she can suffer and be

cruel. However, the contrary is true: we see someone suffering or acting deceitfully because we presuppose that she is a human being with free will. Analogously, we can say: The sky is not a human being because it can

neither cry nor suffer. However, once again, the contrary is true: the sky carmot cry because we presuppose that it is not a human being. If we were to presuppose this, we would strain to interpret meteorological changes as expressions of feelings and thoughts. But we do not.

2.4 Consequences for the treatment of presuppositions Kuroda (1989), Ducrot (1972), Thomason (1972), Cooper (1974), Chierchia and McConnell-Ginet (2000) and Kleiber (2012), among others, have remarked on the difference between some general and contingent presuppositions. To my knowledge, apart from Collingwood (1998) and Prandi (2004) who explicitly put forward the idea developed here, the author that presents it in the clearest way is Garcia-Murga (1998). This author distinguishes linguistic presuppositions

from general ones.

However, this distinction is usually made only to drop the general (ground) presuppositions and to focus on linguistic (contingent) ones. The alleged reason (Ducrot 1980) is that the former have no link to the structure of the

A standpoint for presuppositions

15

sentence, while the latter are tied to peculiar linguistic triggersl. The contribution of this book does not lie in acknowledging the existence of general presuppositions, but rather in considering them as prototypical ones. This overturning has two consequences for the treatment of presuppositions. Firstly, the ordinary treatment turns out to be somehow misleading. On the one hand, the contents of contingent presuppositions are not immediately consistency conditions, but simply contingent facts that we assert, question, justify or argue for. Sometimes, thanks to some kind of trigger, these facts can be made to work as consistency conditions of an utterance in a specific text or discourse. By their nature, then, these contents fall into the scope of notions like inference, accommodation or updating. On the other hand, the contents of ground presuppositions are consistency conditions of very general practices: in everyday life, they are always already presupposed. As a consequence, they do not need any trigger and, basically, they are never communicated, asserted, questioned, justified and so on. By their nature, these other contents are placed beyond the scope of notions such as inference, accommodation or updating of a set of infonnation. Now, holding that the prototypical presuppositions are the ground ones amounts to holding that contingent presuppositions are presuppositions only insofar as they behave like the latter. On this premise, the most well-knO\vn notions used to describe contingent presuppositions (namely, inference, accommodation, updating, etc.) do not really apply to them as presuppositions, but as contingent facts. This consequence will be developed in the second part of this book.

1 As is knO\vn, these triggers form a very heterogeneous group. Levinson (1983 : 1 8 8 193), for instance, enumerates thirteen types and he affirms that Karttunen arrived at thirty-one. Just to recall the most well-knmvn examples, let me mention existential presuppositions carried by noun phrases in a referential position (RusseIl 1 905, Strawson 1950); verbal presuppositions: verbs of judgement (Fillmore 1973), factive verbs (Kiparsky and Kiparsky 1973), implicative and aspectual verbs (Karttunen, 1973a), and informational manipulations (Sperber and Wilson 1979).

16

Chapter 2 Secondly, if ground presuppositions are the presuppositions par

excellence, then their ordinary treatment reveals just the top of the iceberg of the whole phenomenon. Studying presuppositions does not primarily mean to describe a set of triggers, but to discover our natural ontology in the track of Strawson's programme of descriptive metaphysics (Strawson 1959 and 1992). This is the submerged part of the phenomenon of presuppositions. This consequence will be developed in the third part of this book. Both the aforementioned consequences, however, rely on the premise that ground presuppositions actually are prototypical presuppositions. The first part of this book is devoted to defending this idea.

CHAPTER 3 THE INDETERMINACY OF PRESUPPOSITIONS

In this chapter, I defend the distinction between contingent and ground presuppositions as a solution to an epistemological paradox. In the classic debate, the notion of presupposition is always construed as something else: a kind of illocution (Ducrot), a kind of propositional attitude (Stalnaker), a kind of implication and so on. These other notions, however, were originally designed to apply to contents that are not presuppositions. Hence, presupposition turns out to be affected by an intrinsic indeterminacy.

Distinguishing

between

contingent

and

ground

presuppositions dissolves this indetenninacy.

3.1 Presupposition as an illocutionary act According to Ducrot (1972:77), both a presupposition and an illocutive act are legal acts (actes juridiques, Fr.), that is, acts that modify social roles among people. If we are at a bus stop, smoking in silence, we are doing nothing wrong, but if a friend walks towards us, says Hello, and we keep smoking in silence, our silence becomes a rude act. This means that our friend, by saying Hello, has assigned us the duty of addressing him back and opens the possibility for us to be rude by not doing so. Similarly, when a speaker asks a question, they assign the receiver the duty of answering (at worst, to answer that they do not know) and open up the possibility of being rude by refusing to answer. When a speaker gives an order, they assign to the receiver the duty to obey and open the possibility of disobedience. And so on. Now, if presupposing is an illocutionary act and if an illocutionary act IS a legal act, what obligations does it impose and on whom? Ducrot (1980:1097) answers that presupposing a content means to impose its

18

Chapter 3

reception on the addressee as the condition for further dialogue. Note that this idea brilliantly accounts for the rhetorical or communicative exploitation of presuppositions, but-paradoxically-it neglects

its

simplest

functioning.

In its

simplest functioning, a

presupposition is already accepted by all the participants to the communicative exchange. Hence, defining it as "to impose to accept certain information... " is pointless. Indeed, such a definition makes sense only if the presupposition is not already presupposed. A similar objection can be raised against all the conceptions of presupposition focusing on its rhetorical exploitation. For instance, construing a presupposition as a kind of inference, or the updating of a common ground of knowledge, can make sense if the addressee does not share the presupposition yet. But if the addressee already accepts the presupposition, there is simply no inference to draw and no update to make. Indeed, all the rhetorical or infOlmative uses of presuppositions are precisely based on the exploitation of something that, normally, does not need to be inferred or updated in a common ground of knowledge. The same objection, moreover, affects the very notion of "trigger". Consider the utterance George stopped smoking. Is the predicate to stop

doing something a presuppositional trigger? If the idea that George smoked needs to be triggered, then it was not a shared presupposition. If the idea that George smoked is already presupposed, then there is no sense in saying that to stop smoking triggers it. So, the very notion of "presuppositional trigger" turns out to be self-contradictory. Be that

as

it may, let us come back to Ducrot's illocutionary

conception of presupposition. As Strawson (1964) points out, in order to work, an illocutionary force must be disclosed, that is, recognised. Now, as Ducrot (1972:76) himself acknowledges, presupposition is an undisclosed illocutionary act. But then, how can it put an obligation on the addressee? Indeed, inducing someone to accept something

as

a condition

for further dialogue does not look like an illocutionary act, but rather a perlocutionary one, like forcing someone. Let me develop this point briefly.

The indeterminacy of presuppositions

19

An illocutionary act is not a genuine communicative goal. Consider the following questions: (1)

How can I .. construct a question / make an assertion / make a promise .. in English ?

This is not the kind of question that an English speaker would ask himself during a conversation. The reason is clear: his grammar a priori codes explicit meaning in order to reach these goals. A perlocutionary act, on the contrary, is a genuine communicative goal. Consider (2): (2) How can I.. convince / reassure /frighten .. my listener? These questions are perfectly meaningful. This time, English grammar does not code specialised tools in order to reach these goals, but one has to use, in a creative and unpredictable way, assertions, promises and so on. On this premise, let us consider the following question:

How can I induce my listener to take this information for granted? It

seems to me that this question is like (2) and not like (1). This

question identifies a genuine discursive goal, something that one must be able to achieve by making use of illocutionary acts: that is to say, a perlocutionary act. Indeed, by uttering an assertion, the speaker makes the receiver swallow a pill; once the pill is ingested, once the receiver has accepted that assertion, they can can no more refuse its presuppositions. But, of course, the speaker must succeed in making the receiver swallow that presupposition. If inducing someone to presuppose is a kind perlocutionary act, rather than an illocutionary one, we find an immediate explanation of why the verb to presuppose does not have an explicit performative form without the need to posit a parasitic illocutionary act constant through all different illocutionary acts. Moreover, we understand that, if one looks at grammar trying to list all the structures devoted to manipulating people by making them swallow some information, one reaches Soames's conclusion: One of the most striking lessons of recent work is that there are many kinds

20

Chapter 3 and SOillces of presuppositions; so many that there may be no single theory capable of incorporating them all. (Soames, 1989:602) This is the same conclusion that one would reach after having looked

at grammar and tried to list all the structures devoted to convince, frighten or reassure people. Let us have a closer look at Ducrot's insight. Consider (3): (3a) I stopped smoking/ / I promise I will stop smoking / Did he stop

smoking? (3b) I1he smoked According to Ducrot, by stating (3a), the speaker performs two layers of actions: (a) a first layer of different actions consisting of ordering, promising,

asserting and asking (in the same way as, in greeting someone by saying Hello /, he shows that he recognises the addressee); (b) a second constant, meta-communicative action, consisting of obliging tbe receiver to take (4b) for granted (in the same way as, by saying Hello /, he obliges the addressee to answer). The layer (b) identifies presupposition. My point is that these layers cannot be consistently qualified as actions at the same time. Of course, in the abstract, ordering, promising, asserting or asking are all actions like obliging. However, in the linguistic exchange at stake, ordering, promising and so on in (a) count as discursively relevant actions or argumentative moves, while obliging in (b) does not. This difference is made implicitly, but clearly, by Ducrot himself, who observes the existence of a linking or concatenation rule (loi

d 'enchainement, Fr.). According to this rule, all argumentative connections take place at the level of (a) and not at the level of (b). That is to say, when actions (a) take part in the argumentative chain of discourse, 'action' (b) is banished below tbe threshold of discursive relevance. Consider, for instance, the following examples:

The indeterminacy of presuppositions

21

(4a) Paul drank a whole bottle ofBordeaux. Hence. he felt a little tipsy. (4b) ?Mary discovered that Paul drank a whole bottle of Bordeaux.

Hence. hefelt a little tipsy. In (4a), the consequence perfectly leads on from Paul drank a bottle of

Bordeaux. In (4b), on the contrary, the same consequence has trouble leading on from that content because.. this latter is presupposed. So, when a content is overtly asserted-that is, posited-it can be used to make argumentative actions (such as drawing consequences), but when that very content is presupposed-that is, covertly asserted-it is no longer disposable for the previous argumentative actions. Now, it seems to me that this does not prove that we are confronted with another kind of argumentative action or illocutive force; rather, it is the best proof that one carmot consistently talk about "action" or "illocutive force" for both cases at the same time. FurthemlOre, note that in a dialogue or a game, an action or a move is something that one can choose to do or not. Now, we can choose to greet or not, to state (3a) (I stopped smoking. .. ) or not and to move a piece in a chess game or not. But, once we have greeted, stated or moved, we carmot choose not to oblige someone to answer back, not to make someone accept (3b) (I smoked... ) and not to make someone move. Once again, this suggests that the fmmer are genuine actions or moves, but the latter are not.

3.2 Presupposition as a propositional attitude According to Stalnaker (1973), presupposing is a disposition, an attitude that a person has towards a fact. As this author \¥fites: [ . . . ] whatever the details of the definition, it is clear that presupposition is a propositional attitude. More specifically, it is an attitude of accepting something to be true. (Stalnaker. 1973 :448 450) The previous quote sketches a picture that, mutatis mutandis, reproduces Ducrot's scheme.

22

Chapter 3 Consider (5): (Sa) I hope Ibelieve I doubt I am sure .. that he stopped smoking/ (5b) He smoked According to Stalnaker, by stating (Sa), the speaker has two layers of

attitudes: (a) a first layer of variable propositional attitudes (believing I doubting

I being sure) whose propositional content is he stopped smoking; (b)

a

second constant propositional attitude consisting of. . presupposing, taking for granted, accepting, not discussing, havingfaith in . . the truth of (5b). .

Once again, my point is that (a) identifies genuine propositional attitudes, but (b) does not. Note that standard propositional attitudes like (a) are intentional attitudes and are expressed by everyday words, but there is no clear word for (b), just as there was no explicit perfOlmative fmm for Ducrot's alleged presuppositional illocutive force. This is not by chance. One can believe, fear or hope that George stopped smoking and-in doing so-certainly take for granted that George smoked. This latter, however, is not an attitude that can be grasped independently from the former because believing, doubting, hoping, etc. that someone stopped smoking actually

are taking for granted that he smoked. So, beliefJear, hope, etc. identify positive propositional attitudes, but presupposing that, taking for granted

that, accepting that, etc. do not. A consequence is that, if one tries to analyse presupposing as a propositional attitude, one carmot avoid analysing it in in telTIlS of some other, genuine, propositional attitude: namely, belief. This explains a paradox intrinsic to Stalnaker's approach. On the one hand, he explicitly states that presupposing is neither having a certain belief, nor simply behaving as if one had a certain belief, but rather being ready to behave as if one had that belief (cf. Stalnaker 1999:52). Along this path, however, the alleged 'presuppositional' attitude becomes more and more rarefied. Hence, on the other hand, Stalnaker grounds the whole

The indeterminacy of presuppositions

23

definition of presupposing precisely on beliefs. This is confirmed by his very notion of updating a presuppositional set or common ground (cf. also Heim 1991 and Veltman 1996), whose whole point is precisely to fonnalise the way in which beliefs are injected into the discourse when they

are

presuppositions.

The

consequence,

however,

IS

that

presupposition is not a positive notion, but rather a way in which a piece of infonnation is introduced into the discourse. Let us have a closer look at a particularly refined suggestion about how to formalise the updating of a common ground of beliefs through a presupposed proposition. Jayez (2004:87-98) defines assertion and presupposition as follows: Asserting P means to intend that the other participants in the communicative exchange update their belief states with P. Presupposing P means to intend that the other participants in the communicative exchange update their belief states with the proposition that the speaker already believed P. Leaving aside the criticism against the very idea of presupposition-as­ updating (cf. §3.1), if we consider something less contingent than the fact that someone smokes, a tension between the role of presupposition and the notion of belief is highlighted. Consider the following utterance: (6) Cecile is sad According to the logic behind the aforementioned definitions, a speaker asserting (6) should intend that his interlocutor updates his set of beliefs with the proposition: Cecile is sad and, before asserting this, the

speaker already believed that Cecile is a human being (and that human beings have feelings) . Now, maybe, this may work from a purely technical point of view, but, from the point of view of common sense, it is odd. As Wittgenstein points out: Does a child believe that milk exists? Or does it know that milk exists? Does a cat know that a mouse exists? (Wittgenstein 1 969:478)

24

Chapter 3 In my opinion, the moral to draw is the following. Presupposing means

taking something for granted. This does not imply that this something carmot be seen as a belief, but it implies that it is not relevant as such when it is taken for granted by some other belief. It is obviously possible to consider presupposed content to be like a secondary belief; however, doing this means disregarding its working as a presupposition. As a matter of fact, the more that content is a contingent fact, the easier it is to do so. Indeed, Stalnaker and Ducrot share the same assumption: namely, that one can only presuppose contingent facts. A contingent fact is expressed by a proposition. A proposition is the intentional content of propositional attitudes like belief or illocutionary forces like assertion. Therefore, studying presuppositions without ultimately relying upon those notions turns out to be practically impossible.

3.3 The indeterminacy of presuppositions Ducrot's and Stalnaker's positions share the same pattern. They put forward a notion-illocutionary act or propositional attitude-and apply it at two levels, (a) and (b). The second level identifies presuppositions. The difference between (a) and (b), however, is not a matter of nature: if we elicit (b) from its relation to (a), we are left with two propositional attitudes or two illocutionary acts. So, in these approaches, presupposition is nothing in itself, but is reduced to something else: a secondary illocutionary act or a secondary propositional attitude. The same pattern is shared by the approaches constructing presuppositions as background

inferences opposed to foreground ones (cf. Thomason 1990), or non-main­ point implications opposed to main point ones (cf. Simons 2007). Now, from a discursive point of view, levels (a) and (b) turn out to be mutually incompatible. This is shown by an observation usually made in order to undermine the very distinction between them (cf. Wilson 1975). Consider an utterance like (7): (7) Her husband is approaching.

The indeterminacy of presuppositions

25

Spontaneously, (7) would be interpreted as foregrounding the information that someone is coming towards us on the background that someone is married. In this sense, the former is communicatively relevant, while the latter-the presupposition-is not. Now, this balance can be reversed. Consider the following dialogue: (8) - Is that woman married?

- Well, look, her husband is approaching us. In (8), it is the information that someone is married, namely (7)' s presupposition, that is foregrounded, while the idea that someone is approaching is kept in the background. This remark, however, is not an argument against the distinction between posited and presupposed content, that is, the separation between (a) and (b), but the best proof that these levels, from a discursive point of view, are mutually incompatible. 'When the fmmer is communicatively relevant, that is, the object of illocutionary acts, propositional attitudes and so on, the other is not, and vice versa. This reminds us of what Epicure writes about death: Death is nothing to us, seeing that, when we are, death

has not come, and, when death has come, we are not. From a discursive point of view, presupposition is nothing to the utterance presupposing it. If this is true, then describing presupposition in terms of illocutionary acts, propositional attitudes and so on-which are appropriate to what is discursively relevant-amounts to describing presupposition through notions that apply to it when it is not a presupposition. Hence, presupposition turns out to be affected by an intrinsic indeterminacy and the ideas of presupposition-as-illocution and presupposition-as-attitude turn out to be ultimately inconsistent. I do not want, by any means, to minimise the contribution of the aforementioned ideas to the understanding of presupposition. Rather, I want to point out that they are insightful precisely insofar as they are brilliant metaphors: that is, insofar as they rely on the fact that

26

Chapter 3

presupposition is not really an illocution and is not really a propositional attitude.

3.4 Dissolving indeterminacy The indetelTIlinacy of the notion of presupposition can be dissolved if one makes the hypothesis that there are contents working as presuppositions

necessarily, and others that work as presuppositions contingently. The fmmer (ground presuppositions) are never contingent facts, but only presuppositions. Since they are not facts, strictly speaking, they are not propositions either. Hence, they are not natural objects of propositional attitudes, illocutionary acts, beliefs or implications. This point is clearly made by Collingwood (1940). The latter (contingent presuppositions) are first and foremost contingent facts. As contingent facts, they actually are the objects of propositional attitudes, illocutionary acts, beliefs, implications and so on. When they happen to work as presuppositions with regard to some utterance or speech act, they simply keep these features besides their new function of presupposition. The indetenninacy stems from hijacking the features of contingent facts on the function of presupposition. This indetenninacy, however, disappears if one returns to contingent presuppositions what belongs to them as contingent facts and what belongs to them as presuppositions. Let me insist on a point. Of course, in principle, nothing prevents us from studying contingent presuppositions as illocutionary acts, propositional attitudes, beliefs or implications. Indeed, heuristically, this is very insightful. However, if we do so, presuppositions are like diminished versions of other notions (not discursively relevant, non-main-point,

secondary, backgrounded illocutionary acts, attitudes, beliefs, implications . . . ) and lose the name ofpresuppositions.

CHAPTER 4 PRESUPPOSITIONS BEYOND COOPERATION

In this chapter, I argue for the distinction between contingent and ground presuppositions as the consistent development of a widely accepted conception of presupposition: the neo-Gricean framework (cf. Simons 2003, 2004, 2007, 2010 and Sbisa 2007). In this framework, taking for granted some infOlmation for the coherence of further dialogue-that is, presupposing-is seen as a natural consequence of the cooperative principle (cf. Grice [1975] 1989, 1981). The main idea can be traced back to the link between presupposition and felicity conditions stressed by Austin (1962), Searle (1978) and Fillmore (1989). Once this link is made, it is just one step towards rooting it in the cooperative principle. Since questioning the felicity condition of a speech act amounts to considering the interlocutor unreasonable by breaking dO\vn the interaction, and since the interlocutor is assumed to be reasonable, one tends to accept a priori all the felicity conditions granting the consistency of that act. According to Sbisa (2007:54-55), this is the source of the phenomenon of presupposition itself. In my opinion, if consistently developed, the reasoning behind this conception goes beyond the limits of the very notion of cooperation and implies a hierarchy from contingent to ground presuppositions.

4.1 Presupposing vs. Presupposed content As seen in Chapter Three, Ducrot analyses presupposition as a specific act, defmed by a specific illocutionary force, different from asserting, questioning, ordering, promising and so on. From this perspective, when uttering Did he stop smoking?, a speaker actually perfOlTIlS two actions:

28

Chapter 4

asking something and presupposing, that is, setting a piece of infOlmation as the condition for further dialogue. As a matter of fact (as remarked in §3.2 discussing Stalnaker's approach), it is possible to analyse presupposing not as an action distinct from asking, promising, ordering, asserting and so on, but rather as these very actions. In order to show this point, let us distinguish between presupposing and presupposed content. Consider the idea that there is a king of France. How can one presuppose it? For instance, by asking if the king of France is bald. Here, presupposing is uttering a question. Suppose that a friend of ours loves the city of Florence. How can we presuppose it? For instance, by promising them that we will go to Florence together. Here, presupposing is promising. Suppose that I own a bike. What action could presuppose this? For instance, selling the bike to someone. Here, presupposing is selling. Now, the previous actions can be considered from two different points of view. If we look at them as asking, promising or selling, we focus on their goal. If we look at them as presupposing, we focus on their felicity conditions. Presupposing, then, is not an action separable from asking, promising, selling, asserting and so on, but it is these very actions looked at from the point of view of their consistency conditions. Indeed, every action is presupposing if looked at from the point of view of its felicity conditions.

4.2 Beyond felicity conditions I maintain that if one looks deeper and deeper into the consistency conditions of a practice, one can remark that the presupposed contents increase in generality, but the nature of presupposing remains the same. If this is true, by following up the very reasoning of the neo-Gricean account, one can identify presuppositions that are far beyond both felicity conditions and the scope of the cooperative principle. Let us consider an action such as selling a bike to someone (example borrowed from Prandi 2004:227-228). Imagine that you are bargaining

Presuppositions beyond cooperation

29

with me in order to buy my bike. We disagree about the price. Then, for some reason, you suspect that the bike that I am trying to sell you is not really mine (because, for instance, I have stolen it). At that very moment, our practice of bargaining for the price breaks dO\vn and a new one begins: a quarrel. The idea that I owned the bike was then a felicity condition of our bargaining. This is the core intuition of the neo-Gricean account. Looked at from the point of view of its goal, bargaining is trying to obtain the best price; looked at from the point of view of its felicity conditions,

bargaining is presupposing that I owned the bike. Now, imagine that, while bargaining with me, you suddenly address the bike, beg the bike to come with you, and try to seduce it with a love song. Clearly, our interaction breaks dO\vn again. So, after all, looked at from the point of view of its consistency conditions, bargaining is also presupposing that bikes are not possible interlocutors. For each scene, the structure of presupposing is exactly the same. In both cases our practice breaks do\Vll: as a quarrel as opposed to a bargain, and as a demented way of acting

as

opposed to a consistent one. Of course,

there is difference. This difference, however, does not concern the nature of presupposing, but rather the presupposed content: the ownership of an item vs. the difference between humans and inanimate beings. The fmmer is on the surface: it works as a consistency condition with respect to some commercial exchange. The latter is much deeper: it works as a consistency condition with respect to every action. Despite their difference

III

generality and depth, they both work as consistency conditions.

4.3 The limits of cooperation Let us examine the behaviour of notions such as belief, attitude, felicity condition and cooperation through the previous scenes. Someone observing us bargaining over the price of the bike could say that we both seem to believe that I own the bike, or that I have the attitude of someone O\Vlling it. But would one say that we have the attitude of believing that bikes are not human beings? Clearly not. So,

as

we rise in

30

Chapter 4

the hierarchy of consistency conditions, at some point, the ascription of propositional attitudes or beliefs stops being relevant. In other words, propositional attitudes and beliefs can be ascribed to someone only insofar as they apply to contents that are not consistency conditions of every act. Hence, the limits of the applicability of these notions identify a turning point. Below it, there are contingent presuppositions, which are the usual felicity conditions. Above it, there are ground presuppositions, which are beyond felicity conditions. This turning point also marks the limit of the notion of cooperation. If two people disagree on the price of an item, we can reasonably hold that they share the common goal of concluding the bargaining. In this sense, they do cooperate. But what about the common goal of accepting the distinction between inanimate beings and humans? Here, speaking about cooperation is out of place because we do not face a goal to reach together, but a precondition for having common goals.

4.4 A consistent development of a 'Gricean' approach If the line of reasoning sketched above is sound, then a conception of presupposition based on the cooperative principle is not wrong but partial: it fits for presuppositions below the aforementioned turning point, but not above it. Its consistent development, however, leads to overstepping that very conception and to acknowledging the existence of ground presuppositions. In other words, the same reasoning that recognises the function of presupposition to the felicity conditions of a speech act implies that there are some basic ideas working as presuppositions of every act. And if there are ideas working as consistency conditions for every action, it seems to me that they should be considered the consistency conditions par excellence.

CHAPTER 5 PROTOTYPICAL PRESUPPOSITIONS

In the first part of this book, my goal is to distinguish between ground and contingent presuppositions and to argue that the fmmer are the prototypical ones. In Chapter Three, I presented this idea as a way out of the problem of the indeterminacy of presuppositions. In Chapter Four, I presented it as the consistent development of a notion of presupposition grounded on the cooperative principle. In both cases, I focused on some classic and influential views on presuppositions. In the present chapter, I will instead focus on the core features of the phenomenon of presuppositions itself. My argument is very simple. If there is a crucial phenomenon defming presupposition, and if there are some propositions that exhibit it in the clearest possible way, then these propositions are the prototypical presuppositions.

5.1 Survival under negation A crucial feature of presuppositions is undoubtedly their "survival under negation". As a matter of fact, this denomination is somehow misleading because a presupposition not only carmot be negated, but it carmot be asserted either. Be this as it may, consider the examples (1): ( l a) George has stopped smoking/ (1 b) George has not stopped smoking / (2) George had smoked In uttering ( l a), intuitively, one affirms that George does not smoke (any more) and negates that George smokes (again), by relying on (2). In

32

Chapter 5

this sense, we can say that in uttering (1), (2) is neither asserted nor negated, but taken for granted or presupposed. This is the founding intuition of the whole phenomenon of presupposition and the point of contention in the debate. Note that (2) behaves in this way only relative to (1); in itself, that very idea can be perfectly asserted or negated. General basic distinctions, reflected in classic selection restrictions, show the same phenomenon: (3a) George is sad/ (3b) George is not sad/ (4) George is a human being. In uttering (3), (4) is taken for granted and, even more clearly than (2), neither is asserted nor negated. This fact has been stressed. 'What has not been stressed enough though is that this time no trigger is needed: to stop

doing something is a trigger, but to be sad is not. So, here, we face the classic presuppositional behaviour without the need of any trigger. Now, if we extract (4) and try to assert or negate it in isolation, we find an interesting fact: (4a) George is a human being/ (4b) George is not a human being/ In everyday discourse, utterances (4a) and (4b) would not be interpreted as asserting or negating that George belongs to the species of Homo sapiens. Rather, by taking this idea for granted, they would be interpreted on a meta-discursive level as follows: (Sa) George surrenders to temptation / has limited patience / has

empathy Ihas weaknesses, etc. (Sb) George never surrenders to temptation / has unlimited patience /

has no empathy / has no weaknesses, etc. Of course, if forced by some peculiar context (for instance, in a

Prototypical presuppositions

33

biology class), one can interpret (4a) and (4b) literally. However, the discursive data suggest that one nOlmally does not. So, an example like (4a) exhibits the core feature of presuppositions­ that is, being neither asserted nor negated, but taken for granted-not only as regards some other utterance like (3), but also in itself If this is true, then the intuition which leads to considering (2) a presupposition relative to (1), leads to considering (4a) a presupposition in itself: an absolute

presupposition (Collingwood 1998). A relative presupposition is neither asserted nor negated only by an utterance which presupposes it.

An

absolute presupposition is instead

neither asserted nor negated even by an utterance trying to express it. In fact, it cannot be expressed, but only presupposed (Fasciolo 2012b). As Collingwood (1998) puts it: [ . . . ] absolute presuppositions are never [ .. ] propounded. I do not mean that they sometimes go illlpropounded [ ... ] I mean that they are never propounded at all. To be propounded is not their business; their business is to be presupposed. (Collingwood 1998:32 33)'

5,2 The inaccessibility of presuppositions If one looks at the classic presuppositional tests (cf. Ducrot 1972, Chierchia and McConnell-Ginet 2000), one notices that they all stress that presuppositions are, in some way, discursively inaccessible. Now, absolute presuppositions

exhibit

such

a discursive

inaccessibility directly,

Indeed, contrast (a) and (b): (a) Today, Isaw / realised that .. Alice 's luisband cheated on her. (b) Today, I saw /realised that... Alice is a human being. Utterance (a) is neutral. Hence, the content of the subordinate clause is an object of perception or knowledge and someone asking Did Alice 's luisband cheat on her? exhibits a true desire to know. Utterance (b), on the other hand, strikes us as odd, or as describing a sort of epiphany. Hence, the content of an absolute presupposition is neither an object of perception nor of knowledge. Moreover, someone uttering (b) does not extinguish the curiosity of a scientist, but expresses a sort of wonder: the wonder of a saint facing the creation, or of a parent facing his child, or of anybody facing a miracle, that is what Wittgenstein (1922) refers to by using the word "Mystic".

2

34

Chapter 5

independently from any other utterance and without the need of any trigger. Relative presuppositions, on the contrary, exhibit it only indirectly. In the previous paragraph, I have illustrated this point regarding assertion and negation. Let me now present some consequences of it concerning: propositional attitudes (§S.2.1) argumentative links (§S.2.2) question and explication (§S.2.3). 5.2.1 Propositional attitndes

Consider (6): (6a) Ifear that my wife has discovered that I have cheated on her. If ! state (6a), the object of my fear is not the fact that I cheated on my wife. In this sense, (6a) 's presupposition-I cheated on my wife-is inaccessible to that propositional attitude. However, besides (6a), this very fact can be the natural content of a fear: My wife fears that I have cheated

on her. This is not true for something like a judge is a human being. Consider (7): (7a) Ifear that the judge is a human being If I state (7a), most likely, I do not fear that the judge belongs to our own species, but, for instance, that he can be corrupted or frightened by a gangster. So, A judge is a human being looks directly inaccessible as a content of fear, while I cheated on my wife can become discursively inaccessible only by means of a trigger like to discover and relative to (6). 5.2.2 Argumentative links

I have already mentioned Ducrot's concatenation rule in Chapter Three, §3.1. Consider (8a):

Prototypical presuppositions

35

(Sa) Paul stopped smoking. His health will improve. As stressed by Ducrot, from an argumentative point of view, the second utterance does not lead on from the presupposed content (Paul has

smoked) because this latter would contradict it: (Sb) ?Paul has smoked Then his health will improve. In (8a), the relevant premise of the second utterance is rather the posited content (Paul does not smoke) while the presupposed one is kept in the background. However, besides (Sa), the idea that Paul has smoked can of course plainly work as a premise of an argument: Paul has smoked

a lot. He ruined his lungs. Once again, this is not true for A judge is a human being. Consider (9a): (9a) The judge is a human being. He will accept our offer. In (9a), the second utterance does not argumentatively lead on from the idea that A judge belongs to the human species, but rather from some moral characteristic which presupposes it, for instance: (9b)

Thejudge fearsfor hisfamily. Then he will accept our offer.

So, A judge is a human being is directly inaccessible as an argumentative pivot while Paul has smoked can be put outside the flow of argumentation only by means of some trigger and relative to some utterance like (Sa). 5.2.3 Questioning and explication

Consider (10): ( lOa) ?Why did Alice discover that her husband cheated on her

because her husband had actually cheated on her. ( l Ob) ?Why is Alice sad? Because she is a human being. These sequences are both odd. The reason is that Alice 's husband cheated on her and Alice is a human being are not answers to the

36

Chapter 5

questions at stake, but consistency conditions of all their possible answers. However, the first consistency condition, in itself, can be relevantly questioned, while the latter carmot: (1 l a) Why didAlice 's husband cheat on her? ( l Ib) ? Why is Alice a human being? So, Alice is a human being is directly inaccessible to questioning, while Alice 's husband has cheated on her is not. As Collingwood (1998) puts it: [ . . . ] any question involving the presupposition that an absolute presupposition is a proposition, such as the questions "Is it that true?" "What evidence is there for it?" "How can it be demonstrated?" "What right have we to presuppose it if it can't?" is a nonsense question. (Collingwood 1998:33) Stating that the content of a ground presupposition cannot be questioned means that it cannot be justified either. If one tries to, the result can be: fOlmally circular, conceptually circular or not relevant. Let me briefly illustrate each case by considering some possible answers to (11 b). Consider answers like: ( 1 1 b') Alice is a human being because .. she is a human being / ir is so

/ God made her like that. In this kind of answer, the explanans actually repeats to or refers to the

explanandum. This is a fonnally circular explication. Consider answers like: ( 1 1 b") Alice is a human being because .. she has feelings / has

free will ! can be evil. Here, the explanans is not a cause of the explanandum, but rather its consequence: Alice has feelings, free will, etc. because she is a human being and not vice versa. This is the intuition pointed out in §2.3. So, those answers do not justify the idea that Alice is a human being, but rather the

Prototypical presuppositions

37

right to state it. In fact, they can be paraphrased as: I say [think, suppose,

etc.} that Alice is a human being because .. This is a conceptually circular explanation. Consider answers like: ( l l b ' ' ') Alice is a human being because .. her DNA is so and so. Here, we find Moore's "naturalistic fallacy", that is a fmm of reductionism.

From the

standpoint of everyday life,

a specific

configuration of the DNA is not the definition of a human being, just like a frequency of light is not the definition of a colour. This implies that such an answer oversteps the domain of common sense relevance. A variant of the previous answer is a statement like: ( l lb"") Human beings have feelings for such and such evolutionary

reason. Once again, this level of explanation would be out of place in our everyday lives. Indeed, even an ethologist truly convinced of (11 b '" ') will never look at a friend hugging them from that point of view. And if they do, they would be considered a sort of psychotic, that is, someone not sharing the mainstream of our fmm of life. 5,2,4 Prototypical presuppositions

The conclusion seems to me clear. On the one hand, there are facts that exhibit the core features of presuppositions, namely, being inaccessible to assertion, negation, propositional attitudes, argumentation, question and explication exclusively if they are embedded in some sort of trigger and relative to some other utterance or speech act. On the other hand, there are ideas that exhibit the previous core features of presuppositions openly. Now, if one had to ascribe the title of presupposition to one of them, it would be natural to choose the latter. Or, one can recognise the function of presuppositions to both of them, just pointing out that the former carry it out contingently (they are contingent presuppositions), while the latter

38

Chapter 5

carry it out systematically (they are ground presuppositions). I have chosen this latter solution.

5.3 Why are there contingent presuppositions? Since the content of a ground presupposition shows a presuppositional behaviour in itself, our access to it is immediate or a priori. No trigger is needed to make it a presupposition because it already is one. Since the content of a contingent presupposition does not show presuppositional behaviour in itself, but only in respect of another utterance, our access to it is mediated or a posteriori. It needs some trigger to make it a presupposition because, in itself, it is not one. A contingent presupposition, then, enjoys a double life. On the one hand, relative to the utterance for which it works

as

a presupposition, it

looks inaccessible. On the other hand, independent from this utterance, the content of a contingent presupposition is perfectly accessible. This is the source of the ambivalence of contingent presuppositions which fuels the debate between supporters and debunkers of presupposition. Note that, from this point of view, a puzzling fact from the literature on presuppositions can be solved. Surprisingly, nobody seems to be able to answer a very simple question: why are there presuppositional triggers? Karttunen and Peters (1979) write: At present we have no answer to questions like [ .. ] "Why are there words like even which mean something but which have no effect on truth conditions?" [ ... ] we concentrate on HOW this aspect of meaning can be described in an explicit way. (Karttunen and Peters 1979: 15) Actually, the answer is straightforward. There are presuppositional triggers in order to make presupposition contents, which, in se, are not presuppositions. It is precisely since they are not presuppositions that it may be useful to make them presuppositions with, for instance, the intention of making the receiver swallow them. This answer may seem like a platitude, but this platitude can be grasped only by distinguishing between contingent and ground presuppositions and by acknowledging

Prototypical presuppositions

39

that these latter are the prototypical ones. I concede that a linginst's interest can begin after grOlmd presuppositions. After all, a linguist is concerned with the ways and the aims of presupposing contents that are discursively relevant. However, I carmot concede to reserving the label presupposition to contingent presuppositions and to put the ground ones in some other section. The reason for this is that the former are presuppositions in as much as they resemble the latter. This is not purely a matter of labels. Linguists and philosophers who consider ground and contingent presuppositions to be separate phenomena expose themselves to the indetelTIlinacy problem. Because of the aforementioned double life, all of the discursively interesting features of contingent presuppositions stem from their differences as regards ground ones, that is, from the fact that, in themselves, they are not presuppositions. As a consequence, one is led to consider as essential to presuppositions some features that they have insofar as they are not presuppositions.

5.4 Why have scholars focused on contingent presuppositions? I see two main reasons. Firstly, presuppositions are detected by their violations. Therefore, the more easily a presupposition is frustrated, the easier it is seen. On the one hand, in the course of a dialogue, it is nOlTIlal to make our interlocutor realise that he seems to speak as if, for instance, he believes that someone is married when, in fact, it is not so. Contingent presuppositions fOlTIl specific discursive contexts, which are intrinsically dynamic. Everyday discourse spontaneously presents cases of frustration and modification of contingent presuppositions. On the other hand, in the course of a dialogue, it is not usual at all to make our interlocutor realise that they seem to speak as if they believe that trees have feelings. Ground presuppositions constitute the ground of every changing context. Everyday discourse does not offer instances of frustrated ground presuppositions. Ifwe want to find them, we must manipulate utterances to produce conceptual conflicts and

40

Chapter 5

metaphorical discourse3• Secondly, the contents of contingent presuppositions are nothing but the facts we live by and we are concerned with: the facts that we can justify, state, negate, question, infer, prove, disprove and so on. None of them, considered by itself, is a presupposition. Each of them, thanks to some linguistic trigger, can be contingently used

as

a presupposition as

regards some other utterance or speech act. As a consequence, linguistic triggers attract attention. The contents of ground presuppositions, on the contrary, are not the facts we live by and we typically want to communicate, demonstrate or contradict. Moreover, there are no triggers to make them presuppositions. Therefore, they simply stay on the ground, unnoticed.

3 To my knowledge, the linguist that highlighted the epistemological privilege of the conceptual conflict in the clearest way is Prandi (1987, 2004, 2017)

CHAPTER 6 A HIERARCHY OF PRESUPPOSITIONS

Ground and contingent presuppositions identify two poles of a hierarchy. At the bottom of the hierarchy, contingent presuppositions work as consistency conditions of single speech acts. At the top of the hierarchy, ground presuppositions work as consistency conditions of all practices and form the bedrock of our form of life (cf. Wittgenstein, 1969:97). In Chapter Five, I argued for this hierarchy on the basis of the very intuition which revealed presuppositions. In this chapter, I try to give some substantial content to it. I propose to identify at least three levels: (a) discursive presuppositions (b) epistemic (encyclopaedic, sociaL) presuppositions (c) ground presuppositions. I am mainly interested in separating the two poles (a) and (c). Hence, the observations concerning (b) will be limited to highlighting that dichotomy. No doubt, layer (b) could be analysed more fully. In order to distinguish (a), (b) and (c), I will offer three substantive criteria concerning: (i) the functioning of predicates like to ignore, to discover, to teach, etc. (§6.1) (ii) the limits of felicity conditions (§6.2) (iii) the distinction of three kinds of tautologies (§6.3).

42

Chapter 6

6.1 First criterion: to forget, to teach, to learn . . . The content of a contingent presupposition can be consistently ignored,

discovered, known, believed,jorgotten, remembered, reminded, but neither *taught nor *leamed. The content of an epistemic presupposition can be consistently ignored, discovered, taught, learned, but neither *forgotten nor *remembered. The content of a ground presupposition carmot be consistently *ignored, *discovered, *taught, *leamed, *forgotten or

*remembered. 6.1.1 Discursive presuppositious

Let us begin by considering the trite fact that someone smoked. One can construe consistent examples where this content is the direct object of verbs like to ignore, to discover, to know, to believe, to forget, to

remember or to remind. For instance: ( l a) Who discovered that George smoked? (1 b) Anna ignored that George smoked ( l e) Anna forgot that George smoked Note that, with verbs like to teach, the result is unnatural: (Id) ?He taught Anna that George smoked

?Ha insegnato adAnna che Giorgiofomava. (It) ( l e) (?)Anna learned that George smoked'

?Anna ha imparato che Giorgio fumava. (It) Examples (Id) and ( l e) sound odd. The idea that someone smoked, then, is not the kind of thing that we normally teach (and learn): it is a piece of gossip or Doxa, but not a piece of knowledge or Episteme. This is In English, the verb to learn has two different meanings Gust like the verb apprendre in French), roughly: to know by studying and to become aware of something. Here, the relevant meaning is the former. In Italian these meanings are clearly separated: imparare vs. apprendere. 4

A hierarchy of presuppositions

43

the typical content of a discursive presupposition. 6.1.2 Epistemic presuppositions

Let us now focus on an idea such as The eanh is round. Here, we have content that can be consistently discovered or ignored, and also taught or

learned. The following utterances are perfectly consistent: (2a) Once people ignored / did not know that the earth was round' (2b) Who first discovered that the earth is round? (2c) 1 taught my children that the earth is round And they learned it. However, that the earth is round is not the kind of thing that we normallyforget, remember or remind each other of. In everyday discourse, examples like the following ones would be perceived as odd: (2d) ?My sonforgot that the earth is round vs. My son forgot that his uncle hates cheese. (2e) ?My son remembered that the earth is round vs. My son remembered that his uncle hates cheese. (2f) ?1 remindedmy son that the earth is round vs. I reminded my son that his uncle hates cheese. These intuitions suggest that The earth is round is not simply a piece of Doxa, but it counts as encyclopaedic knowledge, which is a portion of the Episteme. This piece of knowledge has two sides. On the one hand (by teaching and learning it), it can practically enter into our shared background; in this sense, its acquisition is de facto contingent. On the other hand, when it becomes a part of our encyclopaedic background, it acquires a sort of necessity de jure.

5 In English, to ignore is usually to choose not to know. Here, it is used as simple synonym of not to know.

44

Chapter 6 So, knowledge like The earth is round is not relied upon as subjective

mental content (a memory), but rather as an inter-subjective epistemic one. This is the kind of content of an epistemic presupposition6• 6.1.3 Grouud presuppositious

Let us turn to

an

idea such

as:

Human beings have feelings. Consider the

following examples: (3a) ?Once people ignored / did not know that human beings have

feelings. (3b) ?Whofirst discovered that human beings have feelings? (3c) ?In school, it should be taught that human beings havefeelings. (3d) ?My children learned that human beings have feelings. (3e) ?But then theyforgot it. (3f) ?Some years later, however, they remembered it Here, the data are clear. Either examples (3) are ridiculous or they would be interpreted non-literally by triggering the presuppositional phenomenon stressed in Chapter Four, §4.1. For instance, (3c) could be

In school, it should be taught.. kindness, compassion, respect for people, which presupposes precisely that humans are sentient interpreted

as:

beings with feelings.

6 Of course, when it is taught or learned for the first time (when it has not become a part of our backgrOlUld yet), even knowledge like The earth is round can be forgotten and remembered. Think about a very naive student having an oral exam, for instance. Hence, from this point of view, an episternic presupposition looks like a discursive one. However, this fact must not mislead: an epistemic presupposition has a different nature compared to a discillsive one. On the one hand, after it has become a part of Oill shared backgrmmd knowledge, it is odd to remember or forget the content of an epistemic presupposition. On the other hand, the content of a discursive presupposition never becomes public background knowledge (that is, it is not something that we learn by teaching). This ambivalence of an epistemic presupposition is perfectly consistent with its intermediate position between discillsive and grmmd presuppositions.

A hierarchy of presuppositions

45

So, our reactions to (3) suggest Human beings have feelings is a much more stable idea tban the most stable encyclopaedic knowledge. In fact, it is its very ground. This is a ground presupposition.

6.2 Second criterion: the limits of felicity conditions A contingent fact working as a presupposition assumes the fmm of a felicity condition of a speech act: this is a discursive presupposition. A public or social fact working as a presupposition can assume the form of a felicity condition of a speech act, but it is more than that: this is an epistemic or social presupposition. An idea concerning our natural ontology cannot even assume the fmm of a felicity condition: this is a ground presupposition. 6.2.1 Discursive presuppositions

Consider ( l a): ( l a)

?Since George never smoked in his life, I ask you: Did he stop (smoking) ?

Utterance ( l a) sounds odd because the subordinate clause contradicts a felicity condition of the question, namely tbe fact tbat George smoked. This contradiction stems from the clash between the idea expressed by the subordinate clause and the one triggered by to stop. In this sense, the contradiction has a discursive source. Hence, we can say tbat in (1 b): (1 b) Since George smoked all his life, I ask you: Did he stop (smoking) ? the felicity condition of the question has a discursive source too. Here, we are confronted with a discursive presupposition. 6.2.2 Epistemic presuppositions

Consider (2a): (2a)

?Since France has a king, I ask you: Is he bald?

46

Chapter 6 Once again, utterance (2a) would sound odd: a felicity condition of the

question is then contradicted. However, this time, the subordinate clause does not clash with a presupposition triggered in the principal clause (namely, the referential pronoun he), but simply with our shared knowledge: we know that France is not a monarchy, but a republic. Let us contrast (2) with (1). In (1), the contradiction takes place horizontally: in order to perceive it, we have to parse the whole example. In (2), by contrast, the contradiction takes place vertically: in order to detect it, we do not have to parse the whole example, but we stop at the subordinate clause. This is a sign that we evaluate it as regards some extra­ linguistic content. The source of the contradiction, then, is not discursive, but epistemic. If this is true, then, in (2b): (2b) Since France has a president I askyou: Is he bald? the felicity condition of the question has an extra-discursive source too. Here, we are confronted with a piece of encyclopaedic background that happens to offer a felicity condition to a speech act, but it cannot be reduced to that7• 6.2.3 Ground presnppositions

Let us now turn to (3): (3) ?Since Alice is a human being, I ask you: Is she sad? Note that, even if the subordinate clause is sound, example (3) is immediately odd. Let us contrast (3) with (2). In (2), the oddity stemmed from the fact Indeed, you are immediately disposed to accommodate an utterance like: Did George stop smoking?-even if you do not actually know who I am talking about but not Is the present king ofFrance bald? which resists accommodation.

7

The reason is precisely the different nature of the content at stake: in one case, it is a matter of private knowledge, while in the other it is a matter of public knowledge. It is easier to accommodate the former than the latter.

A hierarchy of presuppositions

47

that the subordinate clause contradicted some shared knowledge. In (3), the oddity stems from the fact that the subordinate clause affirms a truth. This suggests that such a truth is so basic that it cannot be seen as a felicity condition of a specific act even if it provides a ground for its consistency. This is a ground presupposition.

6.3 Third criterion: tautologies A presupposition works

as

a tautology. For example, if we both know that

a friend of ours is a smoker, I can ask you if he stopped. The presupposition triggered by to stop matches a presupposition already there, in some sort of background. Now, presuppositions diverge

as

regards the

'background' that gives rise to such tautologies. A private fact can work

as

a tautology

as

regards a discursive

background: this is a discursive presupposition. A public fact works

as

a

tautology as regards stable shared knowledge: this is an epistemic or social presupposition.

An

idea concerning our natural ontology works

as

a

tautology as regards our natural ontology: this is a ground presupposition. 6.3.1 Discursive presuppositions

Consider ( l a) and (lb): ( l a) Alice regretted that her husband had cheated on her. (lb) Alice did not regret that her husband had cheated on her. Spontaneously, ( l a) would be interpreted as consistent and (lb) would immediately invite (cf. Geis and Zwicky 1971) the so-called "internal reading" of the negation. In this reading, (1 b) would be interpreted Alice was insensitive to her husband's betrayal. In order to perceive ( l a)

as

as:

inconsistent and to block the invited

inference in (lb), we must specify a particular discursive context:

?Alice 's husband never cheated on her and she regretted that he had.

48

Chapter 6

Alice did not regret that her husband had cheated on her because her husband never cheated on her. If this is true, then, in (Ia)'s and (Ib)'s spontaneous readings, de facto, one 'projects' or infers a discursive context according to which Alice 's

husband cheated. In this sense, in ( l a) and (Ib), the presupposition relies on a discursive background and hence it works as discursive tautology. Here, an account of presupposition in terms of satisfaction of a "discursive context" (cf. Karttunen 1973b, 1974) makes sense. These are discursive presuppositions. 6.3.2 Epistemic presuppositions

Consider (2a) and (2b): (2a) ?Alice regretted that Hillary Clinton was elected president of the us.

(2b) Alice did not regret that Hillary Clinton was electedpresident. Note that, in order to perceive (2a) as inconsistent and to interpret the negation in (2b) "externally", we do not need to specify a particular discursive context by linking two utterances like:

Hillary Clinton was electedpresident oOhe US and Alice regretted it. Alice did not regret that Hillary Clinton was elected president because Hillary Clinton was not elected Indeed, (2a)'s inconsistency and (2b)'s external reading are immediate. Why is this so? Because, here, some shared public knowledge is at stake. Let us contrast example (2) with (1). On the one hand, the presupposition at stake in (1) belongs to private and contingent facts, without any impact on our shared background knowledge; hence, there is no hatm in accommodating it. On the other hand, the presupposition at stake in (2) belongs to public knowledge, and accommodating it would have a dramatic impact on our background knowledge.

A hierarchy of presuppositions

49

This fact implies a consequence concerning (2c) and (2d): (2c) Alice regretted that Donald Trump was elected president of the us.

(2d) Alice did not regret that Donald Trump was elected president of

the

us.

If, in order to perceive (2a) as inconsistent and to read the negation in (2b) as external, we do not have to specify a discursive context, then, when we interpret (2c) as consistent and we read the negation in (2d) as internal, it is pointless to say that we infer: Donald Trump was elected

president of the

us,

or that we project backward a context like: Donald

Trump was elected president ofthe US and Alice regretted it. The reason is that such a context is not supposed to be rooted in some discourse, but is supposed to already be there in our shared encyclopaedic knowledge. In this sense, the presupposition at stake in (2) (namely, that Donald Trump was elected) does not simply rely on a discursive background, but rather on an epistemic one. Hence, this presupposition works as an epistemic tautology. Here, an account of presupposition in telTIlS of satisfaction of a "discursive context" (cf. Karttunen 1973b, 1974) is misleading. These are epistemic presuppositions. 6.3.3 Ground presuppositions

Consider (3a) and (3b): (3a) 7The moon smiles. (3b) The moon does not smile. (3c) Alice smiles. (3d) Alice does not smile. Let us immediately contrast (3) with (2). As regards examples (3), the same remarks made for examples (2) hold, and even more evidently. In order to perceive (3a) as inconsistent or to read the negation in (3b) as

50

Chapter 6

external, we certainly do not need either to specify a coordination like: The moon is not a human being and it smiles, or to add: The moon does not

smile because it is not a human being. Hence, it is clearly absurd both to hold that, facing (3d), one infers Alice is a hwnan being, and to hold that, in uttering (3c), one projects backward a discursive context like: Alice is a human being and she smiles. However, between (3) and (2) tbere is a striking difference. In itself, a coordination like:

Donald Trump was elected president of the he had been

us and Alice

regretted that

is discursively consistent. Instead, a coordination like:

?Alice is a human being and she smiles IS discursively pointless. After all, one can assert the result of a presidential election, but one is not supposed to assert the distinction between humans and non-humans. If this is true, this time, the presupposition relies on background much more basic than public knowledge: natural ontology. The ideas tbat Alice is a human being and human beings have feelings and smile, work, then, as ontological tautologies. Here, an account of presupposition in telTIlS of satisfaction of a "discursive context" (cf. Karttunen

1973b,

1974)

IS

totally

absurd.

These

are

ground

presuppositions.

6.4 Contingent vs. Ground presuppositions The previous remarks stress an intelTIlediate status of epistemic presuppositions as regards discursive and ground presuppositions. Focusing on epistemic presupposition, then, opens out the distinction between contingent and ground presuppositions. On the one hand, like ground and unlike discursive presuppositions, epistemic presuppositions are not immediately accommodated, they are

A hierarchy of presuppositions

51

not triggered by invited inferences and the utterances violating them are either immediately felt to be inconsistent (if positive), or immediately undergo an external reading (if negative). On the other hand, like discursive and unlike ground presuppositions, epistemic presuppositions do not manifest themselves in selection restrictions and are hosted by the same triggers as discursive ones. Moreover, in some circumstances (typically, when they have not already entered our background), they can be immediately accommodated. This is because, in the end, both discursive and social presuppositions are knowledge. In this sense, they both oppose themselves to ground presuppositions: the fmmer are more or less contingent, whereas the latter are not contingent at all. Let us come back to the following propositions: (1) Paul smoked (possible discursive presupposition) (2) The earth is round (epistemic presupposition) (3) Human beings have feelings. (ground presupposition) When they work as presuppositions, these propositions are kinds of tautologies. A tautology is a necessary and a priori truth. A discursive presupposition like (1) holds as a necessary and a priori truth only in respect of a certain discourse. In order to see it as true or false a posteriori, that is contingent, it is sufficient to imagine another discursive context. Without a specific discursive context or trigger, one would neither think of (1) as an a priori truth, nor as a presupposition. A social or epistemic presupposition like (2) holds as a necessary and a priori truth in respect of socially shared background knowledge, more stable than discourse. In order to see it as true or false a posteriori, we must imagine another course of history or nature. A ground presupposition like (3) holds as a necessary and a priori truth as regards ontology. This time, we cannot really see it as true or false a

52

Chapter 6

posteriori because we carmot really have access to another ontology8. This is the reason why (3) looks like an a priori truth innnediately. Now, this book began by noting tbat, in the very debate about presuppositions, it is pointless to make an inventory of presuppositions. If one focuses on propositions like (1), tbe reason is clear: (1) has notbing in itself that makes us think of it as an a priori truth. On the contrary, since propositions like (3) are immediately recognisable as a priori truths, it is perfectly reasonable to look for them in order to list them. Note, moreover, that the previous tautologies are not formal, but substantial; that is to say, they are not a priori true in virtue of their fmm, but of their content. Two examples of fOlTIlal tautologies, instead, are:

A bachelor is an unmarried man. not-(A and not-A}. The former is a lexical formal tautology, while tbe latter is a formal logical tautology. [f propositions (1) to (3) are substantial tautologies, tben tbe study of presuppositions carmot be fOlTIlal, and it involves the analysis of content. The illusion tbat tbe study of presuppositions could be carried on by disregarding the content involved has two interconnected causes: the prejudice tbat there is only one kind of presupposition (namely, tbe one exemplified by (1)) and the assumption tbat any content can be seen as knowledge updated into the context. Indeed, if presupposition is studied in tenns of knowledge, and if all content is asslUlled to be knowledge, it is just one step further to studying presupposition independently from content, that is, fonnally. In the same way, if presupposition is studied in tenns of updating, and one assumes tbat context can be updated with every kind of content, then, once again, it is just one step further to studying presupposition formally, tbat is, independently from tbe nature of what is updated. But not all content is knowledge and not every content can be updated.

8 Of cmu-se, we can imagine, for instance, stars smiling and chatting like hlUllan beings. However, in this hypothesis, stars would actually be hlUllan beings and we would be imagining human beings in the fmm of stars.

CHAPTER 7 A MODEL FOR PRESUPPOSITIONS :

THE LOOKING-GLASS

In Chapters Two to Six, I tried to defend the distinction between contingent and ground presuppositions, and the idea that the latter are prototypical presuppositions, both as regards some comprehensive theories on this phenomenon and as regards the core features of the phenomenon itself. If the reader is disposed to accept my main idea, we are at a crossroads: either we explore its implications for the classical issues around presuppositions, or we focus on the study of ground presuppositions. The first alternative will be developed in the second part of this book (Chapters Eight to Eleven) and the second in the third part (Chapters Twelve to Fifteen). Before doing this, however, let me picture a model of the phenomenon of presupposition which highlights the leading ideas of the following parts.

7.1 The paradox of the looking-glass Let us imagine a friend of ours, Paul, smiling and a looking-glass mirroring him. What does the looking-glass reflect? Certainly, it reflects the fact that our friend is smiling. However, does it also reflect that he is a human being capable of having feelings? It seems odd to hold this. The same holds for the relationship between language and ground presuppositions. Consider the utterance (1): (1) Paul smiled The utterance (1) is the analogue of the surface of the looking-glass. A proposition like Paul is a human being expresses the consistency

54

Chapter 7

condition of (1), but tbis consistency condition is not expressed by (l)'s meaning any more than the picture of someone smiling depicts the idea that he is a human being. An intuitive way to see this could be to insert (1) as a direct object of a verb like to dream: (2) George dreamed that Paul smiled In (2), did George also dream tbat Paul is a human being? And if George dreamed that Alice discovered that Paul cheated on her, did he also dream tbat Paul cheated on Alice? Be tbis as it may, the model of tbe looking-glass has an advantage over the dream. Imagine breaking the looking-glass. If we do so, the reflected image becomes distorted. By destroying the surface of the mirror, we destroy the condition under which that image can exist. However, the integrity of the mirror's surface is not a possibility condition of the scene itself, which has its O\Vll. On the one hand, the fact that a certain mirror is intact is a contingent fact, far more contingent than the idea that human beings have feelings. On the other hand, when the scene of our friend smiling is reflected, there is an overturning: the integrity of the mirror becomes essential, while the coherence conditions of the scene itself become subordinated to it. This overturning is the paradox of the looking-glass. This paradox may have two interpretations, one negative and one positive.

7.2 Negative interpretation If one only stares at the mirror, one could easily mistake the integrity of its surface for a consistency condition of the scene reflected because it is shared by all images. The more a presupposition concerns the surface of discourse, the more it depends on linguistic expressions functioning as triggers, and the more it seems crucial. A striking example of this is the so-called "existential presupposition" (cf. § 1 1 .4). Vice versa, the more a presupposition concerns an experience's consistency conditions, the less it depends on linguistic triggers and the less is remarked.

A model for presuppositions: the looking-glass

55

From this point of view, the function of the paradox is negative: it invites one to clearly separate presuppositions which govern textual or discursive coherence from those which govern our

experience's

consistency. This insight will orient the second part of the book.

7.3 Positive interpretation Once it is made clear that a discourse surface's integrity conditions are not the possibility conditions of the reflected scene, the mirror becomes a valuable instnnnent. The reason is precisely the overturning highlighted by the looking-glass paradox. By reflecting ground presuppositions in selection restrictions, language subdues them to its 0\Vll fOllllal mould. Therefore, language becomes autonomous from ground presuppositions and allows us to create inconsistent meanings which reveal them. On the one hand, in our everyday lives, ground presuppositions are at work directly. As a consequence, our experience cannot offer any example of a violated ground presupposition because it is ruled by them. Ground presuppositions, then, are practically invisible. On the other hand, in language, ground presuppositions are at work indirectly, that is, as reflected in utterances. As a consequence, language can produce examples of frustrated ground presuppositions. This time, ground presuppositions can be detected and described (cf. Chapter Twelve). In this sense, the function of the looking-glass paradox is positive: it points to the language (the conceptual inconsistency) as a natural laboratory for studying ground presuppositions. As Prandi (2004) states: As consistency criteria escape from direct experience, the only way to analyse them is to pay attention to the effects of their silent work, that is, to focus on instances of transgression that ignore, or intentionally break, essential conceptual bOlUldaries. Among these instances, inconsistent linguistic expressions enjoy an indisputable privilege. (Prandi 2004: 1 04)

PART 11 ISSUES ON PRESUPPOSITIONS

CHAPTER 8 POSITED AND PRESUPPOSED CONTENT

The first issue of the classical debate that

I would like to tackle is tbe

distinction between posed and presupposed content. Sometimes, the fmmer can be formulated much more easily than the latter; sometimes, the opposite is true. In my opinion, such a variable balance does not challenge the idea of presupposition, but rather detects different layers of presuppositions.

8.1 Some difficulties in the distinction Consider two classic examples of triggers: ( l a) Alice did not discover that Paul cheated on her. (2a) Alice did not manage to repair the car. These utterances contain a factive verb (Kiparsky and Kiparsky 1973) and an implicative verb (Karttunen 1973a). According to the test of tbe negation, their posed content is sometbing like (1 b) and (2b): (lb) Alice does not know that Paul cheated on her. (2b) Alice did not repair the car. Their presupposed content is something like ( lc) and (2c): ( l c) Paul cheated on Alice. (2c) The car was somehow difficult to repair. Now, if we compare (1) and (2) in more detail, two observations can be made.

60

Chapter 8

The first is as follows. The presupposed content of the implicative verb to manage is quite marginal as regards the posed one. Consider (2d): (2d) Alice managed to repair the car without any effort. The absence of contradiction in (2d) suggests that (2b) and (2c) are not linked by any logical or semantic relationship. On the contrary, the presupposed content of the factive verb to discover clearly provides a ground for the posited one. Consider (Id): (Id) ?Alice discovered that Paul cheated on her, but Paul never

cheated on her. The inconsistency of (Id) stresses that (lb) and (lc) are linked by some kind of logical or semantic relationship. Let us move on to the second observation. On the one hand, it is very easy to isolate the posed content of the implicative verb, but it is harder to isolate the presupposed one. If one tries, as we did in (2e), the result is questionable. According to Halliday (1970), (2c) does not carry an ideational function in respect of (2a), but rather an interpersonal one. On the other hand, the presupposed content of the factive verb is very clear, but it is difficult to independently state its posed content. If one tries, as we did in (lb), the result will contain another presuppositional trigger (like

to know). Thus, we are left with the following question: how can these facts be accounted for? There are two options: either we dismiss the distinction posed/presupposed or we reinterpret it.

8.2 First option: rejecting the distinction The first option rejects the whole opposition between posed and presupposed. Observations like the first, for instance, have led Wilson (1975) to develop non-truth-conditional semantics. Observations like the second show that, in linguistic practice, the posed and the presupposed content cannot be clearly detached. From this fact, one may be tempted to conclude that the very distinction between posed and presupposed content

Posited and presupposed content

61

should be abandoned. It seems to me that such a conclusion is hasty. Consider a felicity condition. Theoretically speaking, the sincerity condition of a promise is distinct from its preparatory conditions (for instance, the idea that the promise must be right for the interlocutor). However, from a practical point of view, a promise is not a list of felicity conditions, but rather an indivisible entity. These two perspectives (theoretical and practical) are not contradictory, but complementary: the fact that it is reasonable to separate two felicity conditions does not mean that this must concretely be possible. The same remark holds for the posed/presupposed distinction. Consider the foundations of a building. Clearly, its roof and walls carmot hold without the foundations. However, it is not only possible-but crucial-to conceptually separate the foundations from the walls and the roof. This leads to the second option, which consists of reinterpreting the opposition posed/presupposed.

8.3 Second option: reinterpreting the distinction I suggest that the variation in the difficulty in formulating the posed and presupposed content does not reject the distinction posed/presupposed, but rather detects a difference in the presuppositions at stake. Let us examine the details. On the one hand, the fact that the presupposed content of a verb such as to manage is difficult to fOlTIlUlate is consistent with the idea that, here, the presupposition is created or coded by language. Thanks to words like

to manage, to spare, to deprive, mum, dad and so on, the language packs together infOlmation carrying out an interpersonal function (the presupposed content) and infonnation carrying out an ideational function (the posed content). These fonns of information, in themselves, are umelated: we can think of the latter independently from the former. Therefore, those words cause some infonnation (the presupposed) to enter into the orbit of some other information (the posed) and to work as a

62

Chapter 8

contextual requisite for the appropriate use of the word itselP. One could label the fmmer "connotative meaning" and the latter "denotative meaning". Be that as it may, it is because there are words such as to manage, to spare, mum and dad that some infOlmation becomes their presupposition; without these words that presupposition is not even expressible1D• On the other hand, the fact that the presupposed content of a verb like

to discover is easy to explain is consistent with the idea that, this time, the presupposition is not created by language, but rather inherited by it from reality. It would be absurd to hold that it is the presence of the verb to discover that makes the truth of a fact a consistency condition of a discovery. That a true fact is a presupposition of a discovery is, so to speak, in the nature of things. This is the reason why, here, we carmot fOlTIlUlate the posed content independently from the presupposed one. Note that this is even more evident for ground presuppositions: how could we possibly paraphrase the posed content of Alice is sad independently from the idea that she is a human being and that human beings have feelings?ll

9 It seems to me misleading, then, to analyse the presupposition of a verb like to spare as an adpositive clause, a relative clause, a parenthetic clause or any other form of aside clause (Wilson 1975: 144). An aside clause is not a requisite for the use of an utterance. Instead, the presupposition of to spare is a requisite for the appropriate use of that word. So, to spare gives to a certain information precisely the flUlction that that information does not have if it is an adposition. 10 This is evident when the packed information concerns the interpersonal flUlction. However, this does not imply either that the presupposition should necessarily concern the interpersonal function, or that the presupposition should be necessarily inexpressible beyond the trigger. The word bachelor, for instance, packs "adult man" and "not married", but "adult man" has an ideational function and can be clearly formulated. This being said, if "adult man" becomes a presupposition as regards "not married", it is only thanks to the existence ofword bachelor. 11 There are some words which offer structure to host and create presuppositions from the content at stake. Consider, for instance, the following examples: (a) Although it rained all night, the streets are dry. (b) It rained all night, but the streets are are dry. (c) It rained all night and the streets are are dry. The utterances (a), (b) and (c) presuppose something like (d): (d) ifit rains all night, streets should be wet.

Posited and presupposed content

63

8.4 Lexical presuppositions Following the second of the aforementioned options enables us to identify another kind of contingent presupposition. Consider an example like (3a): (3a) Mary deprived Paul offish. The fact that Paul loves or hates fish is not a motif to wonder if Mary actually deprived him of fish, but rather if it is appropriate to use that very verb rather than to spare. Contrast (3b) with (3c): (3b)

?Since Paul loves fish, I wonder ifMary deprived him of it.

(3c) Since Paul hates fish, I wonder

if saying that Mary

deprived him of it is appropriate. Example (3c) is perfectly clear, while (3b) is somehow odd. Now, consider an example like (4a): (4a) Paul stopped smoking. The fact that Paul smoked is not simply a motive to wonder if it is pertinent to use that verb, but rather a motive to wonder if he actually stopped. Indeed, (4b), contrary to (3b), is a perfectly felicitous utterance: (4b) Since George smoked a lot I wonder ifhe stopped Hence, this time, we are not simply confronted with a precondition for talking appropriately, but with a precondition for a process to take place. These are the kinds of discursive presuppositions so far considered. If this is true, to spare, to deprive, to manage and so on identify another kind of contingent presupposition, namely, preconditions for the appropriate use of a term. I label them lexical presuppositions.

This contingent presupposition is coded by although in Ca), inferentially invited by but in Cb) and completely inferred in Cc).

CHAPTER 9 PRESUPPOSITION AND TRUTH CONDITIONS

In this chapter, I discuss the link between presuppositions and truth conditions. This question is usually raised, not with respect to the presupposition itself, but to an utterance presupposing it, namely: If a

presupposition is false, is the utterance presupposing it false too or is it truth-valueless? Such a question assumes that a presupposition can be true or false. This assumption must be challenged. Hence, my question will be: Does the assertion of the content of a presupposition have truth conditions?

9.1 Truth conditions vs. ground presuppositions Let us begin by considering contingent presuppositions: ( l a) George stopped smoking. (1 b) George continues to smoke. ( l c) George smoked A contingent presupposition like ( l e) is a contingent fact and, as such, has specific truth conditions: the assertion George smoked is true if, and only if, George smoked. However, since only assertions have truth conditions, and since a presupposition is not really asserted, the fact that ( l c) can have truth conditions is not relevant when it works as a presupposition. Let us now turn to a ground presupposition: (2a) Alice is sad (2b) Alice is happy.

66

Chapter 9 (2c) Alice is a human being. As seen in Chapter Five, §5.1, the distinctive feature of the content ofa

ground presupposition like (2e) is that it is never asserted. Hence, its eventual truth conditions are never relevant. Moreover, what would they be? Of course, by applying the Tarskian formula, one can say that Alice is

a human being is true if and only if Alice is a human being. However, this fonnulation simply gives the illusion of truth conditions. Indeed, concretely, what is the state of things that verify or falsify (2c)? Once again, as seen in Chapter Five, §5.2.3, we cannot answer by presenting the actions or feelings that we usually ascribe to people because they already presuppose that they apply to a person. So, both in the case of a contingent presupposition and in the case of a ground one, the answer to our question (namely, Does the assertion a/the

content of a presupposition have troth conditions?) is no. In the first case, the content of the presupposition can have truth conditions not as a presupposition, but as the content of a contingent assertion. In the second case, truth conditions are never relevant because the content of a ground presupposition is never asserted. The fact that a ground presupposition does not have truth conditions highlights a hiatus between truth and truth conditions: the former does not need the latter. In other words, something can be a truth and have truth conditions (like someone smoked), or be a truth and not have truth conditions (like human beings have feelings). A truth without the need of verification-that is, an unconditional truth-is precisely the definition of an a priori necessary truth. In light of the previous remark, consider the following quotation from Wittgenstein: If true is what is grOlUlded, then the grOlUld itself is neither true, nor false. (Wittgenstein 1 969) Faced with this statement, we have a dual intuition. On the one hand, it seems to express a sound and deep insight. On the other hand, one could well say that something like Human beings have feelings is true and not

Presupposition and truth conditions

67

false. If we separate truth from truth conditions, these intuitions are no longer contradictory. Wittgenstein's "ground" is a non-truth-conditional truth (that is, a ground presupposition); what is grounded is a truth­ conditional truth (that is, an a posteriori contingent truth)". The truths that we argue about in our everyday lives are these latter. Moreover, the notion of truth should be distinguished from the adjective true. The fact that one can perfectly say that Alice is a human

being is true just like one says that George smoked is true does not mean that the fmmer is like the latter. For instance, the fact that one can perfectly well say This is a rose for both a painted rose and a real one does not mean that a painted rose is like a real one. Rather, our use of the word

rose takes that difference for granted and it would clearly be absurd to think that language should distinguish two words for rose. In the same way, our use of the words true (or false) takes the difference between a necessary a priori truth (a ground presupposition) and a contingent a posteriori truth for granted; it would be absurd to think that language should not allow, in one of these cases, the use of the adjective true.

9.2 Presuppositions and necessary conditions Strawson (1952) considers presupposition to be a necessary condition. Intuitively, the fact that Paul had cheated on Alice looks like a sort of necessary (but luckily for him not sufficient!) condition for Alice to discover it. This is so because this fact is a contingent presupposition. If we consider a ground one, we realise that the concept of presupposition is reluctant to be interpreted as a necessary condition. Let me illustrate this point with an example. I have a diabetic cat. A necessary (but not sufficient) condition for my cat to be in good health is that it has one insulin injection a day. But I would not be inclined to say that another necessary (not sufficient) condition for my cat to be in good health is that it is a living being. Of 12 We can answer in a similar way to another ofWittgenstein's questions: "This body has extension." To this we might reply: "Nonsense!" but are inclined to reply "Of cmu-se!" "Why is this? (Wittgenstein 1953:252).

68

Chapter 9

course, in a sense, this latter is even more necessary than the fmmer. However, from a discursive point of view, it is not at all a relevant necessary condition. Imagine, for instance, my vet saying:

Dear Marco, for your cat to be in good health, it is not sufficient that it has an insulin injection each day, your cat also has to eat specific biscuitsfor diabetic pets/ ? Dear Marco, for your cat to be in good health, it is not sufficient to be a living being, it also has to have an insulin injection each day (and eat

biscuits) /

9.3 Presuppositions and necessity Let us come back to the example ( l a) (George stopped smoking) and let us consider the logical definition of presupposition in terms of possible worlds: Utterance U presupposes P if and only if: (a) in all possible worlds in which U is true, P is true (b) in all possible worlds in which U is false (not-U is true), P is true. Facing this definition, the following argument has been put forward: if what is true in all possible worlds is necessary; if between U and non-U,

tertium non datur; and if P is a presupposition of U; then P is a necessary truth. But, of course, the fact that Paul smoked is not a necessary truth at all! If one adopts a functional conception of presupposition, and distinguishes contingent from ground presuppositions, such an argument makes no sense. As regards the consistency of the notion of presupposition, it is necessary that the fact that George smoked is a necessary, a priori truth in respect of the assertion that he has stopped smoking. But it is not necessary that it is a necessary a priori truth beyond that assertion. As Epicure puts it: Necessity is an evil; but there is no necessity for continuing to live with necessity.

Presupposition and truth conditions

69

Since a presupposition is a necessary a priori truth within the limits of the practice in respect of which it works presuppositions

are

relative

a

as

pnon

a presupposition, contingent necessities,

while

ground

presuppositions are absolute ones. In other words, a functional notion of presupposition implies a relational notion of a priori and necessity (Prandi 2004:241). This latter preserves the existence of an essential link between a priori and necessity: it simply subordinates this link to the limits of a certain practice (which can cover a speech act or all our life). Now, the existence of an essential link between a priori and necessity has been criticised by Kripke (1980) through his notion of "necessary a posteriori truth". This point, then, requires a discussion. First of all, note that something is not simply known a priori in the absolute, but rather

as

regards knowing something else. In other words, the

fact that P can be learned or discovered does not prevent there being something else with regard to which P holds as known a priori. Let us come back to our examples ( l a) (George stopped smoking). Of course, one can learn a contingent presupposition (that George smoked) through the utterance presupposing it (by accommodation). However, the presupposition still holds a priori as regards knowing if George stopped smoking or not. Secondly, in my opinion, the expression "necessary a posteriori truth" makes sense only if the adjectives necessary and a posteriori are applied from different perspectives. In order to show this point, I propose to interpret the couples necessary/contingent and a priori/posteriori through the distinction between synchrony and diachrony. Consider a chess match. A chess match can be looked at either from the external point of view of a historian of chess, or from the internal point of view of a player. These perspectives can be opposed through the following utterances:

If the history had been different, the knight would not move like

an

L,

but (say) like an S. would be more coherent (economic, challenging. . .) if the knight moved like an S.

It

70

Chapter 9 Utterances like these sound reasonable if uttered by the historian

looking at the match, but they are out of place if uttered by the player engaged in the match. Indeed, imagine a player moving the knight like an S and justifying himself in that way! n From the historian's perspective, the fact that the knight moves like an L is the result of a historical process or of someone's arbitrary decision; in this sense, it is contingent and a posteriori. From the player's perspective, that very same fact is necessary and a priori; it is a consistency condition making it possible to play chess. So, the fact that the knight moves like an L, although it is not a priori and a necessary truth in the absolute sense, it holds

as

such with respect to playing chess.

Using Saussure's terms, the historian's point of view is diachrony, while the player's point of view is synchrony. From a diachroinc perspective, for instance, the fact that French opposes riviere (a river going into another river) to fleuve (a river going into the sea) is a contingent and a posteriori truth. From a synchronic perspective (from the point of view of a French speaker using those words), that distinction is an a priori and necessary truth, making it possible to speak French correctly. Of course, if someone does not know how to play chess or to speak French, they have to learn how the knight moves and the distinction between

fleuve et riviere. However, as regards playing chess and speaking French, it would be wrong to conclude that they are contingent, a posteriori truths14•

13 Note that, mutatis mutandis, this is analogous to someone behaving consistently with a proposition like Human beings have feelingsfor such and such evolutionary reason (cf. Chapter Five §S.2.3). 14 Consider another well-kno"Wll example: the definition of the meter. Roughly speaking, one can stipulate that a metre is the length of this bar. Note that the circularity of this definition is similar to nouns of colours, substances, smmds and so on. Be this as it may, let us apply the previous remarks to that proposition. On the one hand, from a historical point of view (from the point of view of someone who is not using a stick to measure), that proposition is a contingent, a posteriori truth. It is contingent because a metre could have been defined as the length of another bar. And it is a posteriori because that definition is the outcome of an act of stipulation. On the other hand, from the point of view of someone, here and now, using the bar to measme something, that proposition is a necessary a priori truth. It

Presupposition and truth conditions

71

If we now cross the distinction synchronic/diachronic with the oppositions a priori/posteriori and contingent/necessary, we can distinguish two uses: convergent and divergent. In

convergent

uses,

the

oppositions

a

priori/posteriori

and

contingent/necessary are applied from the same perspective: synchronic or diachronic. In this sense, the fact that the knight moves like an L and that a language has the grammatical rules that it happens to have are, at one and the same time, a priori necessary truths synchronically, and a posteriori contingent truths diachronically:

That the knight moves like an L is known a priori (because I have to know that rule before thinking how to move the knight while playing chess) and it is a necessary truth (because I cannot argue about that rule while playing chess). That the knight moves like an L is known a posteriori (because I have actually learned that rule) and it is a contingent truth (because the history of chess could have been different). In divergent uses, those adjectives are instead applied from different perspectives. So, this time, the fact that the knight moves like an L or that a language has certain grammatical rules can be considered both a priori, contingent truths and a posteriori, necessary truths. The first statement applies a priori from a synchronic point of view, and contingent from a diachronic one15:

is necessary because the fact that a metre could have been the length of another bar is ruled out. It is a priori, because that definition enables them to measure, that is, to experience (a posteriori) that something is or is not one metre long. 15 Kripke's original examples of a posteriori, necessary truths do not concern the grammar of a language, but rather physical laws. This fact, however, does not affect my arglUllent. As regards a specific language, a grammatical rule is necessary and a priori in exactly the same way as a physical law is necessary and a priori as regards nature. The difference is simply the scope of the practice in respect of which they are necessary and a priori. The fact that we can ignore, learn and discover the laws of nature does not challenge the fact that these laws are a priori regarding how nature works; just like the fact that we can ignore, learn and

72

Chapter 9

That the knight moves like an L is known a priori (because I have to know that rule before thinking how to move the knight while playing chess), but it is a contingent truth (because the history of chess could have been different). The second statement, inversely, applies a posteriori from a diachronic point of view, and necessary from a synchronic one:

That the knight moves like an L is known a posteriori (because I have actually learned that rule), but it is a necessary truth (because I cannot argue about that rule while playing chess). In this perspective, talking about "a posteriori, necessary truth" does not really challenge-but rather implies-the essential link between necessity and a priori. Note that, at this point, we can separate "everyday facts", "everyday facts working as presuppositions" and "ground presuppositions". As regards an everyday fact as such, only the notions of a posteriori and contingent apply: hence, these notions cannot diverge a fortiori. This is the reason why (as remarked in Chapter Six §6.4) nobody would consider

George smoked to be a tautology. As regards a ground presupposition, on the other hand, only the notions of a priori and necessary apply; hence, once again, these notions cannot diverge a fortiori. This is the reason why (as stressed in Chapter Six §6.4) the proposition Human beings have

feelings looks immediately like a tautology. When a contingent fact works as a presupposition, instead, both the notions a posteriori/contingent and a priori/necessary apply: the former apply to the presupposition as a contingent fact, while the latter apply to it as a presupposition. Now, this is the only case in which the preconditions for those notions to diverge are fulfilled. In other words, only a contingent presupposition can be seen as an a posteriori, necessary truth precisely because of its double life (cf. Chapter Five §5.3).

discover the grammar of our mother tongue does not challenge the fact that that grammar is a priori regarding speaking that language.

Presupposition and truth conditions

73

9.4 Presuppositions and truth A presupposition is a fact working as an a priori and necessary truth in respect to a practice. If a presupposition can be defined in terms of truth, it seems reasonable that truth could be defined in terms of presupposition. If we do so, truth turns out to be the object of presupposition. This can be illustrated through two observations. The first observation is well known. Consider (4) and (5): (3a) Mary discovered that Paul had cheated on her. (3b) Mary discovered a letter. (4a) It is true that Paul had cheated on Mary. (4b) There was a letter. As stressed by Kiparsky and Kiparsky (1973), a factive verb presupposes the truth of a fact like it presupposes the existence of an object. The second observation is less known. Consider (5a) and (5b): (5a) Paul had cheated on Mary. (5b) It is true that Paul had cheated on Mary. It

seems to me that the difference between (5a) and (5b) can be

highlighted by considering the following examples": (6a) Why did Paul cheat on Mary ? (6b) ? Why is it true that Paul che ate d on Mary? Using the terms introduced in Chapter Five §5.2.3, the question (6a) can receive a genuine answer. The question (6b), instead, sounds odd and can only receive a circular answer, namely:

16

For a discussion about the theories of truth, cf. Horwich (1998).

74

Chapter 9 It is true that Paul cheated on Mary because .. Paul cheated on Mary / He did it.

Hence, as a matter of fact, truth presents something as excluded from argumentation: that is, ready to be considered a necessary a priori truth as regards some practice (in our example, asking if Mary discovered it). In this sense, saying that something is true means that this something can be presupposed. Note that the difference between (6a) and (6b) is neutralised by ground presuppositions, which are excluded from argumentation by default: (7a) ?Why is Mary a human being? (7b) ?Why is it true that Mary is a human being? As pointed out in Chapter Five §5.2.3, (7a) can only receive circular or not relevant answers. The justification of a ground presupposition immediately sounds odd, without the need of the construction it is true

that.. This amounts to saying that a ground presupposition is already relied upon as a truth: it is presupposed from the beginning.

CHAPTER 1 0 PRESUPPOSITION AND INFERENCE

Since the very beginning, presupposition has been analysed as some kind of implication, entailment, inference and so on. The whole issue has focused on how to account for the fact that a presupposed content is implied by both positive and negative assertions and how to characterise this weird implication that is capable of surviving negation. According to Russell (1905), presupposition is an implication of tlie meaning of a noun plirase in a referential position. According to Strawson (1950), presupposition is an implication of the referential use of a noun phrase. These positions diverge radically, but share the same assumption: presupposition is construed as an implication. The following debate increases the list of tlie elements triggering these implications, focusing on the mechanisms used to cancel (block, suspend, etc.) them or to project them through utterances, but it maintains the core assumption. This is shown by the competing analyses of presupposition in terms of conversational implicature, conventional implicature or both. Here are some authors that have developed these issues in detail: Grice (1981), Sperber and Wilson (1979), Wilson (1975), Karttunen and Peelers (1977, 1979), Gazdar, (1979), Kadmon (2001) and Simons (2004, 2007, 2010). However, along this patli, the distinction between presuppositions and implications becomes thinner and thinner lUltil it fades away. I maintain that this outcome is lUlavoidable insofar as two points are not clearly acknowledged. First of all, presupposition-in itself-rejects inference: grolUld presuppositions carmot be inferred, implicated, entailed, commlUlicated etc. If contingent presuppositions can be seen as implications (inferences, entailments etc.), it is only insofar as they are not presuppositions, but contingent facts. Secondly, in the case of contingent

76

Chapter 1 0

presuppositions, presupposition and inference are logically independent notions: the fmmer denotes a function of a content, the latter denotes the access to this content. Hence, one can access through inference a content working as a presupposition. This chapter is devoted to illustrating these points.

10.1 Inference vs. Ground presuppositions Ground presuppositions, contrary to contingent ones, carmot be inferred. More precisely, they can neither be communicated, nor emich some coded meanmg. Contrast the following dialogues:

- DidAlice ever use drugs?

(1)

- Alice stopped smoking ajoint one hour ago. What do you think? - ?Is Alice a human being?

(2)

- ?Yesterday she was sad What do you think? Dialogue (1) is consistent, while dialogue (2) is surreal. So, for the receiver it is coherent to ask and to infer the content presupposed by (1), and for the speaker it is coherent to communicate it because it is just a contingent fact. But this is not true for (2). Consider the following example: (3) Paul got into Alice 's house with the keys. Facing the prepositional phrase with the keys in (3), we can imagine two processes of inferential enriching (prandi 2004:44) according to two different contexts: (3a)

There was a robbery at Alice's house. Paul is the principal suspect. We know he previously stole her keys. Now someone asks: how did Paul manage to get into Alice's house? We answer: Paul got into her house with the keys.

Presupposition and inference (3b)

77

Paul lost his car keys. We wonder where he could have left them. We begin saying: Now, Paul got into Alice 's house with

the keys. then .. The same prepositional phrase with the keys would be inferentially emiched by an instrument-relation in (3a) and in a simple union-relation in (3b). Conceptually, these inferential enrichments are connected in the sense that the instrument-relation is a more complex version of the union­ relation; that is, an instrument is not merely an object present during an action, but an object by which the action can be performed. So, conceptually, we have two 'stations' along the same line travelled by the 'train of inference': one station is the union, the other station is the instrument. Mistaking the inference in (3a) means to get off at the union, the station before the instrument. Mistaking the inference in (3b) means to get off at the instrument, the station after the union. Now, what is the condition to which the train of inference can travel on this line? The answer is straightforward: the idea that the keys are inanimate, concrete entities (things). In (3a) and in (3b), this idea is neither coded, nor inferred. It is not inferred because it is already there, and it is not coded because, in (3), the preposition with does not contain any kind of coding device. Indeed, the distinction between human beings and tools is not a conventional rule established by English, Italian, French or by any other language, but an ontological distinction presupposed by our form of life. This distinction, in (3), delimits the space of possible inferential enrichments (right or wrong) of the noun phrase with the keys.

10.2 Presupposition vs. Access to a presupposition Presupposition and inference are logically independent notions. Consider the following passage from Wilson: Let me first illustrate the two different approaches in action, by considering how each of them deals with sentences (1) (3): (1) Priscilla stopped reading Model Theory for Two-Year-Olds. (2) Priscilla didn't read Model Theory for Two-Year-Olds.

78

Chapter 10 (3) Priscilla has been reading Model Theory for Two-Year-OIds. According to the presuppositional approach, (1) and (2) both presuppose (3), Hence, if either (1) or (2) is true, (3) must also be true, and if (3) is false, both (1) and (2) must lack a truth value. According to the entailment analysis which I am proposing (1) entails (3). Hence, if (1) is true, (3) must also be true. But if (3) is false, (1) will also be false and (2), the negation of (1), will be true. (Wilson 1975: 1 7 1 8) [my emphasis1 This passage is emblematic because it opposes the two approaches­

presuppositional and inferential-as exclusive. This opposition relies on the assumption that presupposition and entailment are logically related: if x

is an entailment, then it is not a presupposition, and vice versa. As a

consequence, from this perspective, presupposition and entailment are different kinds of implications. In my view, on the contrary, the notions of presupposition and inference (implication, entailment, etc.) are not two kinds of the same species, but two completely different animals. The words entailment,

inference, implication and so on denote a logical relation of deduction, abduction or induction. The word presupposition denotes a function (of consistency

condition).

Hence,

inference

and presupposition

are

incommensurable notions because they work on different levels: the fmmer works on a logical

or

argumentative level, whereas the latter works

on a functional level. Note, by the way, that they behave like forces with a contrary orientation. Since presupposition is a function carried out towards something, it is ideally placed before this something: presupposition is, so to speak, a backward-looking notion. Since inference is a conclusion drawn from something, it is ideally placed after this something: inference, implication, entailment and so on are, so to speak, forward-looking notions. Concepts like non-main-point inferences or background inferences (cf. Thomason 1990, Simons 2007) are affected by intrinsic tension precisely because they try to coerce two opposite forces together.

Presupposition and inference

79

Be this as it may, if we come back to Wilson's quotation, nothing prevents us from holding both that (1) entails (3) and that (1) presupposes (3), because they express different ideas. In other words, nothing prevents us from looking at a presupposition as a kind of logical entailment; by doing this, we simply put into parenthesis the function of consistency conditions of a practicel7• Now, given the logical independence between implication and presupposition, one can distinguish between two questions: (A) What is thefunction of certain content? (B) How do we accede to (or learn about) that content? The notion of presupposition is an answer to (A). The notions of inference, implication, entailment and so on are answers to (B). Questions (A) and (B) are logically umelated, but practically connected. Let us apply them to contingent and ground presuppositions. Consider (4): (4a) George stopped smoking. (4b) George smoked What is the function of (4b)7 The answer is that (4b) carries out the function of a consistency condition in respect of (4a). At this point, we can ask question (B): how can we accede to presupposition? The answer is: through a conventional implicature triggered by to stop doing something. Note that it makes sense to infer (4b) from (4a) only if we ignore it, and, 17 Van Fraassen, (1973 :97) has remarked that a logical definition of presuppositions cannot distinguish between presuppositions and trivial formal tautologies. The problem, then, is how to improve that definition to avoid such a consequence. This problem hinges on the assumption that a presupposition is a (special) kind of implication. From a logical point of view, it is true that George smoked and It is raining or It is not raining are both entailed by George stopped smoking. From this perspective, then, it is useless to discard the second implication as trivial. The difference emerges only at the level of function: George smoked is a consistency condition of the utterance George stopped smoking, but It is raining or It is not raining is not. Only this level reveals the presupposition.

80

Chapter 1 0

indeed, we can ignore this because (4b) is a contingent fact (cf. Chapter Three §3.1). From this point of view, the conventional implicature is the

medium that makes something, which in itself is not a presupposition, a presupposition. Consider (5): (5a) Alice is sad (5b) Alice is a human being and human beings have feelings. The answer to question (A) is that (5b) carries out a function of a consistency condition not only with respect to (Sa), but with respect to our whole ontology: it is a ground presupposition. If this is true, however, this time, it is pointless to ask: how do we accede to (5b)? The reason is that, since (5b) is a grOlUld presupposition, it is already there and it makes no sense to ignore it. Consequently, the whole issue-if (Sa) entails, implicates and so on (5b); that is, the way to accede to (5b)-becomes pointless too. In this sense, a ground presupposition is an immediate presupposition. In the terms of Chapter Nine §9.3, acceding to the presupposition in (4a) simply means to learn a posteriori something holding a priori (and necessarily) as regards another practice. This is possible only for contingent presuppositions: that is, if, and only if, that 'something' does not hold a priori as regards to all practices.

10.3 Presupposition and accommodation If you learn that Mary was married when I say: Mary was married but

then she divorced, you do something analogous to reading how the knight moves in a chess manual, and then achieving checkmate by moving the knight. If you leam that Mary was married when I say: Mary divorced, you do something similar to learning how the knight moves after I put you in checkniate by moving it. In both cases, the fact that Mary was married is presupposition of her divorce. However, in the first case, you accede to that presupposition through its very assertion, while, in the second, you

Presupposition and inference

81

accede to it through a conventional implication. This latter (i.e. learning a posteriori something that holds a priori in respect to some practice) is also known as "accommodation of presuppositions" (Lewis 1988). Now, accommodation represents a danger for the very notion of presupposition. If one adapts their background of knowledge to assimilate a new utterance carrying a new presupposition, then one detects a conflict between two presuppositions (the old and the new). Hence, tbe phenomenon of accommodation puts a blotch on tbe dysfunction of tbe natural use of a presupposition, which should fit a specific background (it is not by chance, then, that Lewis's analogy is based upon an incorrect behaviour). So, we are at a crossroads. If we maintain that accommodation is a crucial feature of presupposition, then presupposition is characterised through its dysfunction. If we refuse to consider accommodation, a crucial feature of presupposition, we lose the possibility of understanding the rhetorical exploitation of presupposition and its discursive interest. Now, the functional notion of presupposition, at the base of the distinction bet\veen contingent and ground presuppositions, solves the dilemma. On the one hand, it does not make sense to accommodate ground presuppositions because they are already there. In this sense, the notion of presupposition excludes accommodation (just like it excludes inference). On the other hand, contingent presuppositions can be accommodated or accessed insofar as they are not presuppositions in themselves, but contingent facts that can be learned or inferred. lOA

Resurrection for presuppositions

Karttunen and Peters (1979) proposed dismantling tbe phenomenon of presuppositions into different kinds of inferences: We believe that a wide range of different things have been hunped together lUlder this single label and that this fact is more than anything else responsible for the continuing controversy about how to analyse presuppositions. To resolve it we propose to do the sensible thing, namely

82

Chapter 1 0 to divide up this heterogeneous collection and to put the particular cases into other categories of phenomena, such as particularized and generalized conversational irnplicatures, preparatory conditions on speech acts and conventional irnplicatures. Since something is already knmvll about the nature of these other phenomena, in this way we may actually be able to explain some of the diverse behavior of different things that various linguists have at one and the same time called presuppositions. (Karttunen and Peters 1979:2) This is the disintegration of the very phenomenon of presupposition,

the so-called Requiem for presuppositions. Once again, presupposition is a function; inference, implication, etc. are ways to accede to a content that can work as a presupposition. As a consequence, presupposition is actually compatible with, and neutral with respect to, different kinds of implications. These implications concern our ways of accessing some content; presupposition concerns the function carried out by this content18• Reducing presupposition to different kinds of implications is simply not relevant for presupposition itself. For example, in Alice did not discover that her husband had cheated on her, the presupposition is accessed through an invited inference, while in Alice

discovered that her husband had cheated on her, it is accessed through a conventional implication. If one just focuses on the way to accede to them, one simply misses their point as presuppositions, that is, the unity of the phenomenon at stake.

10,5 Classification of presuppositions In a way, the aforementioned disintegration of presuppositions is nothing more than a classification focused on our access to some content and not on its function (of presupposition). This attitude is shared even by an 18 Note that, by separating the way to accede to certain information (implication, inference, etc.) from the flUlction of that information (presupposition, communication, allusion, etc.), one can, at one and the same time, understand the intuition behind notions such as non-main-point implications or background, non­ foreground inferences and release their inner tension (cf. §1O.2, in the present chapter).

Presupposition and inference

83

articulated classification like the one proposed by Ducrot (1980): (a) General presuppositions [ . . . ] (b) Illocutive presuppositions [ ... ] (c) Speech presuppositions, grOlUlded on specific morphemes. Between them, one can distinguish: (cl) Existential presuppositions [ . . ] [definite articles] (c2) Verbal presuppositions[ ... ] [verbs like to stop, to discover, etc.] (c3) Construction presuppositions [ ... ] [extrapositions, etc.] (c4) Adverbial presuppositions [ ... ] [adverbs like even, etc.]. (Ducrot 1980:1094 1906) (our translation) Note that, apart from point (a), all other points link presuppositions to their trigger and, hence, offer a classification based not on the presupposition itself, but rather on the modality of our access to it. This is an indispensable step, but it is not the first one to make. Indeed, a classification of something should be based, firstly, on what this something is and not on how we get to it. Otherwise, it is like trying to distinguish breeds of dog by their collars.

CHAPTER 1 1 PRESUPPOSITION AND ANAPHOR

The idea of presupposition as anaphor, put forward by van der Sandt (1992) and developed by Krabmer (1998) and Geurts (1999), forms the climax of a dynamic turn in the debate about presupposition. This turn has its theoretical paradigm in the work of R. Stalnaker and its precursor in Karttunen's

(1974)

notion

of

satisfaction-of-presupposition.

The

importance of the dynamic turn can be measured by considering how it liberated the debate from its major problem, namely, the projection of presuppositions (Langendoen and Savin 1971). Given the impact of the analogy between presupposition and anaphor, I

I will proceed as follows. First of all, I expose the idea of presupposition as anaphor and I raise a first criticism based on the very notion on anaphor (§ 1 1 .1). Then, I raise two devote a specific chapter to discuss it19•

further criticisms: the first is 'external', that is, based on the distinction between contingent and ground presuppositions (§ 1 1 .2); the second is 'internal', that is, concerning the linguistic grounds of that very analogy (§ 1 1 .3).

Finally,

I discuss some consequences as regards the

presupposition of the identification of a referent (§ 1 1 .4).

19 One of the merits ofKarttunen (1 974) and van der Sandt (1992) is that they get rid of the idea of the cancellation of presuppositions. They show that we must distinguish between the presence of a presupposition and the perception of a presupposition. Filters like or, if and and seem to cancel the presupposition because the first member constitutes a micro-context containing the presupposition triggered by the second. As a consequence, this presupposition goes mllloticed. This consequence, however, still holds even if we refuse to believe that presuppositions are anaphors.

86

Chapter 1 1

11.1 Presupposition as anaphor According to van der Sandt (1992): Presuppositions are simply anaphors. They only differ from pronOlUls [ . . . ] in that they contain enough descriptive content ill. establish a reference marker in case discourse does not provide one. In this case, the lexical material will be accommodated. (Van der Sandt 1992:345) [my emphasis] In my opinion, quite to the contrary, this is the best proof that 'presuppositions' are not kinds of anaphors, at least if the word anaphor keeps its usual meaning: . . ] anaphor is an act of reference: By means of an anaphoric pronOlUl, the speaker refers to a referent to which he or she, in the discourse, has already referred to through a previous expression. (Conte [1 996] 1999:19) [my translation] In order to illustrate this quote, consider the following examples: ( l a)

Paul had cheated on Alice and Alice discovered i1 / this / this fact.

(lb)

?Alice discovered i1 / this / this fact

The example ( l a) is informatively good, while (lb) is informatively deficient because, without a co-text, there is no previous textual referent to refer to. The examples (1) show that the function of an anaphoric pronoun is not to introduce a new textual referent, but rather to recall a previously introduced one. Now, let us consider (2): (2a)

?Paul had cheated on Alice and Alice discovered that Paul cheated on her.

(2b)

Alice discovered that Paul cheated on her.

Note that, in (2), it is the example (2b) which is informatively auto­ sufficient, while (2a) is informatively dysfunctional because is affected by

Presupposition and anaphor

87

a repetition. If there is a repetition, however, an infOlmation is re-stated. This suggests that, in (2), the underlined clause does not refer to-does not

recall-a previous information, but it states an infOlmation. Since it is stated, this information can repeat an old one (cf. (2a)) or introduce a new one (cf. (2b))'". So,

as

a matter of fact, the 'presupposition' (using van der Sandt's

expression) works when the anaphoric pronoun does not (cf. (2b) vs. (lb)), and vice versa (cf. ( l a) vs. (2a)). Hence, saying that "presuppositions are simply anaphors, the only difference being that the former establish a reference marker, while the latter do not" seems to me like saying that birds are simply fishes, the only difference is that the former fly, while the latter swim. Indeed, to discover works

as

a presuppositional 'trigger' both in ( l a),

where there is an anaphor, and in (2b), where there is no anaphor21• This implies that anaphor and presupposition are distinct phenomena, which can, but not necessarily, overlap (cf. § 1 l .3).

11.2 Anaphor vs. Ground presuppositions The idea of presupposition as anaphor relies on the possibility of analysing an utterance like George stopped smoking as a conjunction:

George smoked and has now stoppedn.

20

Of course, (2a) must not be confused with something like: Pwl cheated on had Alice andAlice discovered the / that / his cheating. Here, there is an anaphor, but the phenomenon of anaphor is virtually distinct from the phenomenon of presupposition (cf. § 1 1 .3). If, without a context, I say Alice discovered the cheating, the reaction is precisely to spot the lack of a previously introduced referent: which cheating? If, without a context, I say Alice discovered P wl s cheating, there can be accommodation, but accommodation is precisely the reaction to the lack of a previous textual referent: that is, the perception that the anaphora did not work! 21 The pronoun her, of course, is not relevant for the present discussion. 22 Note that this conjlUlction does not really analyse to stop because it repeats it. Moreover, a solution like: George smoked and now he does not is problematic for Van der Sandt's orKarttunen's accounts. In fact, from: a) George smoked and now he does not smoke we can infer b): George stopped smoking. But the first conjlUlct

88

Chapter 11 According to Karttunen (1974), the first member of this conjunction is

the local context satisfying or admitting the presupposition triggered in the second conjunct. According to van der Sandt (1992), it is the antecedent to which the presupposition triggered in the second conjunct is bOlUld. Note that these approaches are complementary. On the one hand, the idea of presupposition-as-satisfaction goes from context to presupposition: the context puts some restriction on presuppositions. On the other hand, the idea of presupposition as anaphor, goes from presupposition to context: the presupposition needs an antecedent in the context. Be this as it may, as already pointed out in Chapter Six §6.3., this analysis does not work for ground presuppositions. For instance, it is pointless to read Alice is sad as:

?Alice is a human being and she is sad The conclusion, once again, is as follows.

It

can be heuristically useful

to consider a contingent fact as a sort of pseudo-antecedent to which the discursive presupposition triggered is 'bound' . However, this analogy works for contingent presuppositions not because they are presuppositions, but because they are contingent facts. So, presupposition as anaphor may help to picture the discursive role of a contingent presupposition, but, as well as presupposition-as-illocution or presupposition-as-attitude, it is no more than an analogy and it is misleading ifit is taken too seriously.

11.3 Anaphor vs. Presupposition: separate phenomena Of course, the previous conclusion can be shared only by someone who already

accepts

the

distinction between

contingent

and

ground

presuppositions. However, it seems to me that the reasoning at the base of the very idea of presupposition as anaphor, is misleading. Let us come back to examples (2a) and ( l a) of a) is neither the antecedent nor the local context of b). If a presupposition is a descriptive anaphor (as held by Van der Sandt) and if in a) the presuppositional trigger is inferred, then we infer an anaphoric element. But one would expect that an anaphoric element is in the text.

Presupposition and anaphor

89

(2a) (?)Paul had cheated on Alice and Alice discovered that Paul had

cheated on her. ( l a) Paul had cheated on Alice andAlice discovered iJ.. Leaving aside the repetition affecting (2a), the simple fact that the sentence Paul had cheated on her-that is, the presupposition-can be replaced by an anaphoric expression in ( l a) does not prove that that sentence is an anaphoric expression too: it simply proves that the anaphoric expression and the sentence occupy the same distributional position. Consider examples (3): (3a) Alice discovered that Paul had cheated on her. (3b) Paul had cheated on Alice and Alice discovered this. What do (3a) and (3b) presuppose? To answer this question, it is not enough to point out the anaphoric expressions her or this, but we must identify their referents. Hence, the presupposition is not these anaphoric expressions. So, examples like (3a) and (3b) do not prove that the presupposition triggered by to discover works like a pronoun, but rather that the identification of a discursive referent made by a pronoun is an independent phenomenon in respect of the presupposition triggered by to

discover. In other words, if a presupposition can exploit anaphoric elements like pronouns or noun phrases-by, so to speak, working on them-then the presupposition does not coincide with these anaphoric elements. Now, if anaphor and presuppositions are separate linguistic phenomena, they can be practically superposed. 'When this happens, it can raise the illusion that anaphor works because of the meaning of the anaphoric expression23• Consider the example (4a):

23 The

same kind of illusion is at the base of Russell's (1 905) and Strawson's (1 950) analysis of presuppositions. Facing the noun phrase the King ofFrance, Strawson (1950) distinguishes its meaning from its uses. These latter, in particular, are

90

Chapter 1 1 (4a) Paul and Mary have a cat. This cat smashed the mirror in the

dining room trying to catch a mouse. Here, the anaphoric link seems to work because the descriptive content of the anaphoric expression (this cat) matches the antecedent (a cat). Since in (4a) there is an anaphor, this leads to concluding that the presupposition triggered by the noun phrase (namely, that there is a cat) is an anaphoric expression. However, in fact, not only does the anaphor not need to picture the presupposition, but it can even contradict it. Contrast (4a) with (4b): (4b) Paul and Mary have a cat lJ. smashed the mirror in the

dining room trying to catch a mouse. In (4b), the anaphoric link cannot take place through the descriptive content of it because this pronoun has no descriptive content at all. Someone asserting (4b) 's second member, however, does presuppose that there is a cat, and not that there is an it. So, the presupposition is not the content of the anaphor. Now, contrast (4a) with (4c): (4c) Paul and Mary have a cat This tsunami smashed the

mirror in the dining room trying catch a mouse. In (4c), there is an anaphoric link between a cat and this tsunami. Clearly, this link carmot take place by virtue of the descriptive content of

this tsunami because it contradicts the one of a cat. However, it is equally clear that someone asserting (4c)'s second member presupposes the same exemplified by the different historical periods in which there have been different kings of France. According to Strawson's example, then, in all the imagined uses, the referent matches the meaning of the noun phrase. In this sense, Strawson's argument seems to tacitly assmne that the meaning of the nOM phrase predetermines a matrix of contexts of use. This assmnption is both unnecessary to the consistency of Strawson's analysis, and false. All that is required is that some referent is identified, independently from the fact that it fits the meaning of the referring expression (that is, for instance, independently from the fact that he is a king ofFrance). Indeed, the referent is identified independently from the (eventual) meaning of the referring expression.

Presupposition and anaphor

91

thing as someone asserting (4a) 's second member, namely that there is a cat and not that there is a tsunami. So, a fortiori, the presupposition is not the anaphor.

11.4 Presupposition of identification On the premise that a discursive presupposition does not coincide with the anaphoric or referring expression which carries it, consider the following examples: (7a) George has a beautiful sister. She is my wife. (7b) George ·s wife is approaching us. She is angry. (7c) George hired a new sales assistant. She is very efficient. In (7), two things should be sharply separated: firstly, the information that George has a sister and a wife and that he hired a sales assistant; secondly, the fact that there is some textual referent for the anaphoric expression she. The first point presents the contents of the discursive presuppositions triggered by she in (7a), (7b) and (7c): these contents are pieces of information and vary depending on the assertion at stake. The second point, instead, is shared by the referential use of she in (7a), (7b) and (7c) independently from the content of the discursive presupposition. Then, the fact that there is a textual referent for the pronoun she is not the discursive presupposition of a specific assertion, but a presupposition which constructs the text: it presides over the integrity of the surface of discourse. As regards this, Bonomi (1979) talks about "presupposition of identication". The presupposition of indentication does not preside over the real existence of something, but rather over the identification of a textual referent. Now, the distinction between the presence of a textual referent for a noun phrase and the fact that someone has a sister, a wife and so on----or that there is a cat, a king and so on-enables us to understand an important

92

Chapter 1 1

point. Since the fmmer constructs a text, from a textual point of view, it is super-ordered to the latter.

This is the reason why referential

presuppositions were the first kind of presuppositions studied, and the reason why they are onmipresent. This ubiquity has been clearly remarked by Austin (1962) and Ducrot (1972): There is a quite obvious difference between traditional presuppositions (such as the presupposition of existence), which are linked to any speech act and others which only concern a particular one. This could suggest that if they both fimction as conditions for use, they do not have this property by the same token. (Ducrot, 1972:49) What is really at stake, in this quotation, is not the existence of something, but the identification of a textual antecedent for a noun phrase. It is only because this latter can picture the former that they can be confused. But the very possibility of their superposition implies their theoretical separation. Here, we can see the looking-glass paradox at work (cf. Chapter Seven). On the one hand, the fact that a noun phrase has a referent

or

a

textual antecedent is a very contingent fact, tied to a specific utterance. As such, it is more contingent, for instance, than the fact that someone smoked or that France has a king. On the other hand, as regards that specific utterance, the identification of a referent or a textual antecedent becomes hierarchically prior in respect to all these facts. If the identification of an antecedent

fails,

namely,

if the

presupposition of identification fails, then there is no text at all: (8a) 71 saw a dog This chair was broken In (8a), there is no unitary text, but two independent utterances. If a discursive presupposition fails, there is instead a text, but this text is infelicitous: (8b) 7Pau/ never smoked Now, he has stopped

Presupposition and anaphor

93

The frustration of the presupposition, then, relies on the fact that the anaphor works. In other words, from a textual point of view, the presupposition of a trigger like to stop (doing something) is subordinated to the act of reference made through the anaphoric elements that it can contain. So, that presupposition carmot be identified with these anaphoric elements. The same holds, mutatis mutandis, for the relationship between contingent and ground presuppositions. A ground presupposition is much more fundamental than the presupposition triggered by to stop smoking. However, as regards a specific discourse, this trigger plays a much more important role. In (8b) a discursive presupposition fails: the utterance is infelicitous and carmot convey any message. The speech act is, so to speak, aborted. In (8e): (8e) I saw the moon.

It was smiling.

a ground presupposition fails: the utterance is conceptually inconsistent, but this does not prevent the transmission of a message. A speech act can still see the light of day. If the phenomenon of presuppositions was an iceberg and if discourse was the sea, then referential presuppositions would be the top of the iceberg, contingent presuppositions would the part of the iceberg just below the sea level, and ground presuppositions would be the vast submerged part.

PART III TOWARDS A STUDY OF GROUND PRESUPPOSITIONS

CHAPTER 1 2 How TO STUDY AN EXTRA-LINGUISTIC

PHENOMENON LINGUISTICALLY?

The looking-glass paradox, described in Chapter Seven, pictures an overturning, namely, the fact that language submits the consistency conditions of experience to its O\Vll internal constraints. This overturning has two potential outcomes. On the one hand, it can obscure experience's ground presuppositions and induce the mistaking of ephemeral, discursive presuppositions for prototypical ones. This is the negative interpretation which oriented tbe second part of this book. On the other hand, that very overturning is a resource because it can be exploited in order to violate­ and hence

reveal-ground

presuppositions.

This

is

the positive

interpretation of the looking-glass paradox which will orient the tbird part oftbe book. In this chapter we are concerned with the following problem. Ground presuppositions are a priori conditions of experience. As a consequence, they are extra-linguistic par excellence. But how can we linguistically study an extra-linguistic and language-independent phenomenon?

12.1 Exhibited predication Before being a technical concept elaborated by philosophers and linginsts, presupposition is a notion rooted in everyday life. Indeed, there is a common expression which depicts its functioning, namely, to treat

someone like a.. (stranger, slave, animal, thing, etc.). Imagine a wife saying to her husband: You treat me like a slave.l In such a situation, what the wife is complaining about is not that her husband has stated tbe proposition You are my slave, but rather that, through his behaviour, he

98

Chapter 12

seems to assume the truth of that proposition The truth of a proposition like This woman is a slave emerges from-is sho\Vll by-his very actions. In this sense, in the expression to treat someone like a ... , the telTIl filling the suspension points can be seen as the predicative nominal of a predication exhibited by someone's behaviour as regards someone else. This is what I mean by "exhibited predication". Ifwe extend this line of reasoning, all our actions, on closer inspection, exhibit some predication. If someone uses a cannon to shoot at an enemy outpost, in a sense, they treat the cannon as a weapon; hence, we can say that they exhibit a tautological predication like A cannon is a weapon. If someone puts flowers into a carmon, in a sense, they treat the cannon like a flower-pot: hence, they exhibit a conflicting predication such as A

cannon is a flower-pot (and not a weapon). If a priest tries to comfort a prisoner who is going to be executed, he treats the prisoner as a human being and exhibits a tautological predication like a prisoner is a human

being. If the same priest tries to comfort a fir tree which is going to be cut do'Wll, he treats the tree as a human being and exhibits a conflicting predication like A tree is a human being. And so on24. I maintain that a ground presupposition can be studied in the model of predications of the form X is a r exhibited by our actions. The subject X is the person or the thing which undergoes the action. The predicative nominal Y is the linguistic representation of the presupposition. The whole Note that the idea of an exhibited predication is somehow implicit in philosophical argumentation. Consider the following Heidegger passage: "The hydroelectric plant is not built into the Rhine River as was the old wooden bridge that joined bank with bank for hundreds of years. Rather the river is dammed up into the power plant. "What the river is now, namely, a water power supplier, derives from the essence of the power station. [ ... ] But, it will be replied, the Rhine is still a river in the landscape, is it not? Perhaps. But how? In no other way than as an object on call for inspection by a tom group ordered there by the vacation industry (1954 [1 977:7])'·. The hydroelectric plant, so to speak, 'treats' the Rhine River as a supply and hence exhibits a predication like a river is a supply. The essence of the technique is expressed (according to Heidegger) by the predicative nominal of that predication. Understanding a daily action or the phenomenological impact of a power plant on nature basically means clarifying the predicative nominals of some predications exhibited by them. 24

How to study an extra-linguistic phenomenon linguistically?

99

predication turns out to be tautological if the predicative nominal matches the shared common sense categorisation of the subject, otherwise it is conflicting.

12.2 A difficulty: experience is a priori consistent Three points must be stressed. Firstly, the exhibited predicative nominal is not the presupposition itself, but rather a convenient way to represent it: one must not confuse the object of our study (namely, tbe presupposition) with the means to study it (namely, the exhibited predicative nominal). Secondly, in the expression exhibited predicative nominal, the adjective exhibited is meant to highlight that the predicative nominal does not describe what one does (the content of tbe practice), but ratber what one shows in doing it. In the same sense, when we speak, we do not usually state the grammatical rules that we are in fact following, but we show tbem. Hence, the adjective exhibited does not imply tbat tbe predicative nominal at stake is self-evident; on the contrary, its description can be long and delicate, just like discovering the grammatical rules exhibited when we speak or \¥fite. Thirdly, ground presuppositions are best revealed not by practices exhibiting tautological predications, but rather by practices exhibiting conflicting ones. If one observes a priest comforting a prisoner, the idea that only human beings can be consistently comforted goes unnoticed, but if one observes a crazy priest comforting a fir tree, that idea leaps out. The last point confronts us with a difficulty (cf. Chapter Seven §7.3.). Ground

presuppositions

are

experience's

consistency

conditions.

Experience, in turn, is a priori consistent: it exhibits tautological predications by default. Hence, by looking at experience, it is plainly impossible to find the instances of conflicting exhibited predications which reveal ground presuppositions.

100

Chapter 1 2 A possible way out is the following. Since it is much easier to

construct sentences challenging ontological rules, that is, exhibiting conflicting predications, than to have experiences that do it, we could solve the problem by inquiring into ground presupposition indirectly, through language. In order to do this, we need to pass from a proposition like (A): (A) A rock is an entity with regard to which the practice oftorlure is

insane. to a proposition like (E): (E)

A rock is an entity, which-placed in the direct obiect position ofthe verb "to torture "-produces an inconsistent meaning.

But is this passage licit?

12.3 Sentence meaning as a model of experience's consistency My answer is yes. The transition from (A) to be (E) is granted by the fact that a practice and the structure of the sentence share the same exhibited predication. Consider the following sentence: ( l a) ?He comforted thefir tree. This sentence is conceptually inconsistent. If we make explicit the selection restrictions of the verb to comfort, sentence ( l a) can be analysed as follows: (lb) He comforted [which requires a human being as direct object]

the fir tree [which is a vegetable] . This analysis highlights that 'vegetable' is the category to which a

fir

tree naturally belongs, while 'human being' is the category imposed by the verb to comfort on its direct object. Now, from (lb), it is easy to obtain a conflicting predication where the fonner is the subject and the latter is the

How to study an extra-linguistic phenomenon linguistically?

10 1

predicative nominal: ( l c) A vegetable is a human being. So, in a sense, we can say that, in ( l a), the verb to comfort treats the following noun as a human-noun (through the connection verb-direct object) just like the crazy priest treats the

fir

tree like a human being

(through his behaviour). That is to say, an extra-linguistic action and a sentence's syntactic structure turn out to share the same exhibited predication identified by ( l c). If this is true, then the exhibited predication works as a bridge between an extra-linguistic practice and a sentence. In other words, the sentence's consistency conditions can be considered as a model of those of the extra­ linguistic practice.

12.4 Conceptual conflicts as access to ground presuppositions According to the previous conclusion, the sentence's core syntax can work as a model of experience's consistency conditions. If this is true, we are not obliged to look for crazy

or

impossible experiences, but we can simply

manipulate sentences in order to construct inconsistent complex meanings. This means that language is actually the best laboratory to test our hypothesis about experience's presuppositions, because these latter are, at one and the same time, reflected by the sentence's meaning and subdued in the way in which this meaning is coded. This is the reason behind the positive interpretation of the looking-glass paradox (cf. Chapter Seven §7.3). By core syntax, I mean, in English, the fOlTIlal links between subject and predicate, verb and direct object, verb and prepositional object, and verb and indirect (second) object. This network of grammatical relations is described by Prandi (2004) as a relational kind of coding, which is autonomous as regards all sorts of substantive content. The author opposes it to a punctual kind of coding, which, on the contrary, is functional to the expression of shared cognitive structures rooted in a shared natural

102

Chapter 1 2

ontology. Let me resume this distinction in his O\Vll words: Relational coding depends on the presence of an independent network of grammatical relations. In the presence of relational coding, a given phrase for instance, a nolUl phrase encodes a given role for instance, the agent not in isolation, but as a term of a grammatical relation typically as a subject or object. Punctual coding operates outside the network of grammatical relations. In the presence of punctual coding, a given expression is connected to a given role, not as a term of a grammatical relation, but in isolation. [ . . . ] In the area of plUlctual coding, the formal structure of the sentence is not autonomous, but is only justified insofar as it mirrors the structure of an independent complex concept. [. . .] Expressions belonging to the network of grammatical relations are ready to take content within the framework of a given process, but are as such devoid of inherent content and, above all, are identified irrespective of any content. [ . . . ] Expressions located outside the network of grammatical relations, on the contrary, are both defined, identified and connected to the syntactic structure of the sentence thanks to their association with a substantive role. (prandi 2004:61 62) In the case of a punctual kind of coding, syntax displays various degrees of iconicity as regards independent conceptual structures. 'When coding is relational, on the contrary, syntax constructs a process independently from the consistent natural relations of the concepts involved.

It

is precisely this kind of coding which offers the best access to

ground presuppositions. Now, apart from the technical details concerning the nature of coding, let me stress an epistemological point. In order for the study of ground presuppositions to be possible, it is not required that a language is iconic as regards natural ontology, but rather the opposite. 'What is required is that its syntax embeds at least an autonomous core from natural ontology and hence is capable of revealing it by constructing conflicting complex meanmgs.

How to study an extra-linguistic phenomenon linguistically?

103

Let me sum up. A ground presupposition can be construed as the predicative nominal of a nominal predication exhibited by a practice: that is

to

say, a category given to a person or a thing through a practice. The

best method to describe this category consists of exploring the conceptual conflicts stemming from inconsistent uses of predicates. In order to do this, it is useless to look directly at a proposition like Trees are not human beings; taken in isolation, such a proposition looks false like any other. On the contrary, one has to look at that proposition in its role as a presupposition, that is, indirectly, by examining sentences which exhibit it, for instance:

?He reassured the fir tree.

CHAPTER 1 3 LEXICON AND ONTOLOGY

The method suggested at the end of Chapter Twelve is not new.

It

goes

back to Ryle (1938) and Sommers (1963) which are based, in turn, on Husserl's insights25: [ . . . ] whenever a predicate P is significantly applicable to a thing, then so is its complement non-Po Now this gives us the right to treat predicates as having no 'sign' for the purpose of a type analysis. Thus, any predicate P can be construed as IPI or 'the absolute value' ofP, by which we mean that P spans the things that are either P or non-P but does not span things which are neither P nor non-Po [ . . ] I shall call a class defined by an absolute predicate an ontological class or category. [ . . . ] The category language is embedded in every natural language. (Sornrners 1963:159 160) In the example at the end of the previous chapter (He reassured the fir tree), P is to reassure and non-P is something like to frighten. Hence, Ita

reassure I includes humans, but not vegetables because vegetables can be neither reassured nor frightened. This reasoning depicts the flUlctioning of the very notion of presupposition, and the description of grolUld presuppositions roughly matches the description of Sommers' "category language". However, such an enterprise rlUlS into two main difficulties: polysemy and excessive specijicity. These difficulties, in turn, rely on two assumptions, namely, that lexicon should map natural ontology, and that there is no clear difference between an ontological class and a lexical one.

25 I go back to Westerhoff (2005:40 59) for a discussion.

106

Chapter 1 3 This chapter is devoted to showing how these assumptions are both

false. This will enable us to sketch a map of the relationship between lexicon and natural ontology.

13.1 The po\ysemy objection Let us begin with the polysemy objection. The polysemy objection points out that lexicon offers heterogeneous and inconsistent classes. A predicate like to nourish, for instance, applies to both animate beings and hopes. But one can reasonably argue that animate beings and hopes do not belong to the same category, and that hopes are not animate beings. Hence, lexicon seems to neutralise precisely the ontological distinctions we are interested in. This problem becomes particularly serious when taking into account the pervasive phenomenon of conceptual metaphors (cf. Lakoff and Iohnson 1980).

It

is true that Ryle (1938) stresses the importance of

zeugma; in I nourish my pigeons and my hopes, there are two different meanings of to nourish at stake. But the difficulty remains: because of polysemy and conceptual metaphors, the lexicon seems to largely contradict our ontological intuitions. The polysemy objection relies on the assumption that the lexicon should reproduce ontology: if the predicate to nourish consistently applied

only to living beings-if there was no polysemy-then that predicate would actually hint at natural ontology, but it does not. Note that, from this perspective, either lexicon reproduces ontology or contradicts it:

tertium non datur. In my opinion, this assumption is false. Indeed, there is a third possibility, namely, that lexicon presupposes ontology. Let us focus on conceptual metaphors: ( l a) He nourished the pigeon. (2a) He nourished the hope o/becoming president. The example (2a) is a manifestation of a conceptual metaphor which may be labelled "intentions ARE living beings". Now, let us consider:

Lexicon and ontology

107

( l b) That pigeon became too fat. (2b) ?That hope became too fat. Unlike (2a), the example (2b) is perceived as inconsistent. By passing from (2a) to (2b), then, we turn a conceptual metaphor into a living one. This means that (2b) exhibits a predication that we can picture as (2c): (2c) Intentions are living beings. If this is true, however, this very predication is not exhibited by (2a), which is consistent just like ( l a) and (lb). Surprisingly, then, the label of the conceptual metaphor is the contrary of its exhibited predication, and the verb

"ARE",

in that very label, is a

fake predication. This means that the consistent conceptual metaphor (2a) relies precisely on the idea that intentions (hopes) are not living beings. Hence, the polysemy of to nourish does not really challenge the ontological distinction between animate and inanimate beings, but is grounded on it. This distinction is so deep that the lexicon can extend some uses of predicates without affecting it. If this is true, polysemy and conceptual metaphors turn out to be the best proof not that the lexicon erases the ontological distinctions we are interested in, but rather that it relies on them: these latter provide a ground for its consistency a priori. Consistency, then, is not a property stemming from the lexicon, but from ground presuppositions laying below it; this is the reason why the lexicon can allow itself the luxury of polysemy by creating heterogeneous classes. This answers the polysemy objection, but it leaves a question open. If the lexicon plainly maps natural ontology, then studying ontology has a plain meaning, namely, looking for meanings immediately inconsistent. However, if the lexicon does not map ontology, but presupposes it, what does studying ontology mean? The fact that the lexicon does not passively reproduce ontology, but presupposes it, does not prevent our access to ontology, rather, it implies that this access must be construed.

108

Chapter 1 3 In the cognitive framework on metaphors, one describes a concept like

intention by exploring the range of consistent inferences allowed by the conceptual metaphor "intentions ARE living beings". Hence, for instance, if someone nourishes a hope, then this hope will grow. Doing this amounts to realising a sort of lexical cartography. The job of the cognitive perspective, however, ends here, at the limits traced by Lakoffs "invariance principle", before inconsistent inferences. This is where the study of ground presuppositions begins. This study consists of turning consistent conceptual metaphors into living inconsistent ones by exploring the breaking points of the allowed inferences: a hope, for instance, can be nourished, but it will not get fat and one will never need to put it on a diet. Doing this reveals the presuppositions which ground those very conceptual metaphors. So, the study of ground presuppositions is complementary as regards the cognitive framework: the latter aims to describe the different visions of the world embedded into the lexicons of natural languages, the former aims to describe the same world presupposed through all of those visions.

13,2 The specificity objection Let us now turn to the specificity objection (cf. Smart 1953). This objection points out that conflict leads to identifying both classes that are ontologically relevant and classes that are not. For example, if to smile selects human beings in subject position (because the moon smiles is odd), then to resole selects shoes in direct object position (because she resoled

her gloves is odd too). But shoes are not an ontological class. The specificity objection relies on the assumption that the oddity of

The moon smiles and the oddity of She resoled her shoes have the same nature. The idea is: if The moon smiles and She resoled her shoes were not equally odd, then we could distinguish ontologically relevant classes from those that are not. But, apparently, they are. Note that the specificity objection highlights a paradox of Ryle's and Sommers' approaches to ontology. Indeed, their keystone is the notion of

Lexicon and ontology

109

conflict, but this very notion is taken for granted and never directly analysed. This is shown by the fluctuating terminology:

'oddity',

'deviation', 'non significance' and so on. So, in order to answer the specificity objection, the crucial move is precisely to distinguish different kinds of conflicts26• Let us consider (3) and (4): (3)

?Paul has slaughtered afly.

(4)

?Paul has slaughtered the moon.

Sentence (3) is conflictual. Its conflict concerns the distinction between two kinds of animal: cattle vs. insect& Sentence (4) is conflictual too. Its conflict, however, does not concern classes of animals, but a more basic distinction: animate beings vs. inanimate ones. Asking if there is a relevant difference between an ontological category like animate beings and a specific one like cattle amounts to asking if the kind of conflict at stake in (3) and (4) is the same. Now, between (3) and (4) there are two differences concerning both the horizontal and vertical axes of lexicon. Let us begin with (3). Here, we face a lexical mistake: to slaughter is a verb appropriate to cattle. The appropriate verb for insects, instead, is to

crush. Horizontally, then, that mistake can be corrected by replacing to slaughter with to crush. From a vertical point of view, on the other hand, one can dissolve the conflict by replacing to slaughter with its hypemym to kill: the sentence Paul has killed the fly is consistent. In this sense, we can say that the conflict in (3)-which involves specific classes-can be fixed, either by correction or generalisation towards a hypernym. Let us turn to (4). Note that this time it would be pointless to talk about a lexical mistake. Horizontally, lexicon does not code an appropriate verb to correct to slaughter with. If we replace to slaughter with to crush, the result is inconsistent. If we replace to slaughter with, say, to destroy, we do not actually correct the sentence, but we interpret it. From a vertical 26 To my knowledge, the only author who tried to do something like this is Harrison (1965), and I believe that he is on the right track.

110

Chapter 1 3

point of view, the conflict carmot be dissolved either: the sentence Paul

killed the moon is inconsistent just like (4). So, the conflict in (4)-which involves ontologically relevant c1asseS-caIUlot actually be fixed, either by correction or by generalisation. As Prandi (2004) puts it, we face an "irreversible conflict". These remarks separate two kinds of conflict. Each kind of conflict identifies a different kind of class. On the one hand, there are lexical classes; the violation of a lexical class produces a conflict that can be fixed, and the sentence is perceived as an inappropriate use of a predicate. 'What detelTIlines an appropriate or inappropriate use of a predicate is lexicon. On the other hand, there are natural ontological classes: the violation of an ontologically relevant class produces a conflict that cannot be fixed, and the sentence is perceived as an inconsistent complex meaning. What legislates over consistent and inconsistent uses of predicates is not lexicon, but ratber natural ontology.

13,3 The freedom of the lexicon The discussion of the polysemy and the specificity objections highlight that lexicon presupposes natural ontology, and that consistency is determined by tbis latter and not by the lexicon itself. Since consistency is grounded in natural ontology, and since natural ontology lies below lexicon, lexicon carmot be inconsistent, or, as Prandi (2017) points out, is consistent a priori. This, in turn, implies that the lexicon is free as regards consistency: it can play over consistency without challenging it. More precisely, as regards the relationship between lexicon and natural ontology, we can sketch the following map.

Lexicon and ontology (i)

111

There are predicates that simply match an ontological class: to smile, to suffer, to kill and so on. Here, inconsistency gives us direct access to natural ontology: this is the case envisaged by Ryle's and Sommers' original accounts.

(ii)

There are predicates that include different ontological classes: to caress animate things and dreams, to nourish living beings and hopes, to take drugs, means of transport and roads and so on. Here, we face polysemy. In this case, in order to reconstruct natural ontology, one has to produce inconsistency as suggested at the end of § 1 3 .1 .

(iii)

There are predicates that carve out a lexical class inside the borders of an ontological one which work as a ground: for example, to assassinate political figures, to slaughter cattle inside to kill living beings. When we realise that he slaughtered

the fly is inappropriate, we are pointing out a category (namely, cattle) which is internal to English lexicon and depends on the existence of a specific verb. This category is not an ontological one. The ontologically relevant category, instead, is living being: this latter carries out the function of offering the ground where lexicon can code the previous one. As seen in § 13.2, the violation of a lexical category results in an inappropriate use of a predicate stemming from a conflict that can be fixed; the violation of an ontological category, on the contrary, results in an inconsistent use of a predicate stemming from a conflict that cannot be fixed. Let

us

contrast points (ii) and (iii). Polysemy-a predicate cutting

through different ontological categories---