Seduction, Community, Speech
Pragmatics & Beyond New Series Editor Andreas H. Jucker University of Zurich, English Department Plattenstrasse 47, CH-8032 Zurich, Switzerland e-mail:
[email protected]
Associate Editors Jacob L. Mey University of Southern Denmark
Herman Parret Belgian National Science Foundation, Universities of Louvain and Antwerp
Jef Verschueren Belgian National Science Foundation, University of Antwerp
Editorial Board Shoshana Blum-Kulka Hebrew University of Jerusalem
Catherine Kerbrat-Orecchioni University of Lyon 2
Jean Caron Université de Poitiers
Claudia de Lemos University of Campinas, Brazil
Robyn Carston University College London
Marina Sbisà University of Trieste
Bruce Fraser Boston University
Emanuel Schegloff University of California at Los Angeles
Thorstein Fretheim University of Trondheim
Deborah Schiffrin Georgetown University
John Heritage University of California at Los Angeles
Paul O. Takahara Kansai Gaidai University
Susan Herring University of Texas at Arlington
Sandra Thompson University of California at Santa Barbara
Masako K. Hiraga St.Paul’s (Rikkyo) University
Teun A. Van Dijk Pompeu Fabra, Barcelona
David Holdcroft University of Leeds
Richard J. Watts University of Berne
Sachiko Ide Japan Women’s University
Volume 127 Seduction, Community, Speech: A Festschrift for Herman Parret Edited by Frank Brisard, Michael Meeuwis and Bart Vandenabeele
Seduction, Community, Speech A Festschrift for Herman Parret
Edited by
Frank Brisard University of Antwerp
Michael Meeuwis Bart Vandenabeele Ghent University
John Benjamins Publishing Company Amsterdam/Philadelphia
8
TM
The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.
Library of Congress Cataloging-in-Publication Data Seduction, community, speech : a festschrift for Herman Parret / edited by Frank Brisard, Michael Meeuwis and Bart Vandenabeele. p. cm. (Pragmatics & Beyond, New Series, issn 0922-842X ; v. 127) Includes bibliographical references and index. 1. Pragmatics. 2. Language and languages--Philosophy. I. Brisard, Frank. II. Meeuwis, Michael. III. Vandenabeele, Bart. IV. Parret, Herman. V. Pragmatics & beyond ; new ser. 127. P99.4.P72S43 2004 306.44-dc22 isbn 90 272 5370 6 (Eur.) / 1 58811 557 7 (US) (Hb; alk. paper)
2004057692
© 2004 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa
Table of contents
Introduction: Mind the gap: Pragmatics and cognition today Frank Brisard
1
Selected books and articles by Herman Parret in philosophy of language and pragmatics
19
Part I Pragmatics (standard and not so standard)
23
Ordinary time Frank Brisard
25
Performatives as constatives vs. declarations: Some recent issues Robert M. Harnish
43
‘First/second vs. third person’ and ‘ªrst vs. second/third person’: Two types of ‘linguistic subjectivity’ Yoshihiko Ikegami
61
De la nécessité de prendre en compte la dimension praxéologique à tous les niveaux de l’organisation des discours Eddy Roulet
75
From Optative and Subjunctive to Irrealis Johan van der Auwera & Ewa Schalley
87
Quand dire, c’est “faire rire aux dépens”: Notes sur l’ironie Maurits Van Overbeke
97
Part II … & beyond: Art, mind, and community Pragmatics and evolution David Holdcroft
115 117
vi
Table of contents
Expressive language games Antonio J. Marques Language as pragmatics: Studying meaning with simulated language games Domenico Parisi
129
139
Sharing… But why a language or world? Bart Vandenabeele
151
Identity as denial of diversity Jef Verschueren
171
Conceptual innovation in art: Three case studies on Leonardo da Vinci, William Turner, and Henry Moore Wolfgang Wildgen
183
Index
197
Mind the gap Pragmatics and cognition today* Frank Brisard University of Antwerp
1.
Introduction
Paul Grice, in a fatherly moment, warns us “against the devil of scientism, who would lead us into myopic over-concentration on the nature and importance of knowledge” (Grice 1991: 161). On a more fabulous note, he adds the following little tale. Of course, it is taken entirely out of context here: The very eminent and very dedicated neurophysiologist speaks to his wife. ‘My (for at least a little while longer) dear,’ he says, ‘I have long thought of myself as an acute and well-informed interpreter of your actions and behaviour. I think I have been able to identify nearly every thought that has made you smile and nearly every desire that has moved you to act. My researches, however, have made such progress that I shall no longer need to understand you in this way. Instead I shall be in a position, with the aid of instruments which I shall attach to you, to assign to each bodily movement which you make in acting a speciªc antecedent condition in your cortex. No longer shall I need to concern myself with your so-called thoughts and feelings. In the meantime, perhaps you would have dinner with me tonight. I trust that you will not resist if I bring along some apparatus to help me to determine, as quickly as possible, the physiological idiosyncracies which obtain in your system.’ I have a feeling that the lady might refuse the proŸered invitation. (ibid.)
* This introduction was delivered as a speech at a colloquium in honor of Herman Parret’s academic retirement, held on April 23rd, 2004, at the Higher Institute of Philosophy, University of Leuven (Belgium). Numbers between square brackets appearing in the references in the text refer to the numbering in the selective bibliography of Parret, immediately following this introduction.
2
Frank Brisard
This quote is, in some important sense, not very representative of Grice’s general stance in matters of science and philosophy. Grice is not particularly known to have engaged in any explicit polemics against contemporary academic practices. If anything, in the work inspired by his conversationalist hypothesis, he was bent on reconciling, and thus salvaging, traditions (of ordinary-language philosophy and formal logic, for instance) more than on seceding them and/or defeating one or the other, and in this his only fear, perhaps partly instilled by what seemed to be going on around him at the time, was that he should become overly sophisticated in applying his own technical machinery (as remarked on one occasion by Hilary Putnam). Still, the philosophy of Grice, which is constructive more than it is destructive or “critical”, might simultaneously serve as a reliable baseline, to which one may return time and again, for anyone who wishes to understand and evaluate newly established insights in the ªeld of pragmatics. I submit that something very similar appears to hold for Herman Parret’s oeuvre, which oŸers substantive proposals in the epistemics of verbal communication and discourse yet at the same time functions as an elaborate, meta-pragmatic comment on the ideological premises and implications of extant analyses in pragmatics. There is little chance that Parret’s explorations in philosophical and linguistic pragmatics will be remembered for their excessive formalizing. If there is one constant in Parret’s critical appreciation of much of the work that has been done in empirical pragmatics over the past thirty years or so, it is exactly the intuition that too much formalism, in modeling linguistic interaction or in any other related type of analytic endeavor, will lead to a depersonalized, disembodied view of symbolic behavior. In particular, the emphasis, in the course of 20th-century linguistic history, on matters of propositional knowledge and attitudes, ascribed to an idealized, individualized agent acting in complete sovereignty, tends to reduce the full potential of human abilities (what we know, will, can, and must) to one, rather poorly motivated, modality, that of beliefs, which for some reason is awarded the highest position in the hierarchy of mankind’s distinctive skills (Parret 1976a [30], 1983a [47]).1 This, now, reeks of objectivism and essentialism, two qualities that can reasonably be ascribed to certain variants of Cartesian rationalism, including within linguistic thinking. Parret’s answer to such reductionism is, naturally, to embrace the tension that may arise between (accounts of) theoretical and practical — nonpropositional, volitional — knowledge, rather than negate it, and thus “put man back into language” (or subjectivity into discourse). Indeed, many of the more prominent dichotomies that have been posited in modern linguistics
Mind the gap: Pragmatics and cognition today
(and semiotics) may be said to refer to a dual structure in the human faculty of reason(ing), as worked out in Kant’s ªrst two Critiques. This tension typically manifests itself in attempts to ªgure out the relation between language and cognition, between linguistic praxis and its corollary mental reality, and thus, in the end, between linguistics and psychology, as two intimately linked disciplines. The “gap” in the title of the present paper relates to this very problem of which is to constitute which, in the study of meaning and mind. It serves to remind us of the Kantian Klufte that stand between purposeful action and the beliefs that are thought to underlie and justify it (or, more precisely, between alethic and practical reasons for good behavior). Any review of Parret’s contributions to an “integrated” pragmatics will have to pay attention to his peculiar, original views of language use as calculated, strategic action, fueled by the passions and community-building, because this is obviously where his heart-felt philosophical interests lie. Still, it is interesting in its own right to consider which private demons Parret has also sensed the need to exorcize throughout his career, if only because that might provide us in turn with some tools with which to take up the fresh challenges stemming from recent scholarly developments, especially in the so-called cognitive sciences and philosophy of mind. So now I suggest we do both. I will not refrain, in what follows, from blending my own voice, as a sort of backing vocal, with Herman’s, for such is the nature, I feel, of proper paraphrase.
2.
Truth
In the work of Parret, the demon that we should be ever watchful against is not so much that of scientism, in all of its epistemological conceit, but more speciªcally that of truth. Once more, we ªnd ourselves in the company of those Oxford philosophers who, like John Austin, were more concerned with the act of telling the truth, than with an abstract conception of truth itself. Austin’s (1962) speech-act theory, in fact, with its e¹cacious distinction between locutionary and illocutionary acts,2 did not only manage to shift the focus away from a context-free, semantic discussion of linguistic meaning towards an action-centered approach that can truly be called pragmatist, in the (Meadean) sense of the speaking self as a psychosocial actor. It also relates the truth-saying goal of canonical “constatives”, whose job it is to report knowledge, to a broader notion of successful felicitation (Austin’s “happiness”) that accommodates the many alternative performative functions of language
3
4
Frank Brisard
directly. In the case of constatives, we may attribute an illocutionary force that is to be equated with a belief attitude, also implying a nontrivial practical commitment on the part of the speaker. It is only with Searle’s subsequent elaborations that speech-act theory has been progressively re-semanticized, so to speak, at least according to such notable critics as Habermas, because there the relevant object of analysis is yet again exclusively conªned to the domain of propositional truth, and discursive activity reduced to information transfer. Austin, in contrast, previously saw that without some “maxim” of truth which rational speakers and hearers are always already supposed to adhere to, there can be no feeling of trust among interlocutors (quite necessary, one would think, for forming any kind of belief), so that no real speech community can be imagined, either. Accordingly, truth, not as a referential relation to reality and other possible worlds but as an interactional given, goes a long way towards founding the communicative power of utterances, much more so, in any case, than logic. “Veridiction”, as envisioned in Greimas’ structural, neo-Hjelmslevian semiotics (cf. Greimas & Courtés 1979; see also Parret 1978 [35], 1983b [12], 1989 [56]), forms a conceptual square uniting the poles of (non)being and (non)appearing. Truth, then, is the positive conjunction of being and appearing, the perfect opposite of falseness. The illusion of reference, i.e., an utterance’s relation to an object, is eŸectively replaced here by a methodological focus on communication, in which sense eŸects are studied that derive from a heterogeneous ensemble of modal and illocutionary procedures. Insofar as such a semiotics succeeds in reconstructing the dynamics of understanding, its conditions of production and reception, and in confronting the creativity of discourse, it transcends the limitations inherent to the older structuralist approaches. However, to bring down this most metaphysical of debates, on the notion of truth, to the level of “enunciated discourse” and claim that truthtelling, in the immanence of that discourse, is solely a function of the speaker’s believing something to be true or false, must look like an extreme act of “futilitarianism” to anyone familiar (and comfortable) with the classical traditions in philosophical methodology, say from Aristotle to Kant. For now there are no more referents external to language, and there is no longer any objectivity that may be universally validated, and if it turns out that one is not concerned with the nature of reality, the proper subject of philosophy, then surely one must be dealing with its mere representation or appearance, or so it might be argued. But if nature as we experience it is not identical to something like an objective, purely physical and causal reality,3 then it is indeed legitimate
Mind the gap: Pragmatics and cognition today
to concentrate on the structure of that world which we take for granted, the common-sense world we routinely live in. Within this world, the transmission of what holds true, i.e., of our own certainties, is positioned as the central problem of semiosis, and this is something going on between two actors: a speaker/enunciator, engaged in “making-to-know”, and a hearer/enunciatee, who “receives-as-true”. The speaker’s “persuasive doing” does thus not just involve a representation of what is the case, but rather an active intervention on the belief system of the hearer, whose “interpretive doing” bears the same marks of agentivity, assessing and summoning under which conditions she is willing to accept a given interpretation. This structure leads to the establishment of a set of cooperative principles on which interlocutors must depend, or, in Greimas’ words, on a “ªduciary contract” in which a true belief is shaped by having interpretive coincide or accord with persuasive doing. Persuasion, then, is deªned as an activity by which one makes believe that what appears, really is, and interpretation basically concerns a similar movement from appearing to being, via inference. Lying (Parret 1979 [37]) is the conjunction of appearing and nonbeing. The lie is a “discursive mask” (Parret 1994a [25]: 228), in that the speaker lies when she knows one thing (an epistemic proposition) and consciously communicates another — as with St. Augustine’s and St. Thomas’s deªnitions of lying in terms of a discursive sequence contra mentem ‘against the mind’. Discovering the intention to lie therefore seems crucial in any attempt to unmask the liar, as the mere recognition of a falsehood in itself does not su¹ce to accuse someone of deception: they may have simply produced a metaphor or used irony, in which cases a false proposition is uttered with due emphasis on the blatancy of its contradicting an obvious, and presumptively shared, belief. Lying, unlike ªgurative speech, is an act that confronts the trust placed in the speaker, and it certainly does not lead to new knowledge in a hearer who does not recognize the lie. This connection between beliefs and the feeling of trust that they require also challenges the Kantian distinction, rigidly taken, between theoretical reason (addressing judgments) and applied ethics (set in the modality of desire). Now, it might be well motivated to argue that we should always be truthful, but if we must tell a lie, we should nevertheless implicitly be able to justify it, by providing (possibly deviate) reasons that retain some measure of intersubjective validity. Depending on which reasons work better than others to justify a lie, we can then assume a scale showing the “seriousness” of its mendacity, a typology which any truth-functionalist would deny in principle (Parret 1987 [49]). Lying, of course, is a kind of manipulation
5
6
Frank Brisard
as well and allows us a way into the strategic reasons that give form to symbolic interaction throughout (cf. GoŸman 1969). But lying is no less innocent a form of manipulation than other, perhaps less spectacular or consequential variants. For Parret, even the most ordinary strategies are in fact always “polemic” (Parret 1990 [57]: 2464), and above all linked to a politics of power: if polemos lies at the heart of human interaction, then contractual paciªcation is a continuous necessity. As such, these strategies approximate, if not too closely, the Wittgensteinian conception of “rules”, which are necessarily social because they exist on the basis of the collective (symbolic) sanctioning that can be seen to go on inside a community. Rules are thereby not to be viewed as conditions of production, on the part of any speaker, but as strategies of “understandingwith-right” exhibited in dialogic communion. Considering various types of linguistic manipulation, not all of which, incidentally, need to involve false propositions, we also fairly quickly seem to reach the limits of an intention-based, propositional approach to discourse, even if this is elaborated in a post-Austinian frame that pays ªtting attention to language in/as action. In Greimas’ semiotics, seduction is a subtype of manipulation, next to provocation or intimidation — which is not to be confused with the sort of strategic action described by Habermas (1998: 220Ÿ.), abandoning implicit claims to a cooperative achievement of consensus (and thus eŸectively replacing the use of language to motivate others’ beliefs with the direct exertion of nonlinguistic in¶uence). Strictly speaking, manipulation occurs in diŸerent forms of calculated linguistic behavior but is itself, as an act parasitic on normal language use, still very much inscribed in the assumption that language is being deployed with a built-in orientation toward reaching understanding. This makes much of linguistic manipulation part of a communicative intention, even if this intention may get partly “neutralized”. With seduction, however, we are no longer necessarily concerned with the speaker’s will (to produce “seduction”), or with the hearer’s intentionality to act upon an utterance produced by the speaker, because “the intentional criterion pertains neither to the subject who seduces nor to the subject who is being seduced” (Parret 1994a [25]: 229; emphasis in the original). Instead of jeopardizing the contract of symbolic interaction and thereby virtually returning to an original polemics, as with latent forms of manipulation, the act of seducing someone into an ensuing act can be avowed without losing any of its intrinsic tension. And this is exactly where it diŸers most clearly from those pragmatic phenomena that have traditionally been treated in terms of indirect speech acts (or indirection; cf. Searle 1975): say, directing someone to do something by way of informing
Mind the gap: Pragmatics and cognition today
about that person’s ability to do something (Can you pass me the salt?). In a genuinely judicial-contractual spirit, Austin’s classic answer to the latter set of problems was to point to the essential sincerity condition underlying such illocutionary acts, so that “infelicities” of discourse could be related to the insincerity or disloyalty of a speaker’s intentions, beliefs, and emotions. Instead of introducing infelicity into communication, however, the true seducer might as well be said to deceive (de¶ect, enrapture) without attributing to her some generative competence, as the bearer of an “action plan” that is to be successfully carried out. Rather, “to seduce is to ‘capture the soul’, to be seduced is the pleasure that accompanies rigid subjection” (Parret 1994a [25]: 223), up to the point where the seductive subject, the speaker-hearer, in fact becomes an object. This feeling of subjection and the pleasure accompanying it hint at the essential role played by the passions in the manifestation of purposeful action, symbolic or not (Parret 1986 [48]).
3.
Meaning and understanding
Truth, as something we normally, and normatively, aspire, rightfully occupies a central position in nearly any theory of nonnatural meaning (or, the meaning of natural languages). It is unfortunate, though, that it is the logicist version of truth, as relying on a formal correspondence or agreement with reality, that ªgures in many available theories.4 Correspondingly, there exists no need in this paradigm for the concept of a rational being or practical “reasoner”, mediating between epistemic propositions and what they are supposed to refer to. What remains most unfortunate is the hypostasizing eŸect this choice has on the epistemological orientations of the styles of formal semantics that have found their way into linguistics. A radically autonomous, speaker-oriented conception of semantics has dominated most of the debates on the so-called semantics/pragmatics interface, and has actually triggered and consolidated the distinction between semantics and pragmatics in the ªrst place. Any maximalist notion of pragmatics is thereby ruled out beforehand. Preferably, historical attempts at establishing truly uniªed theories of meaning that go against this secession, or at proposing pragmatics as a perspective rather than a separate object of study, are ignored, brie¶y and ªrmly dismissed (see, e.g., Turner 1999 on Wittgenstein), or even ridiculed, as with Levinson’s (2000: 192–193, 243) presentation of a cognitive “pragmantics”.
7
8
Frank Brisard
For linguists, there are quite big issues at stake here. For example, are we to take the truth-values that ¶ourish in formal semantics and pragmatics in a strictly mathematical sense only, so that we can start doing pragmatics from the relatively straightforward conviction that what we do complements the work of our formal-semantic colleagues (the wastebasket notion of pragmatics as an indexical semantics; cf. Montague 1974)? Or do we take the value in truth-values seriously and suspect an ethical component to truth-telling? Here, we may yet again call upon Grice (not the old, but the fairly new) to provide us with relevant reference points. First, Grice, reconciliatory as ever, still comes close to Austin’s own (nonformalist) preoccupations when he tackles the relation between a correspondence logic and natural language: “In my eyes the most promising line of answer lies in building up a theory which will enable one to distinguish between the case in which an utterance is inappropriate because it is false or fails to be true, or more generally fails to correspond with the world in some favored way, and the case in which it is inappropriate for reasons of a diŸerent kind.” (Grice 1989: 4) Next, in The Conception of Value, Grice (1991) observes that we have the ability to communicate our meaning intentions, such that beliefs are produced in others that may help us meet our own goals. This is an everyday psychological fact, he states. We also expect, as a matter of fact, this ability to be present in others as part of their design, implying that it is always already there and constitutionally immune from modiªcation. These combined facts, now, are not unqualiªed, empirically induced generalizations, but transcendental constructs which we need to assume if we wish to proceed humanly. Grice addresses such structures in view of his constructivist defense of objective value, i.e., of value as part and parcel of the world we inhabit yet none the less constructed by us (and none the less legitimate and irreducible for it). The principle of Humean Projection is used in this respect to explain the growth of conceptualization and representation in progressively more complex creatures. Humean projection, or the propensity of the mind “to spread itself on external objects” (see Baker’s introduction in Grice 1991: 4), is thus a crucial epistemological factor in the genesis of purposeful, reasonable conduct. Any value judgment has a history, or a phenomenology, that starts oŸ with a general mode of “thinking-valuable” (i.e., an attitude). Hume attributed to the human mind a generalized disposition to take the world to be a certain way, projecting into it items which properly considered are just features of our own states of mind:
Mind the gap: Pragmatics and cognition today
‘Tis a common observation, that the mind has a great propensity to spread itself on external objects, and to conjoin with them any internal impressions, which they occasion, and which always make their appearance at the same time that these objects discover themselves to the senses. (A Treatise of Human Nature, Book One, Part III, Sect. xiv) [Reason] discovers objects as they really stand in nature, without addition or diminution; [taste] has a productive faculty, and gilding or staining all natural objects with the colours, borrowed from internal sentiment, raises in a manner a new creation. (Enquiry Concerning Human Understanding, Appendix 1)
Grice is no empiricist, and so what Hume himself saw as an important source of illusion and confusion is reinterpreted in terms of a perfectly legitimate procedure to routinely make sense of the world. Since such procedures do not depend on an individual consciously inventing or exploiting them, they can be taken to carry some of the features that we shall have to resort to in explaining language use: consensus, via a certain commonality of conception, and nonarbitrariness (or internal motivation), for the conceptualization is not a function of private will or whim but of the universal structure of linguistic action. Thus, notions from philosophical psychology are mobilized as theoretical concepts that explain human behavior, justifying the privileged status of a number of items that present themselves as natural preconceptions in the human quest for understanding. Such items will be psychological in substance, but not in any cognitivist or mentalist sense, since understanding (as an ability rather than a competence) is a practice-in-the-world and not an action of the inner life on inner-life ontologies. Consequently, a psychological approach to meaning forces itself upon us that does not claim the autonomy of semantics. More speciªcally, the psychological theory envisioned by Grice contains provisions for a feeling of interest (Interesse, as per Kant) involved in the ascription of mental states to other minds (“mind-reading”), and such interests betray a concern for those others. Accordingly, a truly rational person is not some neutral observer of external goings-on, but a “passionate” participant with a set of strong emotional motivations (also aŸecting communicative interaction) which that person legitimately regards as self-justiªed. One of these, the overall assumption that people use language to refer truthfully to real states of aŸairs, may serve as the basis of semantics, which then becomes the professionalized account of an ordinary principle that might be dubbed “charity” (Davidson 1984). In Parret (1994b [61]), Davidson’s term is interpreted as a linguistic manifestation of the Humean aŸect of curiosity, which is the love of truth. In the process of understanding, the
9
10
Frank Brisard
human economy of passions, which we project unto the world, is regulated by this and related principles, collectively inspiring our trust in the speaker. Against this backdrop, the project of grounding theories of truth and meaning invites an investigation of the underlying pathos that gives rise to meaningful and responsible behavior (and its interpretation). Parret remarks that Hume compares philosophy itself to hunting or playing, experiences which, through associated qualities like uncertainty, di¹culty, vivacity, and the demand for attention management, engender the kind of pleasure that is indispensable to the development of intellectual curiosity, and which are, as such, the prime manifestations of the drive and desires that mark our every activity. In this, speaking and reasoning subjects aŸectively motivated are seized in their global abilities, rising above the restrictive appeal to epistemic contents and the veriªcation of knowledge. And if all beliefs are tainted by feelings, it seems plausible to presume as well that all facts are smudged by values. This is what can be called “ideology”. Truth only has value if there is a subjective investment. Truth value only exists through the economy of an interpretive, “erotetic” modality governed by deductive pragmatic principles. The erotetic modality of questions (or better, of question-answer sequences; Parret 1981 [42], 1988 [19]) refers to the desire or will to know (the truth, and thus value it) and aŸects all cognitive activity. It is, moreover, fundamentally (inter)actional, insofar as it installs “a dialectics of imposition and acceptance, assessment and evaluation: [even] questions for information are requests (or questions for action)” (Parret 1988 [19]: 299; emphasis in the original). Next, the deductive principles typical of (meta)pragmatic analysis (Parret 1976c [33]) come on top of the arbitrary and conventional rules of syntactic distribution and of the constitutive regularities that are to be found in the semantic analysis of speech-act types (i.e., of their conditions of satisfaction), in that these principles do not only address the success but also the possibility of communicative interaction. Independently of any empirical constraints and in line with Grice’s neo-Kantian constructivism, deductive principles or maxims are anticipated, rather than inferred, by interlocutors on the basis of their mutual knowledge of communicative acts. As “primitives”, they provide the necessary psychological basis informing the actional structure of discursive sequences. This a-priori “logic of communicability”, then, is intersubjectively assumed, whereby knowledge of other people’s knowledge of rules of conduct (and how to conform to them) immediately gives us the reason why we, and others, should and actually do
Mind the gap: Pragmatics and cognition today
conform to the same rules. It concerns a contextualized meaning relation spanning at least three dimensions: the acceptance of a signiªer designated by an utterance, of a discursive link between members of a speech community, and a criterion of generality prompting the expectation that the speaker is motivated in speaking by her desire to belong to such a community. At this third level, we discern principles of conversational cooperation (Grice), of charity (Davidson), and of humanity (Quine), all of which rest on the assumption that members of a speech community dispose of similar internal structures of belief and desire. This represents a common-sense ethical stance founded through anthropological analysis. It is because language users — speakers as well as hearers — rely on Humean Projection that they may presuppose in each other, as participants in discourse, the workings of a responsible psychology geared towards veridiction, where telling the truth equals valuing the truth and truth-value is a common good motivating the “person” in toto. That is not to say that we always expect the speaker to tell the truth, or even that we expect her to have the intention of telling the truth, but we do always already suppose that veridiction constitutes the ªnal end or norm for a speaker’s utterance. Understanding both creates its context and is, in speciªc circumstances, determined by various types of context (Parret 1985 [15]). Meaning as understanding, so we have learned in the meantime, can therefore not be studied suitably out of context. And this context cannot in turn be reduced to the indexing of place, time, and person, nor, for that matter, to a set of propositions, no matter how extensive, providing information needed for the calculation of a discourse’s possible inferences. Instead, we need to see that “one’s knowledge of the truth/falsity conditions of an expression lies in one’s capacity to recognize contexts as contexts for the expression, or, more precisely, lies in the ‘interpretation’ of contexts” (Parret 1984 [14]: 587). The interpretation of context boils down to the recognition of relevant individuals, states, and events, “knowledge” that cannot exist except in the practice of interpretation. In the wake of Wittgenstein, “contextualization” is thus portrayed as the property of discourse that is most forcefully constitutive of it. Rather than viewing context as an attachment to linguistic utterances, itself extralinguistic or formulated in the language of thought, real contextuality inheres in every part of discourse, at all conceivable levels of organization.
11
12
Frank Brisard
4.
Community in mind
The nature of the relation between linguistics and psychology is hugely aŸected by such considerations as presented above. Consider, in this regard, a very clear and concise rendition of Parret’s broadly anti-representationalist position: ‘Protè Philosophia’ is epistemology, which constrains psychology, in its turn constraining ontology, just as systems of expressions constrain mental states, and mental states constrain the ontology of the world-for-us. Another mistake, then, would be to look at epistemology as if it were psychology, or to look at knowledge as if knowledge were a psychological state. Epistemology is ‘First Philosophy’, but only in the sense in which Charles Sanders Peirce understood epistemology: as semiotics. (Parret 1982 [11]: 140)
It is, in other words, signs, words, and sentences that determine what counts as a psychological state, and there is no meaningful internal life in the absence of some content that can be either described or expressed, i.e., manifested externally. Paraphrasing La Rochefoucauld’s views on love, one might say that people would not experience anything as people (like feeling happy, for instance) if they never heard or talked about it. We have observed that the branch of psychology that is bound to lead us into the more promising venues of semiotic or conceptual analysis is not of necessity that of scientiªc, experimental psycholinguistics. What we are looking for in a philosophically sound account of meaning and understanding need not separate the facts from their values, as in good science. Not if it turns out that to characterize human reasoning, as involved in the course of interpretation, is immediately to characterize good reasoning, in which every belief, every bit or segment of thought or vocabulary is impregnated with a value. This is so because we have a tendency to attribute purposes to the phenomena we try to read. We turn to Grice for one last piece of (Aristotelian) wisdom: It does not require very sharp eyes, but only the willingness to use the eyes one has, to see that our speech and thought are permeated with the notion of purpose; to say what a certain kind of thing is is only too frequently partly to say what it is for. This feature applies to our talk and thought of, for example, ships, shoes, sealing wax, and kings; and, possibly and perhaps most excitingly, it extends even to cabbages. I suspect that it applies to items which are related, in one or another of a variety of ways, to the notion of life (including that notion itself). (Grice 2001: 35; emphasis in the original)
This tendency, including in discourse, serves to enrich our understanding by the application of forms of transition (reasoning) from a set of acceptances to a
Mind the gap: Pragmatics and cognition today
further acceptance, so as to assure the transmission of values, from any given premises to an inferred conclusion. Truth may be one such value, but not the only one, as there is also practical value, or plain goodness, to be earned. The teleology that marks Grice’s phrasing here may be construed as vehemently, if only implicitly, opposing any naïve form of nativism (“language as a natural product”), in that it views the acquisition of linguistic “re¶exes” as guided by the need to communicate, and to communicate well. Communication then precedes expression or representation, against what is claimed in a standard speech-act account à la Searle, because it is essentially oriented towards the other, as equally noticed by Marty in his Untersuchungen zur Grundlegung der allgemeinen Grammatik und Sprachphilosophie from 1908 (cited in Parret 1976b [31]: 756; emphases in the original): [D]ie Kundgabe des eigenen psychischen Lebens ist nicht das Einzige und nicht das Primäre, was beim absichtlichen Sprechen intendiert ist. Das primär Beabsichtigte ist vielmehr eine gewisse Beein¶ussung oder Beherrschung des fremden Seelenlebens im Hörenden. Absichtliches Sprechen ist eine besondere Art des Handelns, dessen eigentliches Endziel ist, in anderen Wesen gewisse psychische Phänomene hervorzurufen. ‘The expression of one’s own psychology is not the only thing, and not the primary thing, that is intended in speaking purposely. What is primarily envisioned is rather a certain in¶uence or control over the unfamiliar inner life in the hearer. Purposeful speaking is a special kind of acting, whose proper end involves the evocation of psychological phenomena in other beings.’
Let us brie¶y go back to the early days of linguistic pragmatics, when much attention was paid to the issue of indirect meaning and its possible typologies (“cutting up the semiotic pie”, as it were). Grice himself quickly noticed that pragmatic implicatures, conventional or conversational, hide a psychological structure that justiªes them. (At the time, putting emphasis on the hidden aspect of this structure resonated well with concurrent talk about “deep grammars” and other profound levels of linguistic meaning.) Now this structure, i.e., the psychology of rational subjects, needs to be justiªed in its own right, based on its inherent valuation. This takes us to a second dimension, called “metaphysical” by Grice, where it appears that human beings, expressly conceived as biological organisms, attach value to the formal idea of ªnality (without end). This telos represents an absolute value, directly associated with the organism’s will to survive, and reproduce — Kant might have termed this “enthusiasm”, or the passion for passion. All other values “derive” from this original conªguration, which is relevant in everything we do, say, or think, not
13
14
Frank Brisard
so much as an ethical imperative but as a (pre-)rational demand issuing from life itself. In relation to truth and truth-telling, “[l]a Vie nous force à valoriser l’acte de dire la vérité et surtout, la Vie nous impose la charité: interpréter le discours d’un co-sujet, comprendre l’autre et communiquer avec lui, est une quête charitable” ‘Life forces us to valorize the act of telling the truth and above all, Life imposes charity upon us: interpreting the discourse of a co-subject, understanding the other and communicating with her, is a charitable quest’ (Parret 1994b [61]: 235; emphases in the original). If the mind involved in this sort of behavior is a ªrmly biological one, it does not per se function deterministically. That is to say that scientiªc psychology may very well handle certain inferential procedures in the processing of language as if they are causally driven, because they have become deeply entrenched, automatized routines (or subintentional habits), but it cannot get at the core of communication, which ªnds its legitimization in the aŸective community (and in the norms shaped by its collective patterns of behavior), not in the individual. This appetitive core is not exactly intractable, for it is expressed canonically in actual discourse and surfaces most prominently in the workings of rhetorical ªgures, but its relation to reason and the structure of beliefs is bound to be more dialectical than straightforwardly mechanistic. Like all of mental activity, it is semiotically constrained by its own describability, and by the expressivity of its contents. In the Frankfurt School (Apel, Habermas), the sense of community grounding communicative rationality is, moreover, explicitly argumentative and consensual (Parret 1993a [60]), i.e., it is realized in and through collaborative discourse. Every utterance is always implicitly looking to reconstruct its own rationality qua argumentative discourse, and is thus utterly re¶exive on every one occasion. For Parret, however, the grounding of rationality does not stop there. There is still the work of the imagination to be dealt with, insofar as it “presentiªes” objects (Darstellung) as immediate, singular, nonconceptual, and thus ultimately nondiscursive, feelings. Parret follows Kant in relating community to something like a sensus communis, a common sensibility and the baseline for any kind of judgment, including esthetic ones. In fact, Kant’s third Critique may be thought of as a critique of logocentric intersubjectivity and communicability. It proposes that there is a “sensible” way of coexisting (being copresent) that “logically” precedes the community of speaking and arguing individuals.5 In this way, the radically social comes to join the radically sensible, which is not private at all. This implies that the aŸective community is both universally given (leading the way to a certain communicability) and
Mind the gap: Pragmatics and cognition today
self-validating. The aŸectivity sensed in the company of others, and particularly in the shared experience of beauty, gives rise to a fusion community, or a community of fused subjects. Placing oneself outside this community destroys the very possibility to act (speak, etc.) rationally. To use language, on the other hand, in the search for intersubjective understanding is automatically to be admitted into the community, to receive and accept the gift of life and its implicated meanings, which come to us from without. For all these diverse reasons, the category of meaning in natural language should, at the same time, be treated as subjective (what I mean) and as objective (what others mean). It is the social nature of meaning that provides the link between the subjective and objective readings of our utterances. In most cases, both aspects of meaning are determined by social factors (the same social factors) and we should thus strive for an explanation of the correlation between the “internal” and the “external” — which is requisite for bridging the gap we have been staring into — that does not involve trying to make this relation causal.
5.
Conclusion
In these cybernetic times, Parret’s program is primarily a vitalistic one. It does not deny the mental reality of language, but it does distrust the presupposition that all of language production and comprehension revolves around an organic system’s mechanistic processing of information, or propositional knowledge. More is involved in the human struggle for survival than the mere solving of problems, and the instrumental use of symbolic resources, like language, in the process. Discourse also creates things and relations, most important of which are the binding and bonding forces of illocutionary descent that make up speech communities, and thus society, and those are valuable symbolic assets, too. It is, for this reason, entirely futile to study discourse and pretend to explain it while abstracting away from its conditions of possibility, given the constitutive relation between (the act of) enunciation and the meaningful utterance(s) resulting from it. That would be like Galileo describing (planetary) motion, which is ªne until one rises to the challenge of explaining what has been described, of ªnding the causation that is intrinsic to the entities involved and that makes the phenomenon at issue possible: Newton and gravity. We should not forget, furthermore, the role of duties and desires driving our behavior, because an account of so-called “belief change” is incomplete without reference to the ªnal desire, a higher-order one involving a desire to have, and satisfy, a set of concrete
15
16
Frank Brisard
desirables, the fulªllment of which usually (or even critically) requires the communicative employment of language. The ªnal desire in human life, if any, can only be “happiness”. Importantly, the decision that certain items are constitutive of one’s happiness is not a matter of judgment or belief, but one of will. Happiness consists in what we make it consist in, and this is a very serious aŸair, for it demands that we take seriously the appeal coming from life, and whose force we feel in the other as well. Yet happiness develops its own little quirks and intricacies, one of them involving Solon’s dictum “Call no man happy till he is dead”. Let us accept that the boundaries of meta-pragmatic analysis have been reached, and perhaps stretched, with this paradox, which invites us into the realm of serious metaphysics. Let us also accept, and conclude with this acceptance, that this has not stopped Herman Parret, nor will it stop him, from looking further, beyond pragmatics (Parret 1993b [24]), into the sensorial substance of language as a temporal medium, the sonority of the voice it uses, the very corporality of speech, and further still, into the eyes of beauty, which is why we speak to begin with. And all of this, so I may hope, for the sake of his personal happiness, and of that of his co-subjects.
Notes 1. It is usually argued that other such modalities, e.g. in the context of illocutionary or narrative structure, are not amenable to scientiªc analysis because they are simply too complex. With the advent of speech-act theory and other styles of “new rhetoric”, however, the regularity — albeit of a nonnomological type — behind many of the pragmatic (sub)systems that can be distinguished has been amply demonstrated. 2. In French, the terms “locutionary act” and “illocutionary act” are commonly translated as énoncé and énonciation, respectively. 3. This idea is also to be found in contemporary phenomenology and ecological psychology, and it is gradually ªnding its way into (cognitive-)scientiªc accounts, e.g., of perception, as well. 4. On a slightly more sophisticated note, the agreement may also be posited between linguistic expressions and mental representations of real states or events, without this detracting from the essence of such a correspondence view (cf. Parret 1982 [11]). 5. What remains “outside” language is the force or desire to articulate thought (see Parret 1975 [28]: 245).
Mind the gap: Pragmatics and cognition today
References Austin, John L. 1962 How to Do Things with Words. London: Oxford University Press. Davidson, Donald 1984 “Radical interpretation”. In Inquiries into Truth and Interpretation, D. Davidson, 125–139. Oxford: Clarendon. GoŸman, Erving 1969 Strategic Interaction. Philadelphia: University of Pennsylvania Press. Greimas, Algirdas-Julien and Courtés, Joseph 1979 Sémiotique: Dictionnaire raisonné de la théorie du langage. Paris: Hachette. Grice, Paul 1989 Studies in the Way of Words. Cambridge, MA: Harvard University Press. 1991 The Conception of Value. Oxford: Clarendon. 2001 Aspects of Reason. Edited by Richard Warner. Oxford: Clarendon. Habermas, Jürgen 1998 “Toward a critique of the theory of meaning”. In On the Pragmatics of Communication, J. Habermas, 277–306. Cambridge, MA: MIT Press. Levinson, Stephen C. 2000 Presumptive Meanings. Cambridge, MA: MIT Press. Montague, Richard 1974 Formal Philosophy. New Haven: Yale University Press. Parret, Herman 1975 “Idéologie et sémiologie chez Locke et Condillac: La question de l’autonomie du langage devant la pensée”. In Ut Videam: Contributions to an Understanding of Linguistics, W. Abraham (ed.), 225–248. Lisse: Peter de Ridder. 1976a “La pragmatique des modalités”. Langages 10: 47–63. 1976b “Le débat de la logique et de la psychologie concernant le langage: Marty et Husserl”. In History of Linguistic Thought and Contemporary Linguistics, H. Parret (ed.), 731–771. Berlin: Walter de Gruyter. 1976c “Principes de la déduction pragmatique”. Revue Internationale de Philosophie 30: 486–510. 1978 “Les théories linguistiques peuvent-elles être idéologiquement neutres? ou L’histoire d’un démon qui s’appelle Véridiction”. Quaderni del Circolo Siciliano, ‘Strutture Semiotiche e Strutture Ideologiche’: 69–92. 1979 “Eléments d’une analyse philosophique de la manipulation et du mensonge”. Manuscrito 2: 119–152. 1981 “La question et la requête: Vers une théorie anthropologique de l’acte de poser une question”. Revue de Métaphysique et de Morale, 1: 164–179 and 2: 326–346. 1982 “A note on representationalism.” In Knowledge and Representation, B. de Gelder (ed.), 139–154. London: Routledge & Kegan Paul. 1983a “L’énonciation en tant que déictisation et modalisation”. Langages 70: 83– 97.
17
18
Frank Brisard
1983b Semiotics and Pragmatics: An Evaluative Comparison of Conceptual Frameworks. Amsterdam: John Benjamins. 1984 “Regularities, rules and strategies”. Journal of Pragmatics 8: 569–592. 1985 “Contexts as constraints on understanding in a dialogue situation”. In Dialogue: An Interdisciplinary Approach, M. Dascal (ed.), 169–181. Amsterdam: John Benjamins. 1986 Les passions: Essai sur la mise en discours de la subjectivité. Brussels: Mardaga. 1987 “La manipulation et le mensonge”. In Prolégomènes à la théorie de l’énonciation: De Husserl à la pragmatique, H. Parret, 230–278. Bern: Peter Lang. 1988 “The epistemics of the question-answer sequence and its psycho-pragmatic limitations”. In Questions and Questioning, M. Meyer (ed.), 280–303. Berlin: Walter de Gruyter. 1989 “La sémiotique: Tendances actuelles et perspectives”. In Encyclopédie philosophique universelle, Vol. I: L’univers philosophique, A. Jacob (ed.), 1361– 1368 (article 208). Paris: Presses Universitaires de France. 1990 “Stratégie [ling.].” In Encyclopédie philosophique universelle, Vol. II: Les notions philosophiques, S. Auroux (ed.), 2464. Paris: Presses Universitaires de France. 1993a “Die Tranzendentalpragmatik und die Werte der Urteilsgemeinschaft”. In Transzendentalpragmatik: Ein Symposion für Karl-Otto Apel, A. Dorschel, M. Kettner, W. Kuhlmann & M. Niquet (eds.), 316–341. Frankfurt: Suhrkamp. 1993b The Aesthetics of Communication: Pragmatics and beyond. Dordrecht: Kluwer. 1994a “Indirection, manipulation and seduction in discourse”. In Pretending to Communicate, H. Parret (ed.), 223–238. Berlin: Walter de Gruyter. 1994b “La quête charitable de la vérité-valeur”. In Piano: Hommage à Per Aage Brandt, B. Rosenbaum, H. Sonne & A. M. Dinesen (eds.), 219–237. Aarhus: Center for Semiotisk Forskning. Searle, John R. 1975 “Indirect speech acts”. In Syntax and Semantics, Vol. 3: Speech Acts, P. Cole & J. L. Morgan (eds.), 59–82. New York: Academic. Turner, Ken 1999 “Introduction: From a certain point of view (seven inch version)”. In The Semantics/Pragmatics Interface from DiŸerent Points of View, K. Turner (ed.), 1–18. Oxford: Elsevier.
Selected books and articles by Herman Parret in philosophy of language and pragmatics
In English [1]
1971
[2]
1977
[3]
1978
[4] [5] [6]
1978 1980 1980
[7]
1980
[8]
1980
[9]
1981
[10] 1981 [11] 1982 [12] 1983 [13] 1983
[14] 1984
Language and Discourse [Janua Linguarum, Series Minor 119]. The Hague: Mouton. “Conventional implications and conversational implicatures”. In Language and Logic, D. Holdcroft (ed.), 175–196. Warwick: University of Warwick University Press. “A note on pragmatic universals of language”. In Language Universals: Papers from the Conference Held at Gummersbach/Cologne [Tübinger Beiträge zur Linguistik], H. Seiler (ed.), 125–140. Tübingen: Günther Narr. “Illocutionary force representations”. Acta Semiotica et Linguistica 2 (1): 245–256. Contexts of Understanding [Pragmatics & Beyond 6]. Amsterdam: John Benjamins. “Demonstratives and the I-sayer”. In The Semantics of Determiners, J. van der Auwera (ed.), 96–111. London: Croom Helm. “Proper names versus (I, you, we): A universal dimension”. In Wege zur Universalienforschung: Sprachwissenschaftliche Beiträge zum 60. Geburtstag von Hansjakob Seiler [Tübinger Beiträge zur Linguistik 145], G. Brettschneider and Chr. Lehmann (eds.), 81–86. Tübingen: Günther Narr. “The problem of the delimitation of semantics and pragmatics: Round table (with P. Seuren, H. Parret, P. Leonardi, M. Sbisà & F. Récanati)”. Quaderni di Semantica 1 (1): 97–134. (Editor, with J. Verschueren & M. Sbisà) Possibilities and Limitations of Pragmatics: Proceedings of the Conference on Pragmatics, Urbino, 1979 [Studies in Language Companion Series 7]. Amsterdam: John Benjamins. “The pragmatics of semantical theories”. Philosophica 27: 27–51. “A note on representationalism”. In Knowledge and Representation, B. de Gelder (ed.), 139–154. London: Routledge & Kegan Paul. Semiotics and Pragmatics: An Evaluative Comparison of Conceptual Frameworks [Pragmatics & Beyond IV:7]. Amsterdam: John Benjamins. “Shifting in and shifting out: Remarks on deictics”. In Proceedings of the XIIIth International Congress of Linguists, S. Hattori and K. Inoue (eds.), 1068–1071. Tokyo: n.p. “Regularities, rules and strategies”. Journal of Pragmatics (Special Issue on Metapragmatics, ed. by C. Ca¹) 8 (4): 569–592.
20
Selected books and articles by Herman Parret
[15] 1985
[16] 1985
[17] 1986
[18] 1987 [19] 1988
[20] 1989
[21] 1989
[22] 1991
[23] 1992
[24] 1993 [25] 1994 [26] 1994
“Contexts as constraints on understanding in a dialogue situation”. In Dialogue: An Interdisciplinary Approach [Pragmatics & Beyond Companion Series 1], M. Dascal (ed.), 169–181., Amsterdam, John Benjamins, 169–181. “Time, space and actors: The pragmatics of development”. In Developmental Mechanisms of Language, Ch.-J. Bailey and R. Harris (eds.), 131–148. Oxford: Pergamon. Entry “Pragmatics”. In Encyclopedic Dictionary of Semiotics [Approaches to Semiotics 73], Th. A. Sebeok (general ed.), 751–760. Berlin: Mouton de Gruyter. “Can pragmatic meaning be processed by a machine?”. Gengo Kenkyu/Journal of the Linguistic Society of Japan 92: 1–32. “The epistemics of the question-answer sequence and its psycho-pragmatic limitations”. In Questions and Questioning [Foundations of Communication], M. Meyer (ed.), 280–303. Berlin: Walter de Gruyter. “Cognition, the localist hypothesis, and back to Kant”. In Worlds behind Words: Essays in Honour of Professor F. G. Droste on the Occasion of his Sixtieth Birthday, F. J. Heyvaert and F. Steurs (eds.), 37–49. Leuven: University Press. “Paraphrase as a coherence principle in conversation”. In Text and Discourse Connectedness: Proceedings of the Conference on Connexity and Coherence (Urbino, July 16–21, 1984) [Studies in Language Companion Series 16], M.-E. Conte, J. S. Petöª and E. Sözer (eds.), 281–289. Amsterdam: John Benjamins. “Deixis and shifters after Jakobson”. In New Vistas in Grammar: Invariance and Variation [Current Issues in Linguistic Theory 49], L. R. Waugh and S. Rudy (eds.), 321–340. Amsterdam: John Benjamins. (With Jef Verschueren) “(On) Searle on conversation: An introduction”. In (On) Searle on Conversation [Pragmatics & Beyond 21; compiled and introduced by H. Parret & J. Verschueren], J. R. Searle, H. Parret and J. Verschueren, 1–5. Amsterdam: John Benjamins. The Aesthetics of Communication: Pragmatics and beyond [Library of Rhetorics 1]. Dordrecht: Kluwer. “Indirection, manipulation and seduction in discourse”. In Pretending to Communicate, H. Parret (ed.), 223–238. Berlin: Walter de Gruyter. (Editor) Pretending to Communicate [Foundations of Communication and Cognition]. Berlin: Walter de Gruyter.
In French (and German) [27] 1972 [28] 1975
“Nouvelles contributions au débat de l’empirisme et du rationalisme en théorie linguistique”. Communication and Cognition 5: 99–116 & 253–268. “Idéologie et sémiologie chez Locke et Condillac: La question de l’autonomie du langage devant la pensée”. In Ut Videam: Contributions to an Understanding of Linguistics, W. Abraham (ed.), 225–248. Lisse: Peter de Ridder.
Selected books and articles by Herman Parret
[29] 1975
[30] 1976 [31] 1976
[32] 1976 [33] 1976 [34] 1976 [35] 1978
[36] 1979 [37] 1979 [38] 1979
[39] 1980 [40] 1980 [41] 1980
[42] 1981 [43] 1981
[44] 1982
[45] 1982
“La grammaire devant les intentions”. In Beiträge zur Grammatik und Pragmatik [Skripten zur Linguistik und Kommunikationswissenschaft 12], V. Ehrich and P. Finke (eds.), 187–200. Kronberg: Scriptor. “La pragmatique des modalités”. Langages 10 (43): 47–63. “Le débat de la logique et de la psychologie concernant le langage: Marty et Husserl”. In History of Linguistic Thought and Contemporary Linguistics [Foundations of Communication, Library Editions], H. Parret (ed.), 731– 771. Berlin: Walter de Gruyter. “Les paralogismes de l’énonciation”. In Philosophie et langage 1–2, Ch. Perelman (ed.), 97–120. Brussels: Éditions de l’Université Libre de Bruxelles. “Principes de la déduction pragmatique”. Revue Internationale de Philosophie 30: 486–510. “Sémantique structurale et sémantique generative”. In Sémantique et logique, B. Pottier (ed.), 85–108. Paris: J. P. Delarge. “Les théories linguistiques peuvent-elles être idéologiquement neutres? ou L’histoire d’un démon qui s’appelle Véridiction”. Quaderni del Circolo Siciliano, “Strutture Semiotiche e Strutture Ideologiche”: 69–92. “Ce qu’il faut croire et désirer, pour poser une question.” Langue Française 42: 85–93. “Éléments d’une analyse philosophique de la manipulation et du mensonge”. Manuscrito 2 (2): 119–152. “Une théorie linguistique sans concept de signe est-elle possible?”. In A Semiotic Landscape/Panorama sémiotique: Proceedings of the First Congress of the International Association for Semiotic Studies, Milan, 1974, D. Chatman, U. Eco and J. M. Klinkenberg (eds.), 341–344. The Hague: Mouton. (Editor) Le langage en contexte: Études philosophiques et linguistiques de pragmatique [Linguisticae Investigationes Supplementa 3]. Amsterdam: John Benjamins. “Les stratégies pragmatiques”. Communications (Special Issue on Actes de discours) 32: 250–273. “Pragmatique philosophique et épistémologie de la pragmatique: Connaissance et contextualité”. In Le langage en contexte: Études philosophiques et linguistiques de pragmatique, H. Parret (ed.), 7–189. Amsterdam: John Benjamins. “La question et la requête: Vers une théorie anthropologique de l’acte de poser une question.” Revue de Métaphysique et de Morale 1: 164–179 & 2: 326–346. “Le concept de nature humaine dans Chomsky: Table ronde (with H. M. Bracken, T. de Mauro, C. Otero, H. Parret & L. Prieto)”. Quaderni de Semantica 2 (2): 233–280. “Les positions paradigmatiques de la linguistique et son idéologie essentielle”. In Ideologia, ªlosoªa e linguistica: Atti del Convegno Internazionale di Studi, Rende, 1978 [Pubblicazioni della Società di Linguistica Italiana], D. Gambarara and A. d’Atri (eds.), 69–89. Rome: Bulzoni. “Nouveaux éléments de pragmatique intégrée”. Annales de l’Institut de Philosophie et de Sciences Morales (Brussels: Éditions de l’Université Libre de Bruxelles): 61–76.
21
22
Selected books and articles by Herman Parret
[46] 1983 [47] 1983 [48] 1986 [49] 1987 [50] 1987 [51] 1987 [52] 1987 [53] 1988
[54] 1989
[55] 1989 [56] 1989
[57] 1990
[58] 1991 [59] 1991
[60] 1993
[61] 1994
(Editor) La mise en discourse. Special Issue of Langages 70. (“Présentation”: 5–7.) “L’énonciation en tant que déictisation et modalisation”. Langages 70: 83–97. Les passions: Essai sur la mise en discours de la subjectivité [Philosophie et Langage]. Brussels: Mardaga. “La manipulation et le mensonge”. In Prolégomènes à la théorie de l’énonciation, H. Parret, 230–278. Bern: Peter Lang. “L’axiologisation de la pragmatique”. Archivio di Filosoªa LV (1–3; Special Issue on Etica e pragmatica): 13–38. “L’énonciation et sa mise en discours”. Cruzeiro Semiotico 6 (Special Issue on Enunciaçao, enunciada e semiotica discursiva II): 92–101. Prolégomènes à la théorie de l’énonciation: De Husserl à la pragmatique [Sciences pour la Communication 14]. Bern: Peter Lang. “Vers une théorie énonciative de la paraphrase: La ressemblance et l’acte de proximisation”. Modèles Linguistiques 19 (Special Issue on Ambiguité, paraphrase et langage en acte): 33–50. “Empreinte pragmatiste, attitude pragmatique et sémiotique intégrée”. In Semiotics and Pragmatics: Proceedings of the Perpignan Conference, G. Deledalle (ed.), 277–296. Amsterdam: John Benjamins. “La communication et les fondements de la pragmatique”. Verbum (Revue de linguistique, University of Nancy II) 12 (2): 207–219. “La sémiotique: Tendances actuelles et perspectives”. In Encyclopédie philosophique universelle I: L’univers philosophique, A. Jacob (ed.), 1361–1368 [Article 208]. Paris: Presses Universitaires de France. Entries “Compréhension”, “Discours”, “Énonciation (statut de – )”, “Passion”, “Rationalité ( – discursive)”, “Réciprocité”, “Règle”, “Régularité ( – linguistique)”, “Stratégie”. In Encyclopédie philosophique universelle II: Les notions philosophiques. Dictionnaire, S. Auroux (ed.), passim. Paris: Presses Universitaires de France. “Les arguments du séducteur”. In L’argumentation (Colloque de Cerisy, 1987) [Philosophie et Langage], A. Lempereur (ed.), 195–213. Liège: Mardaga. (Editor) Le sens et ses heterogeneities [Sciences du langage]. Paris: Presses du CNRS. (Introduction, “De l’ (im)possibilité d’une grammaire de l’hétérogène”: 11–25.) “Die Tranzendentalpragmatik und die Werte der Urteilsgemeinschaft”. In Transzendentalpragmatik: Ein Symposion für Karl-Otto Apel [Suhrkamp Taschenbuch Wissenschaft 1081], A. Dorschel, M. Kettner, W. Kuhlmann and M. Niquet (eds.), 316–341. Frankfurt am Main: Suhrkamp. “La quête charitable de la vérité-valeur”. In Piano: Hommage à Per Aage Brandt, B. Rosenbaum, H. Sonne and A. M. Dinesen (eds.), 219–237. Aarhus: Center for Semiotisk Forskning.
Part I Pragmatics (standard and not so standard)
Ordinary time* Frank Brisard University of Antwerp
1.
Introduction
When I was working hard on my dissertation, not too many years ago, Herman Parret would ask me time and again if I could specify the nature of the temporal notions and categories that I was scrutinizing. I thought I was dealing with tense, i.e., the grammatical(ized) expression of location in time. But as tense is only a formal linguistic term, designating a class of language objects sharing certain structural properties, what are we to make of the notional domain, if any, underlying them? Would it actually su¹ce to call that domain “grammatical time”? I went through them all, in the text that became my thesis: “linguistic time” (simply the linguistic expression of temporal concerns), “subjective” or “psychological time” (down to and including neural processes of temporal organization in perception, memory, and planning), “objective” or “physical time” (and its connection to logical treatments of tense meaning), without necessarily selecting any of them as directly relevant to the analysis of tense. It seemed obvious that I would have to concentrate on manifestations of time — whichever kind of time, really — in language, which come in two major types. There is ªrst the time of lexicon, where speakers ªnd all sorts of resources for talking about time in more or less explicit manners. This is the time of culture and mythology, at various levels of deliberation, deªning so-called world views and cosmological systems. It is where diŸerences are made between “linear” and “cyclic” models of progress and change, and between assorted understandings of what it means for something to be past, present, or * The author is a Postdoctoral Fellow of the Fund for Scientiªc Research — Flanders (Belgium). I thank Tanja Mortelmans for commenting on a previous version of this paper.
26
Frank Brisard
future. But it is not per se the kind of time to which tenses “refer”. The time of grammar, then, seems to be of a wholly diŸerent nature, re¶ecting universal ways with issues of duration and sequencing (even if it does not by necessity presuppose a “Universal Grammar”). It represents the time that speakers talk in, the time that is needed to conceptualize and organize (and actually utter!) linguistic material, or, better still, the time that constitutes temporal objects (displaying their own temporal structures and relations) and gives them meaning — an epistemic meaning, as I will argue. It is also, in the end, the time that I did not get to name in my subsequent work, not out of some lack of interest but because I plunged into the empirical study of grammatical predications. I am now calling that same time, in the title of the present paper, “ordinary”, not so much as a serious philosophical proposal but rather as a way of acknowledging Herman’s profound in¶uence on my theoretical conceptions of pragmatics, and of the pragmatic analysis of linguistic meaning. Somehow, “ordinary-language philosophy”, a certain cognitive take on the social embedding and meaningful embodiment of language users, and a generally Kantian way of approaching speech and semiosis all came together in my mind and formed a (somewhat hybrid) analytical template that was, and still is, useful in the study of real, contextualized, and strategic language-in-action. The phenomenological move that I am proposing below should be seen in this light, too. It is one, I’m sure, which Herman will appreciate and, above all, come to recognize as largely tributary to his own meditations in time, “that great sculptor”.
2.
The “logic” of tense
The formal study of grammatical time, including tense and aspect, traditionally concentrates on the question of how speciªcally temporal relations can be established between the situations expressed by two consecutive clauses in a discourse: the sequencing problem. Thus, the normal reading of an utterance like (1) would have it that the pushing event occurred before the falling event, that the latter was caused by the former, and furthermore that the falling occurred as soon as its protagonist was pushed, not several seconds or minutes thereafter: (1) I pushed her and she fell.
Ordinary time
In this, it is expressly assumed that the meanings of tense/aspect forms, which are constant, directly determine such ordering and, thus, that their main import to the interpretation of discourse lies in spelling out the metric organization of objective events (mainly in terms of their duration and sequential structure). No accommodations are made in this framework to integrate temporal and other, nontemporal, uses of the same forms and, indeed, there is a systematic intolerance vis-à-vis the occurrence of both referential ambiguity and temporal indeterminacy. When it comes to tense, there are generally two options available within the formalist research program. One is to deªne tense forms in terms of the relations they express with respect to the time of speaking (see also Comrie 1985 for a classic elaboration of this analytical perspective). This constitutes the deictic dimension of tense, and its analysis seems to be of particular relevance to the discussion of such phenomena as consecutio temporum (sequence of tenses). But even there, when two or more clauses contain the same tense form, as in example (2), the deictic dimension of tenses oŸers no conclusive information on the temporal relation between the clauses at hand (here: simultaneity vs. precedence with respect to the main event said): (2) John said that Mary was ill.
In cases like this and many others, it turns out that lots of encyclopedic, nonlinguistic information is called for to solve even the simplest of clausal sequences, just because so much (temporal) information is actually missing from the tense forms themselves, which hardly encode anything directly “conceptual” at all. A second approach is to stress the anaphoric dimension of tense meaning (e.g., Reichenbach 1947, McCawley 1971, Partee 1973), which links a situation to a reference point or antecedent on top of situating it deictically. The reference point itself can usually be inferred from the discursive context or the immediate surroundings of the situation of speech. The use of “imperfect” and “perfect” tenses to refer to a past situation would then be a typical example of such a diŸerence in reference-point construction. The diŸerence is explained by claiming that an imperfect, like (certain uses of) the English simple past, relates the reference point to a past event, with which it is simultaneous, while the (present) perfect has the reference point coincide with the time of speaking and look back upon a past event. The latter conªguration typically prompts the assignment of “current relevance” to a past event, which is often associated with uses of the present perfect. Within and outside of the formalist
27
28
Frank Brisard
literature, then, it has become increasingly clear that any substantiation of the reference-point character of tense meanings will invariably have to include psychological notions of “perspectivization”, and thus leave the domain of tense logic proper.1 A ªrst problem in formal-semantic descriptions of tense concerns the exact nature of the perspectival contrasts between tenses that otherwise denote the same time frame, as well as its implications on the overall conception of tenses as temporal forms. Consider the case of past-tense markers again. From a deictic point of view, past tense should exclusively indicate ‘past’ or ‘precedence’ with respect to the time of speaking (or some shifted deictic center). Yet whenever a situation is presented by means of a past-tense form, the situation can also be linked to an independently provided, or inferable, reference point (in the past). And this is precisely where the French imperfect (imparfait) and simple past (passé simple) diverge: (3) Paul entra. Marie faisait la vaisselle. ‘Paul came in – SIMPLE PAST. Marie was doing the dishes – IMPERFECT.’
Both forms portray a situation as given in the sense of belonging to some facet of reality. But while the simple past presents past events as wholly appearing in the past, cut oŸ from the present, and punctual, the imperfect construes them as “actual”, meaning that they are presented as unfolding and durational, as perceived from an internal perspective. The French imperfect thus presupposes a (nonpresent) center of apprehension that is not immediately related to the primary deictic center and can be directly linked to the psychological notion of a subjective, perceiving mind other than the current speaker’s. The reference point needed to set up this perspective is precisely provided by the preceding use of the simple past. This diŸerence in the presentation of (past) facts between the imperfect and the simple past is one of construal, of course, and it is one that also allows any cognitive analysis of tense sensitive to it to proceed and unify all of the usage types that can be discerned for a given form, also counting the many modal uses that typify tenses. Tense forms are, on the whole, extremely vague in the temporal interval they denote (if any). This constitutes a second formal problem, one that is taken up in Relevance Theory as a speciªcally pragmatic issue of “underspeciªcation” (Wilson & Sperber 1998). Take the use of the English present perfect to refer to a past situation with current relevance, as in (4) and (5):
Ordinary time
(4) I’ve had breakfast. (4') I haven’t had breakfast. (5) I’ve been to Tibet.
Here, the hearer, who supposedly wishes to ªnd out for which speciªc time interval these propositions hold, embarks on a search guided by a heuristic procedure that might be characterized in terms of “optimal relevance”. In the ensuing inferential process, an ordered set of temporal indices (e.g., within the last few minutes, hours, weeks, etc.) is generated that is structured by logical relations of entailment and conforming to the general meaning of the present perfect in English, viz., that the event in question took place somewhere between the time of speaking and the origin of the universe (or, in this case, of the subject referent). Nothing in the meaning of the present perfect itself suggests the availability of one generally “correct” or even “preferred” time interval for this tense; the interval for (5), something like within my lifetime, does not diŸer from that of (4) by virtue of anything that the present perfect additionally intimates, but depends entirely on context. At the same time, the hearer’s eventual choice of interpretation will aŸect the utterance’s truth conditions, in that the proposition at issue will be true or false with varying chosen intervals. We can see this clearly in the progressively declining acceptability of the same utterances under negation, depending on which interval is selected. With example (4'), for instance, it seems as if there is a basic temporal scale at which the utterance can readily be made sense of (presumably some time within the last few hours), and that intervals situated at scales that are either too small or too large invariably yield questionable interpretations (?I haven’t had breakfast within the last few minutes/months).2 Of course, the appropriate level or scale will diŸer for varying predications, since zooming in on the interval that ªts a given predication best requires (nonlinguistic) knowledge of the realworld activities involved, and of the discourse conditions that are holding at the time of speaking, at least if the information provided by explicating the relevant interval is to produce communicative (possibly cognitive) eŸects relevant enough to merit attention. Once more, we notice that precise temporal reference is not an issue in tense semantics, and that, to determine such reference, hearers will be better oŸ if they rely on lexical means and context. The vagueness of the temporal reference eŸectuated by individual tenses does not end here, either, since it also aŸects the hearer’s capacity to calculate, in one way or another, the relations between discursively connected events. In fact, the problem of determining
29
30
Frank Brisard
exact referential values for temporal intervals is tightly linked to the sequencing problem, as Wilson & Sperber (1998: 1–2) demonstrate in their discussion of utterances like (6) or (7): (6) John dropped the glass and it broke. (7) They planted an acorn and it grew.
Despite being presented in exactly the same formal fashion, the events described in each of these utterances are separated by very distinct intervals, in a way that is di¹cult to predict on linguistic grounds only. For one, unlike with events like breaking or falling (example 1), we would not generally expect an acorn to sprout and start growing as soon as it hits the ground. Methodologically speaking, the most consequential problem within tense theory is certainly that of the massive polysemy characterizing virtually all tense forms, regardless of which language one is looking at. This represents a third problem in our discussion. On the whole, meanings that do not agree with a tense’s traditional nomenclature (“allotemporal”), or meanings that are downright “nontemporal” (typically modal), are all treated as derivative of the basic semantics of tenses, which is allegedly temporal and nonambiguous, and as inferable from a tense’s essentially temporal meaning through such pragmatic or generally cognitive principles as implicature (see especially Dahl 1985: 11Ÿ), metaphor, and other mechanisms of “loose talk”. In the worst case, all of these diverging usage types are simply excluded from formalist analysis (cf. Brisard 1999 for discussion). And yet, when we have a closer look at the “allotemporal” tense uses, we ªnd that each and every tense form can refer to practically all of the notional time frames — past, present, and future — that are canonically distinguished (as portrayed in Table 1). Independently of the speciªc tense typologies than can be found throughout the world’s languages, this result can be expected for each attested system (whether tripartite ‘past-present-future’, as in French, or binary, including, e.g., English ‘past-nonpast’ and Hua [New Guinea] ‘future-nonfuture’), because we are looking at the complete usage range of tense markers, not at a subset of theoretically sanctioned usage types. In this sense, ambiguity seems to be the norm in tense semantics, rather than the exception.
Ordinary time
Table 1. Possible semantic values for diŸerent paradigms of tense marking (in English).3 “allotemporal” usage types (simple tenses)
“nontemporal”/modal usage types
PRESENT
‘Past’ (e.g., historical present) ‘Future’ (e.g., scheduled activities, near future)
General validity* / “timeless” Hypothetical Performative Stage and other instructions, photo captions, etc.
PAST
‘Present’ ‘Future’ (both in politeness contexts)
Hypothetical/counterfactual “Optative” Mitigation
FUTURE
‘Present’ (with epistemic uses)
Deontic modality Hypothetical General validity
* The class of general-validity statements comprises any kind of proposition that is qualiªed as holding in some structural way, from habituals to generic/gnomic utterance types.
Likewise, Table 1 shows us how the same tenses can be deployed for the expression of grammatical interests that have little or nothing to do with the location of an event in time. With the exception of deontic types, these mood and modal categories are concerned with the epistemic assessment of a state of aŸairs, that is to say, with the reality status that is attributed to a proposition representing that state of aŸairs. It is therefore not implausible that at least some of the functions of tense, too, can be directly inscribed within a more general, ultimately modal, model of how tenses relate to clause contents.
3.
A cognitive approach
In Cognitive Grammar (CG), tenses are “grounding predications” relating a proposition to the ground. The ground is a cognitive model of the world, and speciªcally of its immediate manifestation at the time of speaking, and includes representations of the speech participants and their constitutive, including mental, environments. The cognitive nature of the ground also allows for a knowledge-oriented focus, in which speakers have some sort of epistemic control over what they are saying. This is indexed by the presence or absence of a modal (re¶ecting a proposition’s real and unreal/potential status, respectively) and, within the realm of reality, by various kinds of formal marking that may be
31
32
Frank Brisard
taken as iconic for the degree of “separation” from the ground that is being expressed. What (indicative) tenses do, in the absence of a modal, is “locate the proªled process within the realm of reality, where the proximal and distal morphemes are susceptible to a temporal construal” (Langacker 1991: 250). Thus, temporal values of tenses, though prototypical in many cases, are treated as interpretations of a schematic value that is qualiªed as epistemic and that holds for all conceivable grounding predications, including modals. What tenses share with modal expressions that can be described as grounding, is that both types of predication indicate the relation between a proposition and the ground (as an assessment of the proposition’s likely occurrence in reality) without, importantly, focusing on this relationship itself. The use of tenses and epistemic modals implies that the speaker is judging a proposition’s epistemic status, as a matter of grammatical habit as it were, without making this the actual point of her message, as depicted in Figure 1. Here, the relation between a “trajector” (subject; tr) and a “landmark” (object; lm) is what is speciªed by a verb’s “proªle”, i.e., it is the relation together with its participants that is actually designated (represented by heavy lines). By itself, however, this proªle does not single out a speciªc instance of the verbal process type, unless it is elaborated in a ªnite clause. This is established by using a clausal grounding predication, which may then reªne the nature of the grounding relation (Rg) at issue. For instance, it may specify that there is a complete coincidence between the proªled process and the ground (G) of the speech event, in which case the suggested interpretation seems to be that the corresponding proposition’s epistemic status is considered as being ‘immediately certain’, and possibly, but not necessarily, this interpretation also leads to a temporal construal in which the given process is viewed as ‘present’. But even with this temporal interpretation, what the grounding predication is not doing, is present the temporal status of a proposition as an issue in its own right. Instead, information about the ground and its relations is treated as implicit and nonsalient, serving only as an “oŸstage” concern. This would make tr
lm
Rg
G
Figure 1. Clausal grounding in CG.
Ordinary time
the meanings of tenses and other grounding predications fundamentally subjective, by virtue of their grammatical status, because the act of relating a process to the ground is a mental one performed by the speaker, and it is to be conceptually distinguished from the “objective” information (the proªle) that the speaker wishes to communicate. The schematic characterization of grounding predications as always involving epistemic judgments is compatible with empirical work in the study of grammaticalization, where it is noticed that many grammatical meanings have lost much of their original contents (“bleaching”) in favor of more abstract, potentially even nonreferential, values, typically described as pragmatic rather than semantic in nature (e.g., Traugott 1989). This is certainly analogous to what occurred in the tense paradigms of numerous languages, where the original, grammatical function of a tense form (to locate situations in time) has often been replaced, or at least vastly enriched, by expressly modal meanings. As a result, the analysis of clausal grounding predications in CG starts from a highly abstract conception of tense and mood meanings, focusing on their respective epistemic contributions. In Figure 2, an “elaborated epistemic model” is proposed for these purposes, specifying the possible “addresses” within and outside reality that can be attributed to grounded clauses. Into this model, we incorporate the conceptualizer C’s realization that reality as she knows it “is not exhaustive of the world and its evolutionary history” (Langacker 1991: 243). The area of “irreality” is accordingly cut up into “unknown reality” on one hand, comprising situations whose reality has not been established (yet) or of which the conceptualizer is entirely ignorant, and that which is treated as “nonreality” on the other: Non-Reality Unknown Reality Reality Known
Known Reality Reality
Immediat
Immediate e Reality Reality C
Figure 2. The “elaborated epistemic model” (Langacker 1991: 244).
33
34
Frank Brisard
Using the tools of CG, a tense semantics based on the subjective assessment of temporal objects seems within reach. But we are still in need of an instrument to analyze the perspectival nature of grammatical markings of time, i.e., the fact that diŸerent tense forms may present distinct viewpoints on the same time interval. In CG, this is usually addressed by pointing out the default reference-point status of the ground, as the origo or starting point for building epistemic grounding relations, and by assuming that this default status of the ground can be overridden by a variety of factors, which may produce a shift in reference point from the ground to another (real or unreal) salient entity. An alternative, closely akin, tactic to solve the same problem consists in the Mental-Space approach, which postulates that language involves the construction of mental spaces, of relations between them, and of relations between elements within them. Reference identiªcation or, in the case of tense, the location of a state of aŸairs in time, is a function of this dynamic construction work, and of the relative accessibilities of the spaces involved. It is explicitly not assumed that grammatical markers in themselves represent or encode such conªgurations. Mental spaces are built in accordance with instructions provided by linguistic expressions (“space builders”), which give us “minimal, but su¹cient, clues for ªnding the domains and principles appropriate for building in a given situation” (Fauconnier 1994: xviii). That these instructions are minimal should lead us to suspect that fairly schematic meanings are at work in the course of space tracking, i.e., of interpreting grammatical structures and morphemes. In fact, in her application of the mental-space frame to the analysis of tense, Cutrer (1994) explicitly introduces such putatively universal terms as “FACT” and “PREDICTION”, which help describe the semantic import of tense markers by extending their purely temporal characterization into the realm of epistemic judgments. Thus, “[i]n moving (mentally) from one mental space to another, we try to keep track of the time shifts and epistemic shifts between the spaces in focus” (Fauconnier 1994: 72; see also Dinsmore 1991 for an explication of the notion of “focus”). In general, then, no conceivably comprehensive account of tense usage can be oŸered without the inclusion of information regarding relative accessibilities and the distribution of mental viewpoints, in contrast with the exclusive focus on the objective location of event times exhibited by formalist approaches. In many cases, moreover, even the viewpoint does not necessarily stand in a temporal relation with the actual base space (the ground). We may conclude that the schematic relation between the diŸerent spaces involved in the interpretation of all tense (and mood) marking should not
Ordinary time
properly call upon the domain of time in the ªrst place, or at least not out of analytic necessity.
4.
Lived time
In contrast with the logicist view, tense can be seen as the grammatical expression of “temporality”, not of time. Temporality is a concept from phenomenology referring to “lived time” and thus shifts the focus of analysis to subjective properties of temporal objects, away from the physical properties of objective time (i.e., duration and succession). Temporal objects display diŸerent “modalities of givenness” (Husserl 1952) and it is these that tenses tend to display or indicate, in line with the general, nonlinguistic observation that the experience of time is constituted by how it aŸects objects, and not so much by how time itself is felt to pass. The speciªcally tripartite structure of time, which is universal but has the potential of being grammatically rearranged in diŸerent languages, is given by the corresponding structure of temporal awareness, which divides into “primal-impressions” of co-present objects, “retentions” of past ones, and “protentions” of future ones. The baseline for this structure is the immediate, actual perception of the now, such that both memories of past and anticipations of future objects are accurately described as “modiªcations” of the primitive mode of corporeal perception: “[…] our retentions, primal-impressions and protentions intend their respective object-phases not only with respect to their temporal locations but also with regard to their extratemporal qualities” (Miller 1984: 138). That the resulting modalities, in linguistic and other types of experience, take on a fundamentally epistemic aspect can be gathered from the suggestion that “every object has its possible varying modes of being valid, the modalizations of ontic certainty” (Husserl 1970: 143). The idea is plainly that being past, present, or future is not a matter of coinciding, or not, with a punctual moment deªned as now, or of being copresent vs. absent relative to an actual consciousness. Instead, the now of the conceiving mind is highly structured and, with respect to temporal objects, it accommodates diŸerences of form rather than content, which should be revealed by a true description of our temporal experience. Thus, both the past and the future, insofar as speciªc past and future objects are intended, are actually “present” to us, though in a diŸerent form4 than the now: the extratemporal qualities of retentions have at least some kind of determinacy, derived from the sensorily ªlled primal-impressions they modify, whereas those of protentions
35
36
Frank Brisard
L
t1 A
t2 B
t C
G
E
t3 D
F
Figure 3. “Perceiving the continuity of a tone”: Husserl’s two-dimensional model of temporal awareness (from Miller 1982: 122).
may suŸer from a uniform indeterminacy, resulting in an intuitively graspable epistemic asymmetry between past and future as such. This kind of information cannot be contained by a one-dimensional representation of the ¶ow of time and calls for an additional axis, fully situated in the present and moving along as time goes by, upon which to project past and future states of aŸairs, as in Figure 3 (see also Brisard 1998: 53Ÿ.). In the ªgure, we notice that past events, such as point A, are projected onto a vertical line, the lower part of which represents a continuous series of past events (e.g., hearing a tone) up till the present moment, such that point F is the modiªed representation of point A in the conceptualizer’s consciousness at time t. Thus, the fact that point F is “twice as past” as point E is rendered by estimating the vertical distance between both points. This way, both referential aspects of temporal objects can be taken into account (i.e., their relative locations on the timeline are provided for, as the projecting function preserves temporal ordering), and a qualitative distinction can be made between past, present, and future states of aŸairs in terms of their modiªed formats of representation. In this respect, then, the vertical line linking points L and F can be interpreted as an indicator of the epistemic status of these objects, the form in which they are “present” at the time of speaking. It would correspond to the ever-evolving face of the growing cylinder of (known and unknown) reality in Figure 2, which inevitably displays a certain internal complexity as well. It is reasonable, primarily on the basis of empirical observations such as presented above, to submit that it is this form that is the focus of the tense-aspect-mood category in its entirety.
Ordinary time
The concept of temporality (or, lived time) speciªcally invites a genuinely uniªed tense semantics, in which the meaning of a tense form is treated as a constant, if not a temporal one. A theory like CG is perfectly capable of substantiating, for purposes of grammatical analysis, the subjective aspects that are necessarily involved in the act of grounding an utterance (i.e., of assessing how accessible its clausal content is from a given base space), thereby removing itself from a tradition of tense analysis based on postulates of physicalism and objectivism. Indeed, both the phenomenological analysis of time consciousness and a phenomenologically inspired analysis of tense meaning require “the complete exclusion of every assumption, stipulation, and conviction with respect to objective time […]. Just as the actual thing, the actual world, is not a phenomenological datum, neither is world time, the real time, the time of nature in the sense of natural science and even in the sense of psychology as the natural science of the psychic.” (Husserl 1990: 4–5) Such a demand for a radical suspension of objective time directly goes against one of the basic assumptions of a logical treatment of tense, viz., the belief that philosophically sound contributions to the study of temporal reasoning should shy away from anything not immediately related to the positivist problem of empirical knowledge. In a truly anti-cognitive spirit, Reichenbach, for one, explicitly attacks the supposition that “unaŸected by physical time, the psychological experience of time retains its a priori [i.e., subjective, FB] character and obeys its own laws” (Reichenbach 1958: 113). At the same time, a properly phenomenological outlook allows us to qualify the status of a so-called cognitive approach to tense. Even though certain claims made in CG could be taken in the strongest neuro-psychological sense of dealing with issues of processing (and thus of processing time), essentially the concept of grounding, central to understanding tense, refers to notions of epistemic control and of practical certainties and predictabilities, more crucially than it does to veriªable positions in conceived time.5 All of the instrumental notions associated with grounding (reality, deªniteness, immediacy, etc.) are at any rate more directly related to meaningful ways of coping with the environment (of surviving, also symbolically) than they are to low-level, automated mechanisms of information processing.
5.
Conclusion (Cognitive, not cognitivist)
A theory of tense that might emanate from the CG approach or related (dynamic) paradigms is primordially situated in the sphere of people’s intentional
37
38
Frank Brisard
actions, and “cognitive” in this context does not reduce, in any naturalized sense, to spelling out mere (transitions between) information states of a physical system like the brain, or to the computational machinery and the logic of its strategies that can be suspected behind them. Rather, the term includes in its scope constructive, creative acts of rational interpretation that immediately incorporate the perspectival nature of embodied meaning. Since any object of an intentional act of mine is constituted by me as from a ¶ux of changing temporal perspectives, corresponding to the object’s successive phases of appearance, the added dimensionality that perspective oŸers is not just an option available to the description and analysis of selected tense forms, but represents a fundamental condition of generating and interpreting meaningful behavior in time. This is what is captured in the conceptualist notion of construal, and it is also what is needed in any account of grammatical categories that is empirically comprehensive and that truly appreciates the subjective side of meaning in grammar.
Notes 1. The relevance of conceptual matters of construal, a notion which is used in cognitive models to account for the special status — in terms of prominence — of certain elements in a given semantic conªguration, is generally not acknowledged in formalist circles. The only place where such interests are provided for within standard generative analyses is in the consideration of syntactic “transformations” of predicate-argument structure (movement, binding traces), like passivization or wh-movement: “Transformations thus provide a layer of meaning beyond who did what to whom; that layer emphasizes or seeks information about one of the participants, while keeping constant the actual event being talked about.” (Pinker 2003: 19) 2. I haven't had breakfast within the last few months could have a plausible interpretation, but then the speaker would probably be talking about her habit of specifically not eating breakfast (as opposed to lunches or dinners) during the day. One would think that this differs from the more canonical interpretation of the utterance in (4'), without any temporal specification, which is that the speaker simply has not eaten yet, not focusing on the fact that the type of meal she missed was breakfast. This would imply that I haven't had breakfast is typically uttered before or around noon, and that any other time of utterance would by itself trigger the search for a more special reading. 3. I have included a slot for the future tense here, even though this ªgure is meant to display the situation for English and English lacks a synthetic future-tense form. This is not in itself problematic, however, since (i) our focus lies on the analysis of grammatical expressions, including analytic ones, for which temporal reference is (perceived as) one of the prototypical functions, (ii) “future reference is never the exclusive property of the single paradigm
Ordinary time
which grammars label — often arbitrarily — future tense” (Fleischman 1982: 1–2), and (iii) similar observations also hold for true future tenses in other languages. 4. As an example, the feeling of duration requires an actual perception in the here-and-now, augmented with the retention of “the same” object of perception as it appeared previously. Retention is thus deªned as the apprehension of a mentally present object that has physically disappeared. The feeling of removal typical of (remembering) past experiences results from the retention, not only of previous primal-impressions, but also of their respective retentions. It is the retention of retention that is responsible for what we perceive as the typical epistemic features of the past, in line with the claim in CG that awareness (i.c., conscious memory) of an experience does not reside in the experience itself or in the neurological activity that constitutes it, but in a higher-order relationship which is in turn constituted by the experience of a higher-order process (of retention; cf. Langacker 1997: 50Ÿ.). 5. “Conceived time” represents time as an object of conception, where the component states of a relational predication are distributed along a one-dimensional axis. “Processing time” talks about time as a medium of conceptualization, i.e., the time it takes to conceive of something (relational or not). Now, the very apprehension of conceived time, within CG, cannot occur without the prior operationalization of processing time, which is also the time needed to establish the feeling of duration and objective relations between events. This makes processing time basic, immanent in all dealings with clausal (relational) predications, and conceived time parasitic on acts of mental organization. It would correspond to the claim in Husserlian phenomenology that objective time is constituted by phenomenological time, or through the constitution of concrete temporal objects. Such constitutive acts are always temporal, too (e.g., the perception of duration presupposes the duration of an act of perception; cf. Kokoszka 1996: 317). The linear nature of objective time can then be seen as arising from the projection of qualitatively distinct phases of representation from a subjective present that is neither singular nor strictly extended (in conceived time). The present, rather, exhibits an original “depth” whose orientation is perpendicular to the plane of objective time (cf. Figure 3; see also Guillaume 1965 for an actual application of similar ideas to the semantic analysis of tense).
References Brisard, Frank 1998 “The peripeties of localism: From Kant to Reichenbach”. In English as a Human Language: To Honour Louis Goossens, J. van der Auwera, F. Durieux and L. Lejeune (eds.), 40–56. Munich: LINCOM EUROPA. 1999 A Critique of Localism in and about Tense Theory. Doctoral dissertation, University of Antwerp. Comrie, Bernard 1985 Tense. Cambridge: Cambridge University Press. Cutrer, Michelle 1994 Time and Tense in Narratives and Everyday Language. Doctoral dissertation, University of California, San Diego.
39
40
Frank Brisard
Dahl, Östen 1985 Tense and Aspect Systems. Oxford: Blackwell. Dinsmore, John 1991 Partitioned Representations: A Study in Mental Representation, Language Understanding and Linguistic Structure. Dordrecht: Kluwer. Fauconnier, Gilles 1994 Mental Spaces: Aspects of Meaning Construction in Natural Language. [Augmented paperback edition.] Cambridge: Cambridge University Press. Fleischman, Suzanne 1982 The Future in Thought and Language: Diachronic Evidence from Romance. Cambridge: Cambridge University Press. Guillaume, Gustave 1965 Temps et verbe: Théorie des aspects, des modes et des temps; suivi de L’architectonique du temps dans les langues classiques. Paris: Champion. Husserl, Edmund 1952 Ideen, Volume 2: Phänomenologische Untersuchungen zur Konstitution. The Hague: Martinus NijhoŸ. 1970 The Crisis of European Sciences and Transcendental Phenomenology: An Introduction to Phenomenological Philosophy. Evanston: Northwestern University Press. 1990 On the Phenomenology of the Consciousness of Internal Time (1893–1917). Dordrecht: Kluwer. Kokoszka, Valérie 1996 “La conception husserlienne de la temporalité entre 1905 et 1910”. Tijdschrift voor Filosoªe 58: 314–341. Langacker, Ronald W. 1991 Foundations of Cognitive Grammar, Volume 2: Descriptive Application. Stanford: Stanford University Press. 1997 “Consciousness, construal, and subjectivity”. In Language Structure, Discourse and the Access to Consciousness, M. I. Stamenov (ed.), 49–75. Amsterdam: John Benjamins. McCawley, James D. 1971 “Tense and time reference in English.” In Studies in Linguistic Semantics, Ch. Fillmore and D. T. Langendoen (eds.), 97–113. New York: Holt, Rinehart and Winston. Miller, Izchak 1984 “Husserl’s account of our temporal awareness”. In Husserl, Intentionality and Cognitive Science, H. L. Dreyfus (ed.), 125–146. Cambridge, MA: The MIT Press. Partee, Barbara 1973 “Some structural analogies between tenses and pronouns in English”. Journal of Philosophy 70: 601–609.
Ordinary time
Pinker, Steven 2003 “Language as an adaptation to the cognitive niche”. In Language Evolution, M. H. Christiansen and S. Kirby (eds.), 16–37. Oxford: Oxford University Press. Reichenbach, Hans 1947 Elements of Symbolic Logic. New York: Free Press. 1958 The Philosophy of Space and Time. New York: Dover Publications. Traugott, Elizabeth Closs 1989 “On the rise of epistemic meanings in English: An example of subjectiªcation in semantic change”. Language 65: 31–55. Wilson, Deirdre and Sperber, Dan 1998 “Pragmatics and time”. In Relevance Theory: Applications and Implications, R. Carston & S. Uchida (eds.), 1–22. Amsterdam: John Benjamins.
41
Performatives as constatives vs. declarations Some recent issues Robert M. Harnish Departments of Philosophy & of Linguistics University of Arizona, Tucson, Arizona, USA
1.
Introduction
Performatives exhibit a special tension between semantics and pragmatics.1 Any theory of them must explain their ‘performative force’ as marked by the performative element of the sentence, within the framework of a compositional semantics. An account of performatives would be easy, or at least easier, were we to give up one or the other. If performative utterances were always just constative in their literal and direct force, then the grammatical form and compositional principles of performative sentences could be straightforwardly declarative (or truth-valuable). On the other hand, if we ignore compositionality, then their performative force could be given by special conventions of force that attach to the performative element in (just) the performative sentence. The problem of performatives is to account for their ‘performative character’2 within a general theory of how such utterances can be communicatively successful (how the intended force of such utterances can be reliably recognized by the hearer), while retaining their ‘compositionality’ or, more precisely, their “innocently compositional” semantics. We distinguish “compositionality” (the meaning of a complex expression is a function of the meaning of its constituents plus their syntactic relations) from “innocence” (an expression makes the same range of meaning contributions to all expressions it is a constituent of as it does standing alone).3 Applied to performatives, the claim of innocent compositionality would be that ‘promise’ in I promise I’ll be there means the same thing as in I promised to be there or He promises to be there. How then to account for the performative force of ‘I promise’ compositionally if it simply
44
Robert M. Harnish
encodes the rules for promising? And how to keep the performative force of ‘I promise’ from infecting I promised to be there and He promises to be there (see Harnish 2002, 1997, 1988, 1979, and 1976)? Although there are at least eight possible (and seven actual) theories of performatives (see Harnish 2002 and 1988), current work seems restricted to one of three major categories: 1. Performatives as, well, performatives, i.e., performatives as governed by special conventions of meaning or use which determine that their literal and direct utterance constitutes the performance of the act indicated by the performative device (see Austin 1962, Searle 1969, Reimer 1995, and Martinich 2002). We can call this the Performative Preªx view. 2. Performatives as literal and direct constatives, but standardized indirect illocutionary acts (see Bach 1975, Bach & Harnish 1979, Harnish 1988, Bach & Harnish 1992, and Harnish 2002). We can call this the Standardized Indirect view. 3. Performatives as declarations, i.e., as acts performed by declaring that one is performing them (see Recanati 1987 and Searle 1989). We can call this the Declarational view. Arguments for these positions are typically buttressed by arguments against the others. Here we will survey two sets of recent arguments regarding ‘constative’ and ‘declarational’ theories, which seem to be in the service of promoting the ªrst type of theory.
2.
Against performatives as constatives
Recently, Grewendorf (2002: 30) has presented a number of arguments “against the claim that performatives constitute the implicit performance of the illocutionary act of stating…”. Although he mentions only Searle (1989) at this point in his discussion, he earlier classiªed Bach & Harnish (1979) (and Harnish 1997, 2002) also as an “assertive approach” (2002: 29). First Argument: (1) If, in issuing a performative, S is both constating and F-ing [where ‘F-ing’ is the act named by the performative], then the performative would be subject to two diŸerent sets of felicity conditions. (2) It should be possible for one set of conditions to be fulªlled and not the other.
Performatives as constatives vs. declarations
(3) So it should be possible for the conditions on F-ing to be fulªlled, but not those for stating. (4) “If this situation can arise in principle, then it provides us with counterexamples to the claim that performatives are statements.” (2002: 33)
We need to look at the last and the second steps of this argument. First, step (4) claims that if (3) is true, then performative utterances are not statements. This does not follow. However, it would follow if (1) is read as: (1') If, in issuing a performative, S is always both stating and F-ing, then… (as before)
That is, (1'), but not (1), would be false if (3) were true, just as (4) claims. But if (1) is read as (1'), then (2) does not follow. Indeed, it follows that (2) must be false. So the argument form is defective. Second, step (2) might be true if the stating and the F-ing were independent acts. But if the F-ing is dependent in some way on the stating, then although one might state without F-ing, one might not be able to manage F-ing without stating. One way of reading the original Bach & Harnish (1979: Chapter 9) proposal is just this way. Performative utterances are a species of indirect acts, and the indirect act is performed via or by way of the direct act. In our original inference it was via the presumed truth of the statement that the F-ing was achieved. So there is a straightforward sense in which the stating and the F-ing are not independent, such that assumption (2) is not satisªed. In reply, someone might say that this may be true from H’s point of view, but that the argument proceeds from S’s point of view.4 And although the theory says that communicative uses of performatives involve the intentionexpectation that H reason from the direct to the indirect act (via the presumption of truth and standardization), the theory must allow that from S’s point of view, S can, in uttering a performative noncommunicatively, be: (SNF) (FNS)
stating, but not F-ing; F-ing, but not stating.
However, so the argument continues, the theory, by characterizing only communicative and conventional illocutionary acts, leaves the case of noncommunicative, nonconventional acts undeªned. Suppose we agree that the theory needs to account for noncommunicative, nonconventional illocutionary acts, and suppose in addition that ‘expressing’ an attitude is logically related to ‘being committed’ to having it, and that one could, in uttering a performative, be committing oneself to the attitudes expressed, even if there is no communicative
45
46
Robert M. Harnish
intention. But now, the objector replies, the stating and the F-ing are independent (though simultaneous) in performative utterances — one is both stating and F-ing, and so (SNF) and (FNS) should both be possible. Yet in uttering a performative these do not both seem to be possible — one cannot be doing just one, or just the other. The answer might be that, because the form has become standardized for F-ing in (con)stating, it is very di¹cult to ªnd circumstances where (SNF) is the case, and (FNS) is ruled out. Second Argument: (1) According to Searle (1969), a preparatory condition on assertions is that S have evidence for the truth of what S asserts. (Grewendorf 2002: 33) (2) “it is… highly unlikely that [the utterer of a performative] when confronted with a response ‘How do you know?’ would be able to give any reasons for the truth of the performative.” (2002: 33)
And more generally: (3) “a further argument… has to do with the fact that we do not react to performatives in the way we react to statements.” (2002: 33)
The argument proves too much. Here are three considerations regarding (1) and (2). First, note that it is not always true that any statement can be responded to naturally by How do you know? in normal conditions — what would you say if someone responded How do you know? to your saying: a. Hi. Let me introduce myself. My name is Tom. b. The shirt costs $5, you gave me $10, so I owe you $5. c. A: Would you like some breakfast? B: No thanks, I already had some.
So even if one could not oŸer a diagnosis of why it might be unnatural to respond to a performative with How do you know?, if that were enough to show they are not statements, that would be enough to show the last sentences in (a) — (c) above are not statements, either. But they are. Behind this point is the qualiªcation made earlier. Bach and Harnish say that the direct illocutionary force of a performative is constative, not necessarily speciªcally an assertion or statement. That means it only needs to meet the minimal requirements for the genus, which in this case would involve expressing (committing oneself to having) a certain belief (Searle 1969 and Alston 2000: committing oneself to the truth of the proposition expressed). But the genus also includes besides epistemically loaded assertions, epistemically neutral hypotheses and guesses. Note that in (1) of the Second Argument, ‘evidence’ is cited, while in (2) ‘reasons’ are. These are not the same. In examples (a) — (c) it is not unnatural,
Performatives as constatives vs. declarations
if pressed, to say that one has reasons for, e.g., believing one’s name is ‘Tom’ or that one has had breakfast, but the request for ‘evidence’ seems out of place. Second, on Bach and Harnish’s view, performatives are cases of standardized indirection, so it is legitimate to ask if the argument proves too much by also counting against noncontroversial (i.e., nonperformative) cases of this phenomenon, such as: a. I’d like the salt. b. Direct: ‘statement concerning S’s desire for the salt’ c. Standardized Indirect: ‘requesting H to pass the salt’
Notice how unnatural the response to (a) How do you know? is. So does that show that no statement has been made? No. Third, performatives can sometimes be followed by remarks suitable for constatives: A: I promise to be there. B: Is that true? Do you promise to be there?
Regarding (3) of the Second Argument, we should not expect performative utterances to have generally the same appropriate reactions as statements because they are also F-ings: A: I promise to be there. B: I hope that’s true (that you promised to be there), maybe you really will be there, and that (for you to be there) would be good.
Third Argument: (1) It should be possible to turn a performative such as I order you to leave into an explicit statement by adding the comment and that’s a statement, and the act of ordering should remain unaŸected. (Grewendorf 2002: 34) (2) “However, this is obviously not the case.” (2002: 34)
Grewendorf gives the following exchange as an example:5 /26/ a. I order you to leave the room. And that’s a statement. b. Thank God.
It is not clear what this example is supposed to show in support of the above argument. Although it is rather unnatural, the reply could be appropriate to just a statement (i.e., ‘thank God you are not ordering, only stating, because I don’t want to leave’), or appropriate to an order (i.e., ‘thank God you are
47
48
Robert M. Harnish
ordering me to leave the room, that’s what I was hoping for’). But it is certainly odder if the utterance is taken to be both a statement and an order. Is this a problem? Again, to test its force against the standardized indirection approach, let’s compare it with the following: a. Could you pass the salt? [core] And that’s a question. [tag] b. Thank God.
This is pretty odd, too. So it looks like explicating the force by appending the performative noun as a tag interacts with the standardized force of the core sentence it is appended to. And this is what one would expect; if the core sentence is standardized for F’-ing, but literally and directly used to F, then trying to use it to F’ by highlighting and emphasizing F poses a con¶ict of salience. Using the standardized form makes F’ most salient and contextually appropriate — appending the tag competes with that and so is pragmatically odd. Note that we do not want this kind of argument to prove too much. The following is odd: *I order you to leave the room. And that’s an order.
But that does not show that the core was not used to order. The moral is that the oddity of tags can be explained in many ways, so the brute fact of oddity does not show much about performativity and constatives. Fourth Argument: (1) If the force of /27/ were a statement that I was ordering you to leave the room: /27/ I order you to leave the room, then I would not be able to make the force fully explicit by using the explicit performative device: /28/ I (hereby) state that I (hereby) order you to leave the room (2) because the full explicit force of /28/ would involve: /28'/ I (hereby) state that I (hereby) state that I (hereby) order you to leave the room,
and so on ad inªnitum. Here are two considerations against this. First, the argument from /27/ to /28/ contains an unstated Austinian premise which is not argued for or otherwise defended:6 (MFE) The force of an utterance can always be made fully explicit by the addition of a suitable performative phrase.
Performatives as constatives vs. declarations
But this need not be true. Maybe there is no performative expression that just ªts the force of the utterance. For one example, consider minor moods such as the tag-imperative: Drive on my lawn will you. [POW!]
Or maybe there is no performative expression that makes explicit just the force of the utterance. For example, consider the illocutionary act of threatening, which cannot be made explicit by the performative: *I (hereby) threaten you (that P).
The upshot is that (MFE) does not seem to be true, and without the (MFE) principle, the regress is blocked at the ªrst step. And why should (MFE) be true? Austin himself noted that: explicitness, in our sense, makes clearer the force of the utterances… The explicit performative formula, moreover, is only the last and ‘most successful’ of numerous speech devices which have always been used with greater or less success to perform the same function… (Austin 1962: 73; emphasis added)
One of these “numerous speech devices… used… to perform the same function” Austin lists is mood, with the introduction: Consider for a moment some of these other more primitive devices in speech, some of the roles of which can… be taken over by the device of the explicit performative. (Austin 1962: 73; emphasis in the original)
The suggestion here is that there are a variety of devices for indicating what one is doing, and among these is mood — a position also taken about 45 years earlier by Frege: We have no particular clause in the indicative sentence which corresponds to an assertion, that is something is being asserted lies rather in the form of the indicative. (Frege 1918: note 3)
Second, even if one modiªed the argument so as not to require (MFE), one might still reply: so what? The regress initiated in /28/ and /28'/ is not vicious, it just iterates indeªnitely (see below). Fifth Argument: (1) In speaking of what we do here and now, we use the continuous present. (Grewendorf 2002: 35) (2) So if a performative were an utterance about what I am now doing, it should have the form: /29/ I am ordering you to leave the room.
49
50
Robert M. Harnish
(3) The continuous present in performatives is limited to such cases as: /30/ a. I am warning you, which “constitute a ‘colloquial emphatic use’ and thus is only exceptionally possible and limited to a few performative verbs” (2002: 35).
In reply, we can ªrst ask whether the generalization in (1) is true? Drawing a conclusion from premises seems to be an activity, and one that can be engaged in now, yet it is ok to say the ªrst, but odd to say the second: a. I conclude from P & Q that P. b. *I am concluding from P & Q that P.
So we cannot infer, just straightaway, that if something resists the present continuous it does not report a present activity. Second, it is not clear if Grewendorf thinks the upshot of this point about /30a/ is that /29/ is not possible, or that it is not a ‘colloquial emphatic use’, or whether he thinks /29/ is possible but most other performatives are not. In any case, it is true that performatives resist the present continuous. But are these continuous forms really “only exceptionally possible and limited to a few performative verbs”? Grewendorf gives no argument or examples, citing Vendler (1972). Since no arguments or examples are in play, let’s propose a contrary generalization: (CG) For any ‘performative’ verb, it is possible to ªnd a sentence in the present continuous containing that verb, and a context, such that uttering that sentence in that context would not be unnatural or odd. It is a good question why performatives resist free use of the present continuous, but as yet it does not tell against the standardized indirect theory. In summary, then, none of the arguments listed tell seriously against the standardized indirect theory, though there are some data left to explain.
3.
Against performatives as declarations
According to Grewendorf (2002: 36), the proposal that performative utterances “constitute the implicit performance of the illocutionary act of declaring” has at least three problems: 1. Can the act of declaring, performed in issuing a performative, be made explicit? It should be:
Performatives as constatives vs. declarations
since every implicit performative utterance can be turned into an explicit performative utterance (with the exception of a few illocutionary acts such as insulting). (Grewendorf 2002: 36)
It is not clear where this principle comes from, and we have already seen that a version of it (MFE; cf. Section 2) has many problems. Grewendorf goes on to say that, if we try to explicate the force of a performative such as: I (hereby) order you to leave the room,
then we get: /34/ I (hereby) declare that I (hereby) order you to leave the room.
Now, /34/ is also a performative, so it should have a force that can be made explicit, and making that explicit would yield: /34'/ I (hereby) declare that I (hereby) declare that I (hereby) order you to leave the room…
and so on: Even though Searle notes the problem of such a regress, he does not oŸer a solution. (Grewendorf 2002: 37)
However, the reason why Searle does not oŸer a ‘solution’ may be that he does not see a problem. Just pointing out a regress is not enough, because not all regresses are vicious — the regress generated by the ‘KK’ principle in epistemology, for instance, is not vicious: (KK) I know that P → I know that I know that P → I know that I know that I know that P… Grewendorf needs to argue that /34'/, unlike (KK), is vicious. He also needs to motivate the regress in the ªrst place. One could reply to this argument the way one might reply to SchiŸer’s (1972) analogous argument (cf. Harnish 1976), that the ‘regress’ in question never gets past the ªrst step, i.e., that /34/ is all the explicit declaring you need. 2. A further problem with Searle’s proposal is said to concern a diŸerence between linguistic and nonlinguistic declarations. With nonlinguistic declarations, the performance of the declaration constitutes the performance of the act ‘declared’:7
51
52
Robert M. Harnish
(NLD) Declaring the meeting adjourned constitutes adjourning the meeting. But with linguistic declarations this is not so: (LD) Declaring that one is ordering H to leave does not constitute ordering H to leave. Grewendorf concludes: declaring is not an illocutionary act per se, it is rather a way of performing illocutionary acts. But to say that an illocutionary act has been declaratively performed says no more than that it has been explicitly performed… This asymmetry… calls into question that the latter [linguistic declarations] are declarations at all. (2002: 37; emphases in original)
On Searle’s behalf, one might say that he does not (or at least need not) say that declaring is an illocutionary act. He need only say that it is a category of illocutionary acts. Perhaps Grewendorf’s point is that, unlike other illocutionary categories, declaring has the particular feature that in order to perform a linguistic declaration one must also perform some other illocutionary act. However, from that it is hard to see why the existence of linguistic declarations is called into question. Grewendorf might be working on the analogy of indirect speech acts. One can request H to open the window indirectly by asserting It’s hot in here. Indirect requests are ways of requesting, not another category of directives. The disanalogy is that Searle oŸers a deªnition of declarations in terms of illocutionary point and word-to-world direction of ªt, just like other categories of illocutionary act. 3. If the utterance of a sentence such as: /27/I order you to leave,
is both a declaration and an order, then the success conditions for both acts would have to be fulªlled. And, according to Grewendorf, it should be possible in principle that, under certain circumstances, the conditions for the directive illocutionary act are fulªlled while the conditions for the declaration are not. However, he thinks that this situation cannot occur on Searle’s analysis: If the conditions for the act of ordering are fulªlled, then the explicit performative utterance (27) constitutes the performance of the act of ordering. Thus, the conditions for the act denoted by the performative verb and the conditions for the declaration must be identical in the case of performatives. But if the conditions for two illocutionary acts are identical, we have to conclude that the two illocutionary acts are identical. (Grewendorf 2002: 38)
Performatives as constatives vs. declarations
This argument is di¹cult to sort out. We do not want it to depend on the fact that distinct properties or features of X might, as a matter of fact, always be coinstantiated. There is nothing problematic about that — consider Quine’s case ‘normal creature with a heart’ and ‘normal creature with a kidney’. If Searle’s declarational theory of performatives is correct, each performative utterance will be both a declaration and something else (denoted by the performative verb). So far no problem; one utterance (or sentence token), with two types of illocutionary act performed (again, think of indirect speech acts as an analogy). Just as in Quine’s example, the set of (normal) creatures-with-a-heart is the same as the set of (normal) creatures-with-a kidney. But from this it does not follow that the conditions for the instantiation of the property of having a heart are identical to the conditions for having a kidney. They have diŸerent ‘deªnitions’. Likewise with ordering and declaring, which have diŸerent deªnitions. In that sense the conditions for the directive illocutionary act can be fulªlled while those for a declaration are not fulªlled, as in Leave the room. But when performatively performed, they overlap (just as creatures with hearts and kidneys). Martinich (2002: 93) begins by claiming that Searle (1989) “abandons the analysis of speech acts that he had presented in Speech Acts and Expression and Meaning”. However, since Searle’s (1969) theory was basically designed for performatives (see Harnish 1979), we will want to see if he can make this claim stick. Martinich contrasts Searle’s and Austin’s use of ‘performative’. For Austin, there are both implicit (mood) and explicit performatives, the latter of which are typically in the present tense and nonprogressive aspect, with the performative element as the main verb. For Searle, only Austin’s ‘explicit performatives’ count, which can have progressive aspect and for which the performative element can be a noun. Searle contrasts: The meeting is over. (nonperformative) The meeting is adjourned. (performative),
and Martinich (2002: 94) thinks this is counterintuitive: “If there is going to be a category of performatives, ‘Leave the room’ intuitively is as performative as ‘I order you to leave the room’.” However, one needn’t share this intuition.8 Since their syntax and semantics are diŸerent,9 why can’t their pragmatics, and so their ‘performativity’, also be diŸerent? According to Martinich, Searle improves on two aspects of Austin’s theory. First, Searle’s schema for illocutionary acts, F(P), resolves them into force and (propositional) content, thereby eliminating the locutionary act (Martinich 2002: 95). Second, Searle modiªes and improves Austin’s illocutionary taxonomy. This modiªed theory:
53
54
Robert M. Harnish
1. is guided by general principles (Martinich 2002: 95); 2. contains several related asymmetries between F and P, including: (a) the ‘basic asymmetry’: “each proposition had at most one illocutionary force per utterance… illocutionary forces were mutually excluding, always semantically and sometimes even syntactically.” (Martinich 2002: 95–96), (b) a performative could never be the content of another performative (Martinich 2002: 96). 3. Paradigmatic performatives were ªrst-person singular active nonprogressive. (Martinich 2002: 96) Martinich gives no citations for Searle believing these at the time of the works cited. More importantly, it is not clear what the ‘basic asymmetry’ amounts to. As worded, it says that, e.g., the proposition:10 (P)
,
can occur with just one force ‘per utterance’ — yet utterance of what (as propositions are not uttered)? A speaker might utter the sentence: (T) The bull is going to charge,
stating the proposition that the bull is going to charge, suggesting that H run, and/or perhaps warning H to run.11 According to Martinich, this earlier theory has the virtue of being capable of dealing with certain cases that undermined Austin’s conception of a performative, and Searle’s primary reason for thinking that performative utterances are declarations is simply that they satisfy his deªnition of them. To this idea Martinich has four objections. 1. Not all performatives satisfy the deªnition of a declaration: The sentence ‘I order you to leave the room’ names an order (which belongs to the category of commissives).12 Since it does not name a declaration, it cannot be used performatively to express a declaration. (Martinich 2002: 97)
But Martinich does not make it clear what the force is of its not being usable ‘performatively’ to express a declaration. Searle is claiming that performative utterances are declarations, not that all declarations are performative utterances. 2. A sentence of the form I state that… would be used to state: Since declaring is not ‘named by the performative verb’ but stating is, the illocutionary act performed cannot be declarative, but is an assertive, contrary to the explicit position held by Searle and Vanderveken. (Martinich 2002: 97)
Performatives as constatives vs. declarations
Again, this might count against the theory if all declarations were performatives, or if no utterance could have more than one force (the ‘basic asymmetry’). But Searle need not (and presumably does not) hold either of these views. 3. According to Searle, the content of I state that P is that ‘the speaker states that P’:13 This would trivialize truth and falsity for explicit performatives. (Martinich 2002: 98)
It is not clear why Searle is committed to this. The content of the declaration may be that S is stating, but that need not be the content of the stating (its content being P). Again, Martinich seems to be assuming that an utterance can have only one force and content (‘basic asymmetry’ again). 4. There is no principled reason to exclude other utterances than performatives, which are not declarations, from the category of declarations. Every utterance creates a linguistic fact, such that: All speech acts are declarations. So he [Searle] has not identiªed what is distinctive about performatives. (Martinich 2002: 98)
It is arguable that being a declaration is not what is supposed to be distinctive about performatives. They are declarations, on this theory, but what is distinctive about them on Searle’s theory is their lexical (they name the illocutionary act performed) and semantic properties (self-guaranteeing and self-referential). Martinich then adds at least two further considerations against the declarational theory of performatives. 1. None of I assert that P, I command that P, etc.: look like the form of a declaration; they look like forms of assertives, directives… (Martinich 2002: 99)
Martinich does not say why they cannot be both, e.g., assertions and declarations (‘basic asymmetry’). 2. Even if performatives were declarations: it would still be necessary to distinguish types of declaration. And the types that would emerge are the same as that appeared in the good old theory… (Martinich 2002: 99)
This is one way to subcategorize declarations, but there are others as well — what is the problem with the possibility of this taxonomy? This leads us to the next criticism.
55
56
Robert M. Harnish
3. Martinich (2002: 99) thinks that “the absurdity of reiterating the old taxonomy inside the new one” can be pressed home by pointing out that on Searle’s theory, an utterance of I command you to close the door does not only constitute a command and a declaration, but ‘derivatively’ a statement, too: How can someone, hearing this theory for the ªrst time, avoid asking, ‘why doesn’t a speaker just promise or state what she has in mind?’ Of course, if I am right, that is what a speaker does in using the relevant sentences. The word ‘declares’ is doing no work. (Martinich 2002: 99)
There are two points here. The ªrst just seems to be an appeal to what a naive hearer thinks the theory of performatives should look like. It is hard to see what the force of that is. The second, that the word declare is doing no work, is more obscure. In one sense that is correct, it is doing no work, because it does not appear in the sentence uttered. It does appear in the theoretician’s account of the utterance, and there it is doing theoretical work — it is assigning a host of features to the utterance. 4. According to Martinich, on Searle’s theory an explicit performative is directly a declaration and derivatively a promise, for instance. How, then, does one promise nonderivatively? [T]he person would have to say ‘I am going to the party’ with the word ‘am’ getting major stress, that stress expressing the illocutionary force. (Martinich 2002: 100)14
He thinks that this consequence is implausible: First, it provides no way to distinguish between statements of ªrm intentions and promises. (Martinich 2002: 100)
But then why should it? That distinction is not recorded in this sentence, since it is provided by context. Second, it puts ‘I am going to the party’ in the category of commissives, even though it looks like it belongs to the category of assertives. (Martinich 2002: 100)
Here, Martinich commits a category mistake. Sentences are not illocutionary acts, utterances of them are (or can be). An utterance of I am going to the party can be an assertive or a commissive or even an expressive — as the contextual case may be. In summary, then, neither Grewendorf (2002) nor Martinich (2002) has presented decisive arguments here against Searle’s view that performative utterances are declarations. Earlier the present author argued against Searle’s
Performatives as constatives vs. declarations
semantics and pragmatics for performatives (see Harnish 2002). It would have been interesting to have further arguments against his pragmatics for performatives as well, but this will have to wait for another occasion.15
Notes 1. For the purposes of this paper a performative sentence is one containing an explicit lexical indicator of illocutionary force (usually the main verb), and a performative utterance is the literal utterance of a performative sentence. 2. Here we include not only the fact that they are used to perform the act indicated by the performative element, but also that they introspectively “feel” as if they are just the performance of that act. 3. These are distinct principles (though occasionally they are confused), as indicated by the fact (or even the possibility) that Frege’s semantics for natural languages was compositional but not innocent, because “indirect” contexts require expressions to have “indirect” senses and references that they, arguably, do not have standing alone. 4. Actually, Grewendorf (2002) tends to adopt the hearer’s point of view. 5. Numbers in ‘//’ are from the original article. 6. Maybe this attribution to Austin should be qualiªed — see below. 7. Grewendorf gives marrying as his example, but in American English there is no performative declarational sentence for marrying. One has to say something like I declare you husband and wife, not *I declare you married. 8. Austin (1962), for one, did not: he distinguished ‘explicit’ from ‘implicit’ performatives, and Leave the room! was an ‘implicit’ performative. 9. For instance, one can consistently conjoin and that’s not an order to the ªrst, but not to the second, in issuing a directive. 10. Maybe Martinich meant to say this: each sentence has at most one illocutionary force. But sentences do not have illocutionary forces, they at best have illocutionary force potential — forces which their utterance would have under certain conditions (see Harnish 1994). If one moves from sentences to utterances of sentences, then one has to control for nonliterality, indirection, and force ambiguity, as in Time ¶ies (see Harnish 1995 for a sample list of force-ambiguous sentences). 11. We will see that Martinich assumes this ‘basic asymmetry’ in some arguments against Searle’s declarational theory, so that it is important to see its limitations. 12. For Searle (and Martinich two pages earlier), orders are directives, not commissives. 13. One could hold a declarational theory of performatives without holding that the above is the ‘o¹cial’ content of the utterance. For instance, on a Kaplan-Perry construal of indexicals, the referent of the indexical is the propositional constituent introduced (the person), not the mode of presentation of the person (being the speaker). See Perry (2001) for a review.
57
58
Robert M. Harnish
14. It is not clear why Martinich thinks this. The sentence could be used to make a promise without any special stress. Nothing in the sentence “expresses the illocutionary force” in any interesting sense of ‘express’. 15. Thanks are due to members of the Graduiertenkolleg, Goethe-Universität, Frankfurt (especially to Christian Plunze), and to Kent Bach, for helpful comments on this material.
References Alston, William P. 2000 Illocutionary Acts and Sentence Meaning. Ithaca: Cornell University Press. Austin, John L. 1962 How to Do Things with Words. Oxford: Oxford University Press. Bach, Kent 1975 “Performatives are statements too”. Philosophical Studies 28: 229–236. Bach, Kent and Harnish, Robert 1979 Linguistic Communication and Speech Acts. Cambridge, MA: MIT Press. 1992 “How performatives really work: A reply to Searle”. Linguistics and Philosophy 15: 93–110. Frege, Gottlob 1918/1994 “The thought: A logical inquiry”. In Basic Topics in the Philosophy of Language, R. M. Harnish (ed.), 142–160. Englewood CliŸs: Prentice Hall. Grewendorf, Günther 2002 “How performatives don’t work”. In Speech Acts, Mind and Social Reality, G. Grewendorf and G. Meggle (eds.), 25–39. Boston: Kluwer. Harnish, Robert M. 1976 “Two consequences of transparent subject position”. Philosophical Studies 30 (1): 11–18. 1979 “A projection problem for pragmatics”. In Syntax and Semantics, F. H. Heny and H. S. Schnelle (eds.), 315–342. New York: Academic. 1988 “Performatives are default re¶exive standardized indirect speech acts”. Acta Linguistica Hungarica 38: 83–106. 1994 “Mood, meaning and speech acts”. In Foundations of Speech Act Theory, S. L. Tsohatzidis (ed.), 407–459. London: Routledge. 1995 “Modularity and speech acts”. Pragmatics and Cognition 3: 1–30. 1997 “Performatives and standardization: A progress report”. Linguistische Berichte 8: 161–175. 2002 “Are performative utterances declarations?”. In Speech Acts, Mind and Social Reality, G. Grewendorf and G. Meggle (eds.), 40–53. Boston: Kluwer. Martinich, Aloysius P. 2002 “On the proper treatment of performatives”. In Speech Acts, Mind and Social Reality, G. Grewendorf and G. Meggle (eds.), 93–104. Boston: Kluwer. Perry, John 2001 Reference and Re¶exivity. Stanford: CSLI Publications.
Performatives as constatives vs. declarations
Recanati, François 1987 Meaning and Force: The Pragmatics of Performative Utterances. Cambridge: Cambridge University Press. Reimer, Marga 1995 “Performative utterances: A reply to Bach and Harnish”. Linguistics and Philosophy 18: 655–675. SchiŸer, Stephen R. 1972 Meaning. Oxford: Oxford University Press. Searle, John R. 1969 Speech Acts. Cambridge: Cambridge University Press. 1989 “How performatives work”. Linguistics and Philosophy 12: 535–558. Vendler, Zeno 1972 Res Cogitans. Ithaca: Cornell University Press.
59
‘First/second vs. third person’ and ‘ªrst vs. second/third person’ Two types of ‘linguistic subjectivity’ Yoshihiko Ikegami University of Tokyo
1.
The controversy over the animacy hierarchy
There is a well-known hierarchical structure, generally called the ‘animacy hierarchy’, constituted by certain nominal classes. Typically, this hierarchy is represented as follows: 1st Person — 2nd Person — 3rd Person — Human — Animate — Inanimate — Abstract
PRONOUNS
NOUNS
It is known, however, that there are two variants of this hierarchy, as shown below: 2nd Person — 1st Person — 3rd Person — Human — Animate — Inanimate — Abstract
PRONOUNS
NOUNS
1st Person 3rd Person — Human — Animate — Inanimate — Abstract 2nd Person
PRONOUNS
NOUNS
62
Yoshihiko Ikegami
Of the three schemes, the ªrst is proposed by Dixon (1979), among others, while the second scheme is preferred by Silverstein (1976) and the third by DeLancey (1981), for example. The diŸerence among the three proposed schemes concerns the relative ordering of ªrst and second person. The ªrst scheme ranks ªrst person higher than second person and the justiªcation for this ordering is sought in the claimed higher degree of “agentivity” of the ªrst person (Dixon 1979: 85). The second scheme reverses the ordering of ªrst and second person on the grounds of “politeness”, i.e. “respect” accorded to the conversational partner (Silverstein 1976: 118, cf. Levinson 1983: 83, Brown & Levinson 1987: 121–122 and Radden 1995: 430). The third scheme places ªrst and second person on an equal footing, because the two are conversational partners in their alternating roles as speaker and hearer (Delancey 1981: 646, Frawley 1992: 95). The animacy hierarchy is known to serve well in characterizing such typological features as the ‘ergative split’ (i.e. the split between nominals employed in ergative-absolutive sentence patterns and nominals employed in nominative-accusative sentence patterns), contrastive singular-plural marking for nominals, and partially also the ‘possession cline’ (i.e. the contrast between inalienable and alienable possession). There is, however, apparently no consensus as to which of the three varieties of ordering between ªrst and second person is most acceptable. One opts for ‘speaker over hearer’, another for ‘hearer over speaker’, and a third for ‘speaker and hearer as equals’. Thus Frawley (1995: 96) poses the following question: What factors determine the diŸerence between a language with speaker over hearer … and one with hearer over speaker…? …Very little is known in answer to these questions, though pragmatic and cultural factors may ultimately be responsible for these diŸerences. Does this suggest that there are two types of culture, one speaker-oriented and one hearer-oriented?
The present paper proposes to address this question.
2.
Contrasting ªrst/second person with third person
It has been widely recognized since Benveniste’s classical paper (1971 [1946]) on the nature of grammatical person that the relationship among ªrst, second, and third persons is not as neat as might be imagined from the sequential
Two types of ‘linguistic subjectivity’
numbering assigned to them. There is clearly a split between ªrst and second person, on the one hand, and third person, on the other. The former two persons involve speaker and hearer, interacting with each other as communicative partners in the context they both share, while third person is not a communicative partner of theirs. Thus the contrast we have here is characterized as ªrst/second person vs. third person. The contrast, ªrst/second person vs. third person, is very well grounded. On the one hand, we have active communicative partners, typically humans, interacting with each other through the use of language. On the other hand, we have non-active entities, not necessarily humans, which are only to be referred to by the speech-act participants. The former and the latter are thus totally diŸerent types of entity. In relation to the main theme of the present paper, it must be noted that the contrast is founded on the notion of language as a means of communication.
3.
Contrasting ªrst person with second/third person
Let me suggest that there is in fact another way of contrasting the three persons, in support of which ample linguistic evidence can also be adduced. The contrast in question features ªrst person vs. second/third person. Helmbrecht (1995: 287), for instance, who investigated the grammatical marking of the three persons across various languages, claims to have detected “a remarkable number of instances where a 1st person is grammatically opposed to 2nd/3rd person”. Such split marking is found, for example, when ªrst, but not second or third, person is marked for plural or dual or when, conversely, gender distinction is morphologically marked for second and third persons, but not for ªrst person. It is also pointed out that the same split tends to emerge in certain predicate categories rather than others: e.g. ‘future’ rather than ‘present/past’, ‘optative’ rather than ‘indicative’, ‘irrealis’ rather than ‘realis’, ‘volitional’ rather than ‘nonvolitional’, ‘direct evidence’ rather than ‘indirect evidence’, ‘assertive’ rather than ‘nonassertive’, and so forth (Helmbrecht 1995: 187–191).1 Helmbrecht seems to suggest in this connection that the prerogative marking of ªrst person as contrasted with second and third person derives from the construal of ªrst person as a ‘self-conscious subject’ (Helmbrecht 1995: 287) rather than as a ‘speaker’. I quite agree. Although Helmbrecht’s investigation does not cover Japanese, the present author’s native language, I can point to a number of features in Japanese which suggest that Japanese is a rather ‘ego-
63
64
Yoshihiko Ikegami
centered’ (i.e. tending toward a ªrst vs. second/third person contrast) language (cf. Ikegami 2001). In the next section, I am going to refer brie¶y to some of such features.
4.
Some ego-centered features of Japanese
4.1 Person restrictions on psychological predicates In Japanese, the subject of a psychological predicate is as a rule (in other words, unless some such stylistic device as free indirect speech is applied) restricted to ªrst person, as shown below:2 (1) (2) (3)
i am glad // i’d like to go. are glad // you’d like to go. ??s/he is glad // s/he’d like to go. ??you
The contrast in acceptability between ªrst and second/third person in Japanese3 is usually accounted for in terms of the general inaccessibility of a person’s private psychological processes (which are directly sensed by the person in question but not by others). More recently, the same point is discussed in terms of “evidentiality” (cf. Chafe & Nichols 1986: the speaker is not supposed to refer to things which he has neither perceived nor experienced — in other words, of which he himself has no evidence). On the other hand, the speaker of English (and of probably most of the Western languages, for that matter) seems to have few qualms about combining a psychological predicate with a second and a third person subject. There are conceivably two ways to account for this: either the speaker is able to split himself psychologically and objectify his counterpart in much the same way as he objectiªes anyone other than himself — in which case the speaker can as much qualify as alter as second or third person — or alternatively, the speaker is ready to empathize with anyone other than himself and is thus able to su¹ciently familiarize himself with any other person’s psychological states and processes. In consideration of other related linguistic features, I opt for the ªrst account. 4.2 A system of demonstratives based on the ‘ego’ – ‘alter’ contrast As is well attested, some languages have a three-tier system of demonstratives, while others have a two-tier system. It is conjectured that the three-tier system is based on the three-person contrast (i.e. ‘ªrst-person related’, ‘second-person
Two types of ‘linguistic subjectivity’
related’, and ‘third-person related’) on one hand, and that the two-tier system is based on the ‘ego’ – ‘alter’ contrast, i.e. ‘proximate (to ‘ego’)’ and ‘distal (from ‘ego’)’) on the other. Japanese is known to have a three-tier system, morphologically marked as ko – so – a. Shoho (1981), however, suggests that the Japanese system is actually an amalgamation of two two-tier systems: one, called contrast type (involving ko and so), which is based on the contrast between speaker and hearer, and the other, called merger type (involving ko and a), which is based on the contrast between speaker/hearer (i.e. inclusive we, hearers being grouped with speakers) and others (in other words, between speech-act participants and non-speech-act participants). Notice that in both systems it is the ego (associated with ko) that occupies the pivotal position. 4.3 The contrast between ‘come’ and ‘go’ and related contrasts Languages are known universally to have some linguistic means to distinguish between two types of deictic movement, typically exempliªed by the contrast between the verbs ‘come’ and ‘go’. It is also known that in its primary uses, some languages apply the verb ‘come’ to a movement toward ªrst or second person, while others restrict the use of the verb to a movement toward ªrst person only. Insofar as I have been able to conªrm this through relevant studies, languages like English, French, German, Croatian, Hindi, Tamil, Nepalese, and Chinese belong to the ªrst type, while languages like Japanese, Korean, Thai, Spanish, and Hungarian belong to the second.4 These two types can be characterized as based on the contrast between ‘speech-act participants’ (i.e. ªrst and second person) and ‘non-speech-act participants’ (i.e. third person), and on the contrast between ego (i.e. ªrst person) and alter (i.e. second and third person), respectively. In some languages, the same contrast is known to be extended to another pair of verbs, ‘bring’ and ‘take’. A peculiar feature of Japanese, however, is that the same contrast is extended to the verb pair ‘receive’ and ‘give’, and according to the survey of more than a hundred languages conducted by Newman (1996: 26–27), hardly any other language is found which extends the contrast as far.5
4.4 Self-address and self-reference It is pointed out in Kunihiro (1982) that in a diary, which is quintessentially a monologic type of discourse, (and, in fact, in speaking privately to oneself in general) Japanese speakers tend to address themselves in the ªrst person,
65
66
Yoshihiko Ikegami
while English speakers do so in the second person, suggesting that a splitting of the self (i.e. the self speaking to the self) is involved in the latter case. He also cites a number of passages in novels where the English originals have a second person in forms of self-address, while their Japanese translations have replaced the second with the ªrst person form, as in the following English sentence (Kunihiro 1982: 7): (4) No matter how objective you tried to be, Mel reasoned, it was hard to be sure of your own motive. (A. Haley, Airport) [The Japanese translation reads something like ‘… I tried to be, …my own motive’.]
As for self-reference, Hirose (1997, 2000) points out an interesting discrepancy between Japanese and English in the use of a kinship term when referring to the speaker. Thus in the English example (5), the speaker uses dad to refer to himself and the word dad here is grammatically treated as a third person, as testiªed by the forms of the predicate verbs (thinks … wants) that agree with it as well as by the use of the third-person co-referential pronoun (he). But in its Japanese equivalent, the word otosan (‘dad’) is treated as a ªrst person, as shown by the psychological predicates6 (omou ‘think’ and hoshii ‘would like to’) which accompany it: (5) Dad thinks it would be bad manners not to wash your hands before dinner. He wants you to be a good boy. [The Japanese translation would read something like ‘Otosan (Dad)… omou (think) …hoshii (would like to)’.]
4.5 Body-part nouns and re¶exives Hirose (1997, see also 2002) points out that there is a systematic contrast between the use of English re¶exive verbs and the ways they are rendered in Japanese. To the construction ‘transitive verb + re¶exive pronoun’ in English corresponds either the construction ‘transitive verb + body-part noun’ or ‘intransitive verb’ in Japanese, as in the following examples: hang oneself – hang neck, shave oneself – shave beard, wash oneself – wash body, kill oneself – cut off life, enjoy oneself – enjoy, pride oneself – pride, worry oneself – worry.
Thus Japanese speakers avoid using re¶exive constructions, which are explicit linguistic markers of the split self (Haiman 1999: 224).
Two types of ‘linguistic subjectivity’
The Japanese nominal zibun (roughly meaning ‘self’) is often referred to as a close equivalent of the re¶exive pronouns in Western languages, and a large number of contrastive studies on the use of this nominal, especially in comparison with the English re¶exive pronouns, have been produced (most recently, cf. Uehara 2002). Among a number of diŸerences in use between the Japanese and English re¶exive nominals is an interesting fact, noted, among others, by Hirose (2002: 392–393): while verbs of physical action in Japanese do not naturally take zibun as their grammatical object (as shown in the preceding paragraph), verbs of non-physical activities like perception and cognition may make good sense when combined with zibun. Thus, as regards Japanese, there is a series of contrasts like the following: *hang zibun *shave zibun *wash zibun *kill zibun
blame (semeru) zibun praise (homeru) zibun despise (keibetsusuru) zibun evaluate (hyoukasuru) zibun
In combination with zibun as object, verbs of non-physical activity (in the right column) sound far more natural than verbs of physical action (in the left column).7 Hirose accounts for this diŸerent behavior by saying that “in nonphysical domains like perception and cognition, you can get outside yourself and look at your self from outside, so the Subject and the Self can be separated” (Hirose 2002: 392).8 It seems to me, however, that a more accurate account is rather to be sought in what the same author discusses in the sentence which directly follows the one just quoted: “In these cases [i.e. cases like blame / praise / despise / evaluate zibun], the Self corresponds not to one’s body or body part, but to rather abstract entities such as a mental image of oneself, social roles, memories, past actions, values, day to day behavior, and so on, which the Subject can treat as external objects” (Hirose 2002: 392). It is not the case that zibun is used as a grammatical object because the Self is ‘easily separated’ here and thus readily objectiªable. The Self stays with the Subject even in the conceptualization of perceived and cognized events. It is simply that the Self is always the most salient entity in those perceived and cognized events and hence it is readily singled out in an event and linguistically encoded, metonymically standing for the whole event it is involved in.9
67
68
Yoshihiko Ikegami
5.
Motivations for the two types of contrast
In Section 4, I have reviewed some linguistic features of Japanese which suggest its markedly ego-centered orientation, together with some others which indicate a notion of the self as resistant to splitting. These two kinds of feature are obviously interrelated. In the use of language, the self typically manifests itself as ‘ego’, i.e. as one who is designated by the ªrst-person pronoun I.“‘ Ego’ is he who says ‘ego’,” as Benveniste (1971: 224) noted. The survey by Helmbrecht (1995), together with my characterization of Japanese as an ego-centered language oŸered above, does seem to show that the ªrst- vs. second-/third-person contrast is as important a typological parameter of language as the ªrst-/second- vs. third-person contrast. It is tempting to speculate about what motivations underlie these two diŸerent types of contrast. The motivation for the ªrst-/second- vs. third-person contrast was already oŸered by Benveniste in his classical paper on person (Benveniste 1971). First and second persons are grouped together because they constitute speech-act participants. As speech-act participants, they are on an equal footing, alternating in their roles as speaker and hearer. It will no doubt be this conceptual schema which underlies the type of animacy hierarchy which opts for equating ªrst and second persons (cf. Section 1). By contrast, the motivation for the ªrst- vs. second-/third-person contrast can best be sought in the characteristically ego-centered way in which humans cognize the world. The person as cognizer is the point of origin, in relation to which he perceives and conceptualizes his environment. Thus the underlying contrast here is ‘ego’ vs. ‘alter’. I suggest that it is this contrasting schema which underlies another variety of animacy hierarchy, in which ªrst person is placed higher than second and third persons. It is to be noted that the latter as well as the former schema is deeply involved in the use of language. The schema ªrst/second vs. third person, as we have seen, is involved in the use of language as a means of communicative interaction. The other schema, ªrst vs. second/third person, however, concerns another aspect of the use of language, namely the process of construing the situation to be encoded before proceeding to encode it in the form of a linguistic message. In construing a situation, the speaker as ego is the point of origin, in relation to which everything else which the speaker ªnds relevant is to be construed and duly encoded. The basic contrast here is thus between ego and alter.
Two types of ‘linguistic subjectivity’
It will be noticed that the contrast just discussed can also be characterized as one between a dialogic use of language and a monologic use of language. A dialogue naturally presupposes the speaker and the hearer; a monologue, by contrast, does not presuppose the hearer. The person engaged in a dialogue is typically playing the role of the speaker; the one engaged in a monologue can be compared to the speaker construing the situation to be encoded — he is in the role of the cognizer, as cognitive linguists would call it. The speaker is concerned with ‘speaking’, the use of language par excellence, while the cognizer is concerned with ‘thinking’, or more exactly, what Slobin (1996) calls, “thinking for speaking”. It is to be admitted that the thinking for speaking process will result in diŸerent formulations of thought, depending on whether the “speaking” here is meant as a dialogue or as a monologue. For my present purposes, I would like to concentrate on the monologic type of thinking for speaking. Such a type of thinking for speaking will naturally be characterized by a markedly ego-centered construal, because of the absence of the hearer as a communicative partner. Now, inner speech may be considered the prototypical instance of the discourse type so characterized, and there are certain linguistic features which researchers have noted about this particular discourse type: “extreme elliptical economy” (Vygotsky 1962: 65), “egocentrism” and “predicativity”10 (Luria 1982: 107–108), and “telegraphic” style (McNeill 1987: 96). It is interesting to note that all these features concern either the ego-centered construal or a relatively high degree of indulgence, on the part of the encoder, in elliptical encoding — either of which is allowed in the absence of an addressee. Although I have no space to oŸer elaborate illustrations, these are in fact wellrecognized features commonly attributed to Japanese discourse in general.
6.
Further questions and some speculations
The foregoing consideration invites us to speculate on the evolution of language. If, as is generally claimed by semiotic scholars (e.g. T. A. Sebeok), language evolved ªrst in its cognizing function (i.e. that of conceptualizing the world) before it was expropriated as a means of communication, it should be possible to ªnd in language vestiges of monologic orientation (e.g. language functioning as a means of self-expression, solitary meditation, egocentric, subjective slant, hearer/reader responsibility11) as well as of dialogic orienta-
69
70
Yoshihiko Ikegami
tion (e.g. language functioning as a means of communication with others, cooperative, intersubjective slant, speaker/writer responsibility). I presume that languages may very well diŸer in the extent to which they have evolved dialogically oriented features (or alternatively, in the extent to which they still retain monologically oriented features). Watching chimpanzees learning a ‘language’ through picture-like symbols, one gets the impression that they are still operating at the lowest stage of monologically oriented language. Some researchers in this ªeld (e.g. Tomasello 1999) suggest that what chimpanzees lack in order to have their ‘language’ develop into a full means of communication, is the behavior of ‘joint attention’, i.e. paying attention to the same object or situation which engages other individuals, thus sharing attention. Human infants, on the other hand, are known to manifest this behavior quite early in their development, ªrst through the gesture of pointing and then by the use of vocal signs. This characteristically human behavior is presumed to have helped develop ‘culture’, as an aggregate of assets to be shared and to be transmitted from one generation to another. On the basis of all these considerations, ªnally, we are now able to evaluate properly the one remaining scheme of animacy hierarchy which we have not yet discussed, namely one which places second person over ªrst person. This one represents a further development of the scheme which places ªrst and second person equally as communicative partners. Successful communication presupposes, at the very least, cooperation on the part of the hearer. And to be able to elicit cooperation, the speaker needs to pay the hearer due respect — a notion claimed to underlie the hearer-over-speaker order in the animacy hierarchy. This can also serve as an answer to the apparently puzzling question: “Why does Japanese, an allegedly ‘polite’ language, not operate on the principle of ‘hearer-over-speaker’ rather than ‘speaker-over-hearer’?” The answer suggested by the present paper is that there is another, competing motivation strong enough to override the hearer-over-speaker principle, namely the markedly ego-centered cognitive orientation.
Notes 1. Notice that common to all these contrastive cases is the fact that the first item of each pair represents a grammatical category which closely reflects the speaker’s subjectivity. 2. Note that, here and elsewhere in the article, example sentences in English in small capitals are meant to refer to original Japanese sentences, of which they are rough equivalents.
Two types of ‘linguistic subjectivity’
3. It is tantalizing to know what other languages behave in this respect in the same way as Japanese. Korean seems to behave in the same way, but not Chinese, for instance. 4. I am extremely curious to know why Spanish, unlike other related European languages, behaves ‘egocentrically’ in this respect. 5. Note also the extensive quasi-auxiliary use of the ‘come’ and ‘go’ verbs and of the ‘receive’ and ‘give’ verbs, and the high degree of lexical diversification of the latter pair of verbs in Japanese. Because of space limitations, I am unable to go into the details of these uses. 6. As discussed in section 5.1, only first-person subjects can combine with psychological predicates in Japanese. 7. The degree of naturalness seems to vary with individual verbs. Thus verbs like respect (sontyoosuru) and assert (syutyoosuru) do not seem to me to go as well with zibun, although they are also attested by Hirose. 8. The terms ‘Subject’ and ‘Self’ are used in Hirose’s account in reference to Lakoff (1996): “the Subject is the locus of consciousness, and the Self is the rest of the person” (Hirose 2002: 391). 9. It may help in this connection to think of a series (represented in English as a metalanguage) like the following: to blame that someone did something → to blame someone’s doing something → to blame someone for doing something → to blame someone. Cf. Ikegami (1981, 1991), where a contrast like this is discussed in terms of ‘kotooriented’ and ‘mono-oriented’ encoding. (Mono refers prototypically to a well-defined concrete entity, or ‘first-order entity’, while koto refers to an event, state, situation, proposition, or ‘second- and third-order entity’. It is argued that typologically, there are languages which tend towards ‘koto-oriented’ encoding (like Japanese) and languages which tend towards ‘mono-oriented’ encoding. 10. By ‘predicativity’ Luria means that the predicate is more important than the subject of the sentence. This also tallies with another feature of Japanese, where the grammatical subject often remains implicit in discourse. 11. Cf. Hinds (1987), who proposes that there are cultures in which success in communication is considered to depend on the way the speaker/writer chooses to behave, and cultures in which the same responsibility is attributed to the hearer/reader. As examples for the latter type, Hinds cites Japan and ancient China.
References Benveniste, Emile 1971[1946] “Relationships of person in the verb”. In Problems in General Linguistics, 195–207. Coral Gables: University of Florida Press. Brown, Penelope & Levinson, Stephen C. 1987 Politeness. Cambridge: Cambridge University Press. Chafe, Wallace & Nichols, Johanna (eds.) 1986 Evidentiality: The Linguistic Coding of Epistemology. Norwood: Ablex.
71
72
Yoshihiko Ikegami
DeLancey, Scott 1981 “An interpretation of split ergativity and related patterns”. Language 57: 626–657. Dixon, Robert M. W. 1979 “Ergativity”. Language 55: 59–138. Frawley, William 1992 Linguistic Semantics. Hillsdale: Lawrence Erlbaum. Haiman, John 1999 “Grammatical signs of the divided self: A study of language and culture”. In Cultural, Psychological and Typological Issues in Cognitive Linguistics, M. K. Hiraga, C. Sinha and S. Wilcox (eds.), 213–234. Amsterdam: John Benjamins. Helmbrecht, Johannes 1995 “The typology of 1st person marking and its cognitive background”. In Discourse Grammar and Typology: Papers in Honor of John W. A. Verhaar, W. Abraham, T. Givón and S. A. Thompson (eds.), 285–297. Amsterdam: John Benjamins. Hinds, John 1987 “Reader versus writer responsibility: A new typology”. In Writing across Languages, U. Connor and R. Kaplan (eds.), 141–152. Reading: Addison-Wesley. Hirose, Yukio 1997 “Hito o Arawasu Kotoba to Shoo [Person-referring expressions and anaphora]”. In Shiji to Shoo to Hitei [Reference, Anaphora, Negation], Y. Hirose and N. Kaga, 1–89. Tokyo: Kenkyusha. 2000 “Public and private self as two aspects of the speaker: A contrastive study of Japanese and English”. Journal of Pragmatics 32: 1623–1656. 2002 “Viewpoint and the nature of the Japanese re¶exive zibun”. Cognitive Linguistics 13: 357–401. Ikegami, Yoshihiko 1981 SURU to NARU no Gengogaku: Gengo to Bunka no Taiporozi eno Shiron [Linguistics of DOING and BECOMING: An Essay towards a Typology of Language and Culture]. Tokyo: Taishukan. [A German version will be published by LIT Verlag, Berlin.] 1991 “DO-language and BECOME-language: Two contrasting types of linguistic representation”. In Empire of Signs: Semiotic Essays on Japanese Culture, Y. Ikegami (ed.), 285–326. Amsterdam: John Benjamins. 2001 “Indices of a subjectivity-prominent language: Between cognitive linguistics and linguistic typology”. Plenary lecture given at the 7th International Cognitive Linguistics Conference, Santa Barbara. Kunihiro, Tetsuya 1982 “Ninsho no Yoho to Kozo: Nichi-Eigo o Taishoshite [The use and structure of person: Japanese and English compared]”. Gengogaku Enshu 82: 7–12. LakoŸ, George 1996 “Sorry, I’m not myself today: The metaphor system for conceptualizing the self” In Spaces, Worlds, and Grammar, G. Fauconnier and E. Sweetser (eds.), 91–123. Chicago: University of Chicago Press.
Two types of ‘linguistic subjectivity’
Levinson, Stephen C. 1983 Pragmatics. Cambridge: Cambridge University Press. Luria, Alexander R. 1982 Language and Cognition. New York: John Wiley and Sons. McNeill, David 1987 Psycholinguistics: A New Approach. New York: Harper and Row. Newman, John 1996 Give: A Cognitive Linguistic Study. Berlin: Mouton de Gruyter. Radden, Günter 1995 “Motion metaphorized: The case of coming and going”. In Cognitive Linguistics in the Redwoods: The Expansion of a New Paradigm in Linguistics, E. H. Casad (ed.), 423–458. Berlin: Mouton de Gruyter. Shoho, Isamu 1981 “KO . SO . A no Taikei [The system of KO . SO . A]”. In Nihongo no Shijishi [Demonstratives in Japanese], 51–122. Tokyo: National Language Institute. Silverstein, Michael 1976 “Hierarchy of features and ergativity”. In Grammatical Categories in Australian Languages, R. M. W. Dixon (ed.), 112–171. New Jersey: Humanities Press. Slobin, Dan I. 1996 “From ‘thought to language’ to ‘thinking for speaking’”. In Rethinking Linguistic Relativity, J. Gumperz and S. Levinson (eds.), 70–96. Cambridge: Cambridge University Press. Tomasello, Michael 1999 The Cultural Origins of Human Cognition. Cambridge, MA: Harvard University Press. Uehara, Satoshi 2002 “Zibun re¶exivization in Japanese: A Cognitive Grammar approach”. In Cognitive Linguistics and Non-Indo-European Languages, E. H. Casad and G. B. Palmer (eds.), 389–404. Berlin: Mouton de Gruyter. Vygotsky, Lev S. 1962 Thought and Language. Cambridge, MA: MIT Press.
73
De la nécessité de prendre en compte la dimension praxéologique à tous les niveaux de l’organisation des discours Eddy Roulet Département de linguistique Université de Genève
Les recherches de ces dix dernières années, marquées par ce que Vernant (1997) appelle le “tournant actionnel” des sciences du langage, ont montré qu’on ne pouvait comprendre ni décrire l’organisation d’une interaction sans tenir compte de la dimension praxéologique, c’est-à-dire de la structure des actions, verbales et non verbales, qui la constituent. Comme l’écrit Vernant: le tournant actionnel qui ªt du discours une forme d’action — déjà chez Peirce ou Frege — s’est opéré lentement. Il produit maintenant une rupture décisive dont il convient de mesurer toutes les conséquences. (Vernant 1997: page de couverture)
J’aimerais montrer comment, dans une approche modulaire de l’analyse des discours (cf. Nølke & Adam 2000, Roulet, Filliettaz & Grobet 2001, Filliettaz & Roulet 2002), on peut décrire avec précision cette dimension praxéologique1 et indiquer à la fois comment elle se distingue de la dimension textuelle et comment elle se combine avec celle-ci pour rendre compte de l’organisation des discours à diŸérents niveaux. Cela me permettra aussi de montrer comment, dans cette approche de la complexité de l’organisation des discours, on peut rendre compte de propriétés particulières de constructions problématiques, comme les constructions parenthétiques, sans recourir à des solutions ad hoc, en exploitant et en combinant des catégories (ici de relations), motivées de manière indépendante dans le modèle. Examinons, pour mettre en évidence l’importance de la dimension actionnelle, une interaction en librairie comme celle de l’exemple suivant, légèrement adapté pour abréger la présentation (libraire = L et client = C):
76
Eddy Roulet
(1) [C] (a) (il entre), (b) bonjour \ (c) est-ce que vous avez La Machine infernale / (d) c’est pour l’école \\ [L] (e) (il regarde sur les rayons) (f) malheureusement pas en stock \\ (g) tu veux que je le commande // [C] (h) volontiers \\ (i) je repasserai la semaine prochaine \\ (j) au revoir \\ (k) (il sort)
On propose généralement en analyse du discours une description de cette interaction qui se limite au dialogue entre les personnages. Ainsi, dans le modèle genevois, on distinguera quatre échanges: deux échanges rituels tronqués de salutations au début et à la ªn,2 et deux échanges centraux d’achat et de commande, avec des constituants (interventions et actes) liés par des relations illocutoires (initiative et réactive) et des relations interactives (argument, commentaire, etc.) (cf. Roulet 1999: chap. 3 et Roulet, Filliettaz & Grobet 2001: chap. 3 et 6); on aboutit à la structure hiérarchique et relationnelle représentée par le schéma suivant (où arg. = argument et com. = commentaire):
Intervention initiative
(b) bonjour
(c) est-ce que vous avez La machine infernale
Acte princ.
(d) c’est pour l’école
Acte sub. arg.
(g) tu veux que je le commande
Echange
Intervention initiative
(h) volontiers
Acte princ.
(i) je repasserai la semaine prochaine
Acte sub. com.
Figure 1. Structure hiérarchique.
Intervention initiative
Intervention réactive
(f) malheureusement pas en stock
(j) au revoir
Echange
Echange
Intervention réactive Intervention initiative
Echange
La dimension praxéologique dans l’organisation des discours
Cette description ne rend compte que partiellement de l’organisation de cette interaction en librairie. Si le schéma propose une description intéressante de la structure des échanges, on observe qu’il ne rend compte ni des relations entre ceux-ci, ni des actions non verbales qui sont pourtant constitutives de cette interaction de service. Pour combler ces lacunes, on peut être tenté de compléter la structure hiérarchique textuelle en y intégrant les actions non verbales. Mais on aboutit à une description hétéroclite, car les catégories de constituants et de relations textuels ne s’appliquent guère à des actions telles que “entrer”, “encaisser de l’argent” ou “sortir”. On peut aussi partir de l’hypothèse que toute activité, même verbale, est de nature actionnelle, et postuler une description intégrée du comportement humain, qu’il soit verbal ou non verbal. La théorie tagmémique de Pike (1954–1960), développée dans les années soixante (cf. Roulet 1974), proposait déjà un instrument d’analyse permettant de décrire dans un cadre intégré des activités verbales et non verbales, que celles-ci soient communicationnelles (comme tendre la main pour demander de passer le beurre) ou non (comme beurrer une tartine), dans un événement de communication comme un petit déjeuner familial ou un service religieux. Mais cette théorie ne peut constituer un instrument de référence adéquat, car elle confond la structure de l’interaction et celle de l’action. Or, plusieurs travaux récents montrent qu’on ne peut ramener la structure des processus langagiers à celle de l’action, car elles sont régies par des principes distincts. Le philosophe Habermas (1987), dans sa théorie de l’agir communicationnel, distingue l’action de la communication, plus précisément la ªnalité téléologique de l’action des procès d’entente qui caractérisent la communication. Charaudeau (1995: 150ss) distingue l’action, qui “s’instaure dans un espace de réversibilité clos”, de la communication, qui “s’instaure dans un espace de réversibilité ouvert”. Vernant (1997: 169) note “que, s’il convient de séparer théoriquement les deux types d’action — extra-langagières et langagières -, elles demeurent pratiquement étroitement imbriquées dans la mesure où nous tenons tout dialogue pour un processus hétéronome qui prend sens et ªnalité dans une transaction eŸective sur le(s) monde(s)”. Enªn, Filliettaz (2002) approfondit et précise la distinction dans le cadre de l’approche modulaire de la complexité de l’organisation du discours développée à Genève; pour lui, “la distinction entre l’activité et la textualité ne peut pas être ramenée à une diŸérence de rang, mais elle doit au contraire recouper une diŸérence de nature” (2002: 223).
77
78
Eddy Roulet
Dans cette optique, il faut distinguer les structures textuelles (formées d’échanges, d’interventions et d’actes), qui, dans le modèle genevois d’analyse du discours, sont le produit émergent de la négociation sous-jacente à toute interaction, des structures praxéologiques (formées d’incursions, transactions, épisodes, phases et actions), qui sont le produit émergent des schémas d’actions, ou représentations praxéologiques, constitutifs de nos connaissances du monde. On ne peut donc ramener l’organisation de l’interaction à une structure unique et homogène, comme le faisaient la théorie tagmémique ou les premiers travaux genevois, qui postulaient un continuum dans la structure hiérarchique de la conversation, de l’incursion à la transaction, aux échanges, aux interventions et aux actes (cf. Roulet 1981); de plus, il faut montrer comment les deux dimensions, textuelle et praxéologique, s’articulent dans une interaction, car elles s’in¶uencent l’une l’autre: “Parce que les formes médiatisées d’action dans le monde ne sont réductibles ni à l’expression d’une ªnalité téléologique ni à l’accomplissement d’une entente mutuelle, la démarche pragmatique se doit de décrire les processus communicationnels aux deux niveaux que sont a) l’émergence des actions dans le monde et b) la régulation dialogique des processus d’intercompréhension. Plus spéciªquement, il s’agit de déterminer comment les ressources communicationnelles liées à l’entente s’imbriquent dans la structure praxéologique des interactions, et quelles sont les modalités de leur mise en relation” (Filliettaz 2002: 216). C’est l’objet, dans le modèle genevois, de l’organisation opérationnelle, qui combine des informations d’origines textuelle et praxéologique. Ainsi, pour revenir à notre exemple d’interaction en librairie, il faut, d’une part, en établir la structure praxéologique, formée de constituants (incursion, transaction, épisode, étape, action) liés par des relations (étape, réorientation, interruption) spéciªques (cf. Roulet, Filliettaz & Grobet 2001: chap. 4), puis coupler cette structure praxéologique avec la structure hiérarchique textuelle établie précédemment pour obtenir l’organisation opérationnelle, représentée par le schéma suivant, qui donne une description plus complète de l’interaction (2001: chap. 7). Ce schéma est un moyen commode pour représenter avec précision la complexité de l’organisation de cette interaction, et montrer comment des activités verbales et non verbales s’articulent dans le déroulement d’une interaction de service. Il permet d’une part de situer les échanges (et du même coup les interventions de salutations, de requête et de question) dans la structure praxéologique plus vaste de l’incursion de service, dont ils constituent les étapes ou, pour le 3ème échange, suite à l’échec du deuxième, une réorientation,
La dimension praxéologique dans l’organisation des discours
organisation opérationnelle organisation opérationnelle structure praxéologique
ENTRÉE
structure textuelle
(a)
Intervention initiative
SALUTATIONS
(b) étape
Echange PHASE (c)
EPISODE TRANSACTION TRANSACTION D’ACHAT D’ACHAT
étape étape
ÉPISODE
Acte princ. princ.
Acte sub. PHASE (d) arg arg. étape
Intervention initiative Echange
PHASE (e)
étape Intervention réactive
PHASE (f)
INCURSION
étape ÉPISODE
Intervention initiative
(g) TRANSACTION DE COMMANDE
ÉPISODE
réorientation réorientation
étape
PHASE (h)
Acte princ.
PHASE (i)
Acte sub. com. comm.
étape SALUTATIONS (j)
étape
Echange
Intervention réactive
Intervention initiative
Echange
SORTIE (k)
étape
Figure 2. Description plus complète de l’interaction.
et, d’autre part, de marquer le rôle des constituants non verbaux comme étapes d’une incursion (pour (a) et (k)) ou d’un épisode de celle-ci (pour (e)). L’analyse de l’organisation opérationnelle présentée ci-dessus donne une description de la structure interne de cette interaction en librairie. Mais, pour comprendre certains aspects de celle-ci, il faut prendre en compte un contexte actionnel plus vaste. L’analyse d’un corpus d’interactions en librairie montre qu’un libraire ne se comporte pas de la même manière s’il a aŸaire à une demande isolée ou si dix écoliers passent successivement acheter le même livre;
79
80
Eddy Roulet
de même le client ne se comporte pas de la même manière s’il achète un livre pour lui ou pour une autre personne. Il faut donc décrire ce qu’on appelle le cadre actionnel d’une interaction, qui déªnit en particulier les statuts sociaux des interactants (par exemple, libraire et client), leurs rôles praxéologiques (prestataire et demandeur de service) et la manière dont les programmes d’activités de l’un (gérer sa librairie) et de l’autre (remplir ses obligations scolaires) se rejoignent à l’occasion de leur rencontre dans la librairie (cf. Roulet, Filliettaz & Grobet 2001: chap. 4). J’espère avoir montré par cet exemple qu’il est nécessaire aujourd’hui en analyse du discours de prendre en compte la dimension actionnelle des discours et que nous disposons d’instruments permettant non seulement de décrire celle-ci avec une certaine précision, mais aussi de montrer comment elle s’articule avec la dimension textuelle. Si la présence de constituants non verbaux dans des interactions a joué un rôle déterminant dans la perception de l’importance de la dimension actionnelle du discours, il apparaît de plus en plus clairement que celle-ci joue dans l’organisation du discours un rôle qui va bien au-delà de ceux-ci. Je pense en particulier à toutes les situations dans lesquelles plusieurs interactions s’entremêlent ou à celles dans lesquelles une interaction est interrompue par une autre. Ainsi des interactions en milieu familial analysées par Filliettaz (2002), où une mère de famille converse latéralement avec diŸérents membres de la famille occupés à des tâches diverses, des dialogues téléphoniques interrompus par l’intervention d’un auditeur, des entretiens interrompus par un coup de téléphone, ou encore des dialogues théâtraux entrecoupés d’apartés. Dans tous les cas, on a aŸaire à des ruptures ou à des incohérences dans la structure textuelle qui ne peuvent être expliquées qu’au plan de la structure praxéologique, par l’alternance ou le con¶it entre des activités diŸérentes. Dans l’exemple suivant, emprunté à Berrendonner (1993: 55–56) et repris ici sous une forme légèrement abrégée: (2) (a) alors on arrive / (b) et ils me font voir leurs épées / (c) et je regarde / (d) et je vois que… (e) [faible et bas] on peut pas soulever un peu / ce truc / là / (f) (on bouge le truc) (g) voilà \ (h) et je vois qu’ils les avaient limées \ (oral, radio) [c’est moi qui introduis les italiques, l’action entre parenthèses et le découpage en actes]
on observe que les actes en italiques ne font pas partie de la narration et ne pourraient être intégrés dans la structure hiérarchique textuelle de celle-ci, par exemple comme des actes subordonnés de commentaires. On a aŸaire, selon
La dimension praxéologique dans l’organisation des discours
Berrendonner, à une “action hétérogène et intercalaire, dotée d’un but propre et distinct, qui vient suspendre provisoirement l’exécution de la période principale” (1993: 55). Si on admet cette analyse, qui paraît correcte, on saisit aussitôt que, dans une approche modulaire, ces actions intercalaires relèvent de la structure praxéologique de l’interview en cours. Le déroulement d’une interview comporte un certain nombre de phases, liées généralement par des relations praxéologiques d’étape ou de réorientation, mais il peut fort bien se produire qu’une action soit interrompue par une action relevant d’un autre niveau d’activité, ici la gestion du matériel d’enregistrement. Il faut donc, pour rendre compte de l’organisation de ce fragment d’interview, postuler la structure opérationnelle suivante:
structure opérationnelle structure praxéologique
structure textuelle
ACTION (a)
PH
ACTION (b) étape ACTION (c) étape PHASE étape étape
EPISO EPISODE DE
acte sub. acte princ. succession
interv. sub.
interv. sub.
acte princ. succession
acte princ. succession succession
ACTION (d)
PHASE interruption
PH (h) étape
interv. sub.
ACTION (e) ACTION (f ) étape ACTION (g) étape étape étape
intervention
interv interv initiative initiative échange
interv interv réactive réactive
acte princ. reformul.
Figure 3. Structure opérationnelle.
Cette structure permet de rendre compte à la fois de la structure textuelle de la narration, avec des constituants liés par des relations interactives de succession ou de reformulation, et du fait que celle-ci, comme épisode comportant diŸérentes phases et actions, liées par des relations praxéologiques d’étape, est
81
82
Eddy Roulet
interrompue par une phase de gestion du matériel d’enregistrement, réalisée partiellement par une structure textuelle d’échange qui est indépendante de celle de la narration. Il est intéressant de relever ici que la construction intercalaire que je viens de décrire correspond à ce qu’on appelle traditionnellement une construction parenthétique. Voir la déªnition de Dupriez (1984: 329), reprise de Fontanier: “insertion d’un segment de sens complet, au milieu d’un autre dont il interrompt la suite, avec ou sans rapport au sujet” et la déªnition proposée aujourd’hui par Apothéloz & Zay (1999: 23): On peut déªnir la parenthèse comme l’interruption provisoire d’un programme discursif P1 par un autre programme P2, et par la reprise et l’achèvement du programme maître P1 une fois P2 achevé (ou éventuellement abandonné). Une parenthèse est donc le lieu d’un traitement en parallèle de deux énonciations indépendantes […].
Si on revient à l’exemple (2), on observe qu’il correspond à la première catégorie de Dupriez, la parenthèse sans rapport au sujet. On peut se demander si la dimension actionnelle ne joue pas aussi un rôle dans la seconde, la parenthèse avec rapport au sujet, illustrée par l’exemple suivant donné par Berrendonner (1993: 55ss): (3) (a) c’est des gens qui vont être emb embêtés / (b) pour parler poliment (c) puisqu’on est dans un colloque universitaire = (d) ce que je ne suis pas = (e) et alors ils vont se dépêcher…
dont le traitement a fait l’objet récemment d’un débat intéressant. Berrendonner notait, à propos des incises de ce type, que leur “intégration est purement chronologique, et non logique” et il ajoutait “il semble donc que [ces incises] ne puissent être représentées adéquatement à l’aide d’un modèle hiérarchique de l’action […]” (1993: 55; précisons que Berrendonner ne distingue pas la structure hiérarchique textuelle de la structure hiérarchique praxéologique). Je rétorquais dans un premier temps dans Roulet (2002: 174) que nous avions toujours admis l’existence, dans la structure hiérarchique textuelle, de constituants subordonnés à fonction de topicalisation, de commentaire (éventuellement méta-discursif) ou d’argument, qui permettaient de rendre compte sans problème de ces constructions. Je proposais l’analyse suivante de la structure hiérarchique et de l’organisation relationnelle de cette séquence:
La dimension praxéologique dans l’organisation des discours
Ap c’est des gens qui vont être emb embêtés Is arg arg.
Is com.
Ap pour parler poliment
Is puisqu’ arg . arg.
I
Ap on est dans un colloque universitaire
As ce que je ne suis pas com . Ap et alors ils vont se dépêcher
Figure 4. Séquence.
Contrairement à ce qu’a¹rme Berrendonner, et comme le montrait déjà Bassarak (1987), les constituants parenthétiques ne sont pas seulement emboîtés, sans lien privilégié à gauche ni à droite, dans l’énoncé. L’intervention introduite par pour parler poliment est clairement subordonnée comme commentaire à ce qui précède. Il en va de même pour ce que je ne suis pas. C’est encore plus évident dans le cas des relatives parenthétiques décrites par Gapany & Zay (1995), où le lien est indiqué syntaxiquement par le pronom relatif, comme dans l’exemple suivant: (4) Un travail discursif possible sur le repère — qui est alors souvent un stéréotype — consiste à le décomposer pour combiner certains de ses attributs avec le repéré.
Ce traitement textuel des constituants parenthétiques n’est cependant que partiellement satisfaisant, car, comme le relève Simon (2003), il ne rend pas compte de cette propriété typique des constituants parenthétiques qu’est l’interruption du discours. Je propose donc de le compléter en prenant en compte, comme pour les constituants parenthétiques sans rapport au sujet, la structure praxéologique et en décrivant l’organisation opérationnelle, qui peut être représentée ainsi:
83
84
Eddy Roulet
ACT (a)
acte princ.
PH
PH Interrupinterruption
ACT (c)
PH étape
PH
acte princ.
ACT (b) acte acte princ. princ.
interv. interv. sub. sub. com. com.
interv. interv. sub. sub. argument argument
interv. sub. argument
interv.
ACT (d) acte Interrupinterrup- acte sub. sub. tion tion com. com. acte princ.
ACT (e) étape
Figure 5. Organisation opérationnelle.
On notera la diŸérence entre le constituant parenthétique de l’exemple (2), qui est lié au reste du discours seulement par une relation praxéologique d’interruption, et celui de l’exemple (3), qui est lié au reste du discours par une double relation, praxéologique d’interruption, et rhétorique ou textuelle de commentaire (ou d’argument, etc.). Les exemples (2) et (3) montrent bien que la dimension praxéologique intervient dans le discours même aux niveaux de la micro-structure où on l’attendrait le moins, comme celui des constructions parenthétiques. Les constituants parenthétiques de commentaire ou d’argument décrits jusqu’ici se distinguent eux-mêmes des commentaires ou arguments non parenthétiques, qui sont liés au reste du discours par une relation praxéologique d’étape; voir l’analyse que j’ai donnée de la séquence suivante, extraite du dialogue (1): (5) (c) est-ce que vous avez La Machine infernale / (d) c’est pour l’école \\
dans la description de la structure opérationnelle de ce dialogue. En d’autres termes, un commentaire parenthétique est présenté par la relation praxéologique d’interruption comme non planiªé, alors qu’un commentaire non parenthétique est présenté par la relation praxéologique d’étape comme planiªé dans le discours du locuteur/scripteur.
La dimension praxéologique dans l’organisation des discours
Il est temps de conclure ce rapide examen. Il me semble que l’analyse que j’ai proposée ici: –
–
–
met bien en évidence l’importance de la dimension praxéologique dans l’organisation des discours (même monologiques, même dépourvus de constituants non verbaux) à tous les niveaux: de l’étude du cadre actionnel à la description du rôle des constituants non verbaux et des relations entre les échanges, jusqu’à l’analyse de micro-structures comme les constructions parenthétiques; montre que, dans une approche modulaire, il est possible de décrire avec précision cette dimension praxéologique, et d’indiquer à la fois comment elle se distingue de la dimension textuelle et comment elle s’articule avec celle-ci dans l’organisation des discours à diŸérents niveaux; illustre bien l’a¹rmation de Rubattel (1990, 309), selon laquelle une approche modulaire “peut parer à la prolifération de nouvelles entités théoriques, qui ne sont en fait qu’une combinaison encore non décrite de propriétés reconnues par ailleurs dans le modèle”; ainsi, dans la description des propriétés des constructions parenthétiques, une approche modulaire évite le recours à des notions ad hoc, puisqu’elle n’utilise que des relations, textuelles et praxéologiques, motivées indépendamment dans le modèle.
Notes 1. Comme la terminologie n’est pas encore ªgée, j’utiliserai ici indiŸéremment les adjectifs actionnel et praxéologique. 2. On parle d’échange tronqué lorsque l’interlocuteur est dispensé tacitement de réagir par le fait que le locuteur ouvre immédiatement un nouvel échange ou s’en va.
Références Apothéloz, Denis and Zay, Françoise 1999 “Incidents de la programmation syntagmatique: Reformulations micro- et macro-syntaxiques”. Cahiers de Linguistique Française 21: 10–34. Bassarak, Armin 1987 “Parenthesen als illokutive Handlungen”. In Satz, Text, sprachliche Handlung [Studia Grammatica 25], W. Motsch (éd.), 163–178. Berlin: Akademie Verlag. Berrendonner, Alain 1993 “Périodes”. In Temps et discours, H. Parret (éd.), 47–61. Louvain: Presses universitaires.
85
86
Eddy Roulet
Charaudeau, Patrick 1995 “Le dialogue dans un modèle de discours”. Cahiers de Linguistique Française 17: 141–176. Dupriez, Bernard 1984 Gradus. Paris: Union générale d’éditions. Filliettaz, Laurent 2002 La parole en action. Québec: Nota Bene. Filliettaz, Laurent and Roulet, Eddy 2002 “The Geneva model of discourse analysis: An interactionist and modular approach to discourse organization”. Discourse Studies 4: 369–393. Gapany, Joël and Zay, Françoise 1995 “Les relatives parenthétiques: problèmes de reconnaissance et de délimitation”. Scolia 5: 31–64. Habermas, Jürgen 1987 Théorie de l’agir communicationnel. Paris: Fayard. Nølke, Henning and Adam, Jean-Michel (éds.) 2000 Approches modulaires: De la langue au discours. Lausanne: Delachaux et Niestlé. Pike, Kenneth L. 1954–1960 Language in Relation to a Uniªed Theory of the Structure of Human Behavior. Glendale: Summer Institute of Linguistics; 2ème éd. rev., La Haye: Mouton, 1967. Roulet, Eddy 1974 Linguistique et comportement humain. Neuchâtel: Delachaux et Niestlé. 1981 “Echanges, interventions et actes de langage dans la structure de la conversation”. Etudes de Linguistique Appliquée 44: 7–39. 1999 La description de l’organisation du discours. Paris: Didier. 2002 “Le problème de la déªnition des unités à la frontière entre le syntaxique et le textuel”. Verbum 24: 161–178. Roulet, Eddy, Filliettaz, Laurent and Grobet, Anne 2001 Un modèle et un instrument d’analyse de l’organisation du discours. Berne: Lang. Rubattel, Christian 1990 “Polyphonie et modularité”. Cahiers de Linguistique Française 11: 297–310. Simon, Anne-Cathérine 2003 La structuration prosodique du discours en français: Une approche multidimensionnelle et expérientielle. Berne: Lang. Vernant, Denis 1997 Du discours à l’action. Paris: PUF.
From Optative and Subjunctive to Irrealis* Johan van der Auwera and Ewa Schalley Center for Grammar, Cognition, and Typology University of Antwerp
This essay aims to shine a little light on the conceptual confusion around the notions of Optative, Subjunctive and Irrealis. The starting point is the Optative, as illustrated in (1). (1) May Herman think a hundred years!
This paper is an essay in linguistics. There is a need, however, to have the matters scrutinized also by a philosopher of language and passion, ideally someone with a classical background as well as interest in the history of the language sciences. Herman, could you please help?
1.
The Optative from a Latin perspective
From a Latin perspective the ‘Optative’ is a use category. More particularly, the term ‘Optative’ is most prominently employed to characterize a use or, if one likes, meaning of the Latin ‘Subjunctive’, the latter being a ‘form’ category. Following Palmer (1986: 39–40; 2001: 108–109), himself following LakoŸ (1968: 172Ÿ.), the Subjunctive can be said to have at least six meanings or uses. (2) Naviget! sail.subj.pres.3sg ‘Let him sail!’
[Imperative]
* A version of this paper was integrated in a talk at the annual meeting of the German Linguistics society (München, February 2003). Special thanks also for the special comments by Andreas Ammann, César Montoliu, and Ceyhan Temürcü.
88
Johan van der Auwera and Ewa Schalley
(3) Sed maneam etiam. but stay.subj.pres.1sg still ‘But I should stay!’
[Jussive]
(4) Sit fur … at est bonus imperator. be.subj.pres.3sg thief yet is good general ‘Though he is a thief … he is a good general.’
[Concessive]
(5) Iam apsolutos censeas … already paid.oŸ think.subj.pres.2sg ‘You may think that they are already paid oŸ…’
[Potential]
(6) Quid agam iudices? what do.subj.pres.1sg jurymen ‘What am I to do, gentlemen of the jury?’
[Deliberative]
(7) Ut illum di … perduint! that him gods destroy.subj.pres.3pl ‘May the gods destroy him!’
[Optative]
It is not relevant here to discuss the appropriateness of each of these labels or the exhaustiveness of the list. The point is that lists like the above are fully in accord with tradition (see e.g. Menge 2000: 157–166 for a recent account) and that ‘Optative’ is one of the traditional labels. The Latin perspective is by no means restricted to an analysis of Latin. Even for English, which hardly has any verbal morphology left, one ªnds Quirk et al (1991: 155–157, 850) writing that the Optative is one of the uses of the English Subjunctive, more precisely a subtype of the latter’s ‘Formulaic’ use. (8) They recommended that the tax be abolished. [Mandative] (9) Be that as it may… [Formulaic, not Optative] (10) Long live the queen! [Formulaic, Optative]
And in a general introduction to morphology Bubenik (1999: 123) also states that “[t]he subjunctive in the main clause is most typically used to express the wish and the subjunctive in this function is called optative”. There is at least one complication. Cross-linguistically, the notion of Subjunctive also goes with another etymon. In German, e.g., the Subjunctive is the Konjunktiv, literally ‘Conjunctive’. It is actually the latter term which seems more faithful to tradition: coniunctivus is the general name of the mood, and coniunctivus subiunctivus, with subiunctivus translating Greek hypotaktikos ‘subordinate’, was originally the name of the uses of the conjunctive in subordinate clauses (Glück 2000, entries for Konjunktiv and Coniunctivus subiunctivus). The term subiunctivus was later used to name the superordinate category, as in the
From Optative and Subjunctive to Irrealis
above descriptions of Latin and English or in the quote from Bubenik. However, the term also shows up in linguistic discussion with its etymological sense (e.g. Bybee et al 1994: 212).
2.
Enters Classical Greek
When we come to Classical Greek, henceforth usually just called ‘Greek’, the story becomes more complex. Classical Greek does not only have a morphological Subjunctive mood, as well as an Indicative and an Imperative, but also an ‘Optative’ (Grk euktikos ‘wish expressing’, with Latin optativus being its translation). The Subjunctive and the Optative both have various uses. Uses accepted by Duhoux (2000: 206–242) are listed in (11). (11) Form: Uses:
Subjunctive Conditional Temporal Relative Hortative Prohibitive Deliberative
Optative Potential Counterfactual Deferential Volitive (Wish) Volitive (Order) Oblique
Like for the Latin listing in (2) to (7), it is not relevant here to discuss the listing in (11) in all its details. For our purposes the main point is this: diŸerent from Latin grammar, Greek grammar employs ‘Optative’ as a form category, and when it comes to listing the uses of this ‘Optative’ form, Duhoux speaks about ‘Volitive’ and ‘wish’ rather than about ‘Optative’. One could also propose to use the term ‘Optative’ for the use, of course, and use another term for the form category. From a Latin point of view, one could further relabel the form category ‘Optative’ as ‘Subjunctive’. This would then give Greek two Subjunctives, which is not insurmountable given that one could call one of them ‘Subjunctive I’ and the other ‘Subjunctive II’. (12) Form: Uses:
Subjunctive I Conditional Temporal Relative Hortative Prohibitive Deliberative
Subjunctive II Potential Counterfactual Deferential Optative Volitive (Order) Oblique
89
90
Johan van der Auwera and Ewa Schalley
Note that there is also a problem with the other subtype of the Volitive, viz. the ‘Volitive (Order)’. If one replaces this with ‘Imperative’, then we obtain an ‘Imperative’ use category, but Greek also needs an ‘Imperative’ form category. The fact that the Greek and the Latin traditions are diŸerent is confusing. Of course, Greek does need one extra label, and one could argue that the term ‘Optative’ is a good choice, just in case the expression of the wish would be the main function of the said form. However, this is only very partially true. Duhoux (2000: 226) reports two studies in which it is shown that the overwhelming majority of Optative (99%) uses in the chosen Classical Greek texts are not used for the expression of a wish. It is only by New Testament Greek times that the wish-expressing use becomes dominant, but not overwhelmingly so: Duhoux’s text counts give us 63,8%. The essential diŸerence between the Classical Greek Subjunctive and Optative is not, according to Duhoux (2000: 226), that the latter is strongly dedicated to the expression of wishes and the former not. Rather, the Optative covers the realm of the possible, and Subjunctive that of the probable (cp. also Sihler 1995: 582 on Sanskrit). If that is true, then the use of the term ‘Optative’ is not very felicitous, at least not with respect to Greek. Another problem arises if one looks at the history of the relevant moods in Greek and Latin, the form categories, not the labels. It turns out that the Latin Subjunctive at least partially corresponds to the Greek Optative. Conversely, the Subjunctive of Greek at least partially corresponds to the Latin Future. (13) sketches some elements of the development of both form categories, relative to Proto-Indo-European (‘PIE’) (Sihler 1995: 590–600) or perhaps a late stage of it, the stage at which these moods had developed, possibly after the separation from Anatolian, which to judge from Hittite (Luraghi 1998: 184) may not have had them yet. (13)
PIE Subjunctive
Lat Future
PIE Optative
Grk Subjunctive Lat Subjunctive Grk Optative
If one adds further branches, one adds variations on a theme. Germanic, for instance, is a bit like Latin, in that its Subjunctive also goes back to a PIE Optative. But Germanic and Latin also diŸer: Germanic, unlike Latin, does not continue a PIE Subjunctive (Ramat 1998: 403). Or take Armenian (Ajello 1998: 220): like Greek, it continues a PIE Subjunctive and the Armenian counterpart is also called ‘Subjunctive’, but diŸerent from Greek, it does not
From Optative and Subjunctive to Irrealis
continue an Optative. This way, Armenian is also a bit like Germanic, in the sense that they both continue only one of either PIE mood. (14) Germ Ø Arm Subjunctive
Germ Subjunctive
PIE Subjunctive
Arm Ø
PIE Optative
Lat Future Grk Subjunctive
Lat Subjunctive
Grk Optative
The lesson is clear: for cross-linguistic purposes, even when one just restricts the study to the Indo-European family, the terminology is less than adequate. We had already seen that ‘Optative’ serves both as a use category and as a form category. We have now reminded ourselves of the fact that the ‘Subjunctive’ of one language may correspond formally to an ‘Optative’ in the next language. No doubt, this discrepancy is due to a meaning change associated with the forms. So even if the label is to refer to a form category, meaning seems relevant too. This is an acceptable practice, but as the example from Greek showed, the relevance is only partial. At least, one might hope that there is no confusion for the specialists. When a Latinist writes about the Latin subjunctive, there should be no misunderstanding.
3.
Other languages
The study of the modern Indo-European and the non-Indo-European languages owes a lot to the classical tradition. Not surprisingly, the confusion around ‘Optative’ and ‘Subjunctive’ spreads beyond the classical languages. Consider the Turkish paradigm in (15). (15) Sg Pl
1 -(y)AyIm -(y)AlIm
2 -(y)AsIn -(y)AsInIz
3 -(y)A -(y)AlAr
For Lewis (1990: 124; 2000: 134) this is a ‘Subjunctive’ paradigm, for Kornªlt (1997: 353) it is ‘Optative’. The label ‘Subjunctive-Optative’ would also be around (Lewis 2000: 134), and for Koopman (1996L 199: 200), the paradigm is ‘Adhortative’. These statements at least share the presupposition that (15) depicts one paradigm, but that is not certain either: Erdal (p.c.) considers the ªrst person ‘Imperative’ and the second and third persons ‘Optative’.
91
92
Johan van der Auwera and Ewa Schalley
Here are two further illustrations of the same confusion: (i) in Georgian the same wish expressing forms are called ‘Optative’ by Cherchi (1999: 15), but Hewitt (1995: 254–258) prefers ‘Subjunctive’, and (ii) a certain Swahili paradigm goes under the label ‘Subjunctive’ for Wilson (1970: 118–121) but ‘Wishing-Form’ for Loogman (1965: 206–209). For Turkish, there is yet another problem. Consider the wish in (16). (16) May he come!
The most common strategy for expressing this wish does not actually employ a form listed in (15), resulting in gel-e, which at least some of the terminology would lead one to suspect, but rather a so-called ‘Imperative’ gel-sin and, better still, a so-called ‘Conditional’ gel-se, the latter together with ke‡ke ‘if only’. That ‘Imperatives’ join the conceptual arena was already clear from the listing in (11)–(12). But we now also encounter the label ‘Conditional’. Here is yet another problem: for Turkish, the label ‘Conditional’ is used for something other than the Subjunctive or Optative. But it turns out that for some languages it competes with these for referring to the same paradigm. A nice illustration is the Baltic supine + ‘be’ construction (Holvoet in print). The number of illustrations could be multiplied. Let us su¹ce with a variation on the theme of confusion shown by the Baltic supine + ‘be’: Hindi has a paradigm that is sometimes called ‘Optative’ (Kachru 1980), sometimes ‘Subjunctive’ (Montaut to appear; cp. also ‘root subjunctive’ in Shapiro 2003: 267), and sometimes something diŸerent still, not ‘Conditional’ this time, but ‘Contingent Future’ (Masica 1991: 281). The conclusion of this section is fairly bleak: for cross-linguistic purposes, terms like ‘Optative’ and ‘Subjunctive’ are not well-deªned.
4.
Enters the ‘Irrealis’
In recent typology, the term ‘Irrealis’ has been gaining a wider acceptance. A telling symptom of the recent vogue can be found when one compares the ªrst and the second editions of Palmer’s Mood and Modality (1986 vs. 2001). In the ªrst edition, Palmer uses ‘Irrealis’ a few times, when he ªnds an author using it, and also when he discusses conditional clauses. In the second edition, however, ‘Irrealis’ and ‘Realis’ occur on page 1, Section 1.1, and are called the ‘basic concepts’. Other recent works, documenting the rise of the label, are Bowern (1998), Bhat (1999), Elliott (2000) and Verstraete (2002).
From Optative and Subjunctive to Irrealis
Bybee et al (1994: 236) date the rise of the term ‘Irrealis’ back to the seventies. Like ‘Optative’ and ‘Subjunctive’, however, the term ‘Irrealis’ has a classical ancestry. While ‘Optative’ and ‘Subjunctive’ were primarily form categories, ‘Irrealis’ primarily if not exclusively functioned as a use or meaning category, thus allowing statements like ‘in this language Irrealis in the past is expressed by the Subjunctive Past’. Traditionally, ‘Irrealis’ contrasted with ‘Realis’ and ‘Potentialis’. In its modern use, there are at least two important changes. First, ‘Irrealis’ is now also used as a form category (see Mithun 1995 illustrating this approach). And secondly, ‘Irrealis’ now contrasts only with ‘Realis’ and there is no explicit argumentation for the demise of ‘Potentialis’. Is the use of the ‘Irrealis’ notion an improvement? We think that as a form category, ‘Irrealis’ is no better or worse than either ‘Optative’ and ‘Subjunctive’. Inevitably, just like for other notional domains, it turns out, and it has been demonstrated by Mithun (1995) and in the theme issue of Anthropological Linguistics (1998, vol. 40, Nr 2) devoted to ‘Irrealis’, that languages grammaticalize and lexicalize conceptual space in diŸerent ways. Just like the formal category called ‘Subjunctive’ in one language may not have the same uses as a ‘Subjunctive’ in the next language, the so-called ‘Irrealis’ marker of one language may have a diŸerent meaning from the ‘Irrealis’ marker of a second language. How about its usefulness as a use category? Bybee (1998: 269) has spoken disapprovingly of its relevance: “it appears that the term ‘irrealis’ is simply too general to be useful, except as a pointer to a very broad domain” (cp. also Bybee et al 1994: 236–240). We think that this judgment is essentially correct. Yes, ‘Irrealis’ is a rather general term. And yet, the term is useful ‘as a pointer to a very broad domain’, almost as broad as notions like ‘modality’ or ‘time’. Of course, having a pointer is not the same as having a good description or explanation. What must happen then is that one goes from the very general (the domain) back to the speciªcs (uses like those mentioned for Latin in (2) to (7) or for Greek in (11)). But at least typologists now have strategies for doing this. They still have to ªnd the uses and list them, but the stress is now on the connections between the uses, and the cross-linguistic generalizations about possible uses and their connections. This method is associated with the term ‘semantic map’. For the general semantic map strategy, see Haspelmath (2003). Related domains of modality and request and order have gotten some semantic map proposals (van der Auwera & Plungian 1998 and van der Auwera, Dobrushina & Goussev 2004), and Verstraete (2002) already proposes a semantic map for the Irrealis.
93
94
Johan van der Auwera and Ewa Schalley
5.
Conclusion
The conclusion need not be as pessimistic as the preceding might lead one to expect. First, a term like ‘Irrealis’ is useful for a general use category, the domain of what is not actual, not real. Within this general domain, meanings or uses will be posited, much like those shown in (11), they will be mapped and thus shown to be connected. For language-speciªc descriptions, linguists will remain free to give language-speciªc grammaticalizations of the uses names like ‘Optative’, ‘Subjunctive’, but also ‘Conjunctive’, ‘Injunctive’, ‘Potential’, ‘Probabilitive’, and many more, as long as they are based on the specialists’ consensus view on the proper use of their terms, so that this or the other Hindi or Turkish paradigm will be called either ‘Subjunctive’ or ‘Optative’ by all specialists. These labels are best kept as form categories, though of course, it is useful if the semantics of the terms re¶ects what is at least a prototypical use of the category. As to the cross-linguistic use of such terms, we have seen the dangers. Though there is nothing wrong about comparing what linguists of two languages have both called ‘Subjunctive’, if their reasons are the same, the history of linguistics shows that there are too many labels around and that linguists of diŸerent languages often use labels for diŸerent reasons. In 1924 Otto Jespersen discussed the labels ‘Optative’ and ‘Subjunctive’ and complained that “[a] precise terminology is a conditio sine qua non if one wants to understand grammatical facts” (1924: 315). This remains true today, both in general and with respect to the employment of terms like ‘Optative’ and ‘Subjunctive’.
References Ajello, Roberto 1998 “Armenian”. In The Indo-European Languages, A. Giacalone Ramat and P. Ramat (eds.), 197–227. London: Routledge. Bhat, D. N. S. 1999 The Prominence of Tense, Aspect, and Mood. Amsterdam: Benjamins. Bowern, Claire 1998 “Toward a typology of irrealis marking in modality”. Paper Australian Linguistic Society. Bubenik, Vit 1999 An Introduction to the Study of Morphology. Munich: LINCOM. Bybee, Joan 1998 “‘Irrealis’ as a grammatical category”. Anthropological Linguistics 40:257–271.
From Optative and Subjunctive to Irrealis
Bybee, Joan, Perkins, Revere and Pagliuca, William 1994 The Evolution of Grammar: Tense, Aspect and Modality in the Languages of the World. Chicago: University of Chicago Press. Cherchi, Marcello 1999 Georgian. Munich: LINCOM. Duhoux, Yves 20002 Le verbe grec ancien: Éléments de morphologie et de syntaxe historiques. Louvain-la-Neuve: Peeters. Elliott, Jennifer R. 2000 “Realis and irrealis: Forms and concepts of the grammaticalisation of reality”. Linguistic Typology 4: 55–90. Giacalone Ramat, Anna and Ramat, Paolo (eds.) 1998 The Indo-European Languages. London: Routledge. Glück, Helmut 20002 Metzler Lexikon Sprache. Stuttgart: Metzler. Haspelmath, Martin 2003 “The geometry of grammatical meaning: Semantic maps and cross-linguistic comparison”. In The New Psychology of Language: Cognitive and Functional Approaches to Language Structure, Volume 2, M. Tomasello (ed.), 211–242. Mahwah: Erlbaum. Hewitt, B. George 1995 Georgian: A Structural Reference Grammar. Amsterdam: Benjamins. Holvoet, Axel in print“Notes on the development of the Lithuanian and Latvian conditional”. Linguistica Baltica 10. Jespersen, Otto 1924 The Philosophy of Grammar. London: Allen & Unwin. Kachru, Yamuna 1980 Aspects of Hindi Grammar. New Delhi: Manohar. Koopman, Dick 1996 Turks: Een nieuwe leergrammatica. Amsterdam: Bulaaq. Kornªlt, Jaklin 1997 Turkish. London: Routledge. LakoŸ, Robin T. 1968 Abstract Syntax and Latin Complementation. Cambridge, MA: MIT Press. Lewis, GeoŸrey 1990 Turkish. Sevenoaks: Hodder and Stoughton. 20002 Turkish Grammar. Oxford: Oxford University Press. Loogman, Alfons 1965 Swahili Grammar and Syntax. Pittsburgh: Duquesne University Press. Luraghi, Silvia 1998 “The Anatolian languages”. In The Indo-European Languages, A. Giacalone Ramat and P. Ramat (eds.), 169–196. London: Routledge. Masica, Colin P. 1991 The Indo-Aryan Languages. Cambridge: Cambridge University Press.
95
96
Johan van der Auwera and Ewa Schalley
Menge, Hermann 2000 Lehrbuch der lateinischen Syntax und Semantik. Völlig neu bearbeitet von Thorsten Burkard und Markus Schauer. Darmstadt: Wissenschaftliche Buchgesellschaft. Mithun, Marianne 1995 “On the relativity of irreality”. In Modality and Grammar in Discourse, J. Bybee and S. Fleischman (eds.), 367–388. Amsterdam: Benjamins. Montaut, Annie to appear A grammar of Hindi. Munich: LINCOM. Palmer, Frank R. 19861/20012 Mood and Modality. Cambridge: Cambridge University Press. Quirk, Randolph, Greenbaum, Sidney, Leech, GeoŸrey and Svartvik, Jan 1991 A Comprehensive Grammar of the English Language. London: Longman. Ramat, Paolo 1998 “Germanic”. In The Indo-European Languages, A. Giacalone Ramat and P. Ramat (eds.), 380–414. London: Routledge. Shapiro, Michael C. 2003 “Hindi”. In The Indo-Aryan Languages, G. Cardona and Dh. Jain (eds.), 250–285. London: Routledge. Sihler, Andrew L. 1995 New Comparative Grammar of Greek and Latin. New York: Oxford University Press. van der Auwera, Johan, Dobrushina, Nina and Goussev, Valentin 2004 “A semantic map for Imperative-Hortatives’’. In Contrastive Analysis in Language, D. Willems, B. Defrancq, T. Colleman and D. Noël (eds.), 44–66. Basingstoke: Palgrave Macmillan. van der Auwera, Johan and Plungian, Vladimir 1998 “Modality’s semantic map”. Linguistic Typology 2: 79–124. Verstraete, Jean-Christophe 2002 “Mood and clause combining in Australian languages”. Ms. Wilson, Peter 1970 Simpliªed Swahili. London: Longman.
Quand dire, c’est “faire rire aux dépens” Notes sur l’ironie Maurits Van Overbeke Professeur émérite de l’Université catholique de Louvain (Louvain-la-Neuve)
L’ironie est communément comprise comme l’action de faire rire aux dépens d’autrui. Dans le langage de tous les jours, l’expression “propos ironiques” renvoie en eŸet à un large spectre d’énoncés allant de la raillerie au sarcasme en passant par la parodie et le persi¶age. Or, l’origine grecque du mot laisse entrevoir un échange verbal dont le rapport avec le rire, voire avec une éventuelle victime, est moins évident. Il s’agit davantage d’une forme plus nuancée de la conversation, qui consiste à questionner en feignant l’ignorance. S’il faut en croire les dialogues de Platon, Socrate la pratiquait de façon magistrale, parvenant à faire formuler par ses disciples des savoirs et des sagesses, qu’il avait apparemment en tête dès l’entame de l’entretien. Aussi la tradition parlet-elle de ce genre de “débat dirigé”, et de la feinte ignorance qui le fonde, comme relevant de l’ironie “socratique”.
1.
Dire, faire dire et faire entendre
Dans l’enseignement moderne, l’évaluation des savoirs passe elle aussi par de telles questions, posées dans le cadre stéréotypé de l’examen, sauf que désormais, celui-ci enlèverait toute vraisemblance à la feinte ignorance de l’interrogateur, si par hasard elle était réelle. Nous continuons néanmoins à pratiquer cette ªgure de style dans des situations d’énonciation moins marquées, comme celles de la communication quotidienne, chaque fois que nous faisons semblant d’interroger notre interlocuteur pour qu’il exprime nos propres avis ou envies, voire nos pensées ou penchants. Des tournures typées marquent alors
98
Maurits Van Overbeke
l’énoncé, dont le motif est moins l’ignorance a¹chée que la quête d’assentiment. En français moderne, par exemple, des expressions comme n’est-ce pas? pas vrai? hein?, qui ponctuent des déclaratifs apparents avec l’intonation interrogative (par exemple c’est dommage, n’est-ce pas?) invitent l’interlocuteur à acquiescer au contenu de la proposition qui précède. Ce qui peut se faire entre autres par la tournure je ne vous le fais pas dire!, formule usée ou rendue opaque par l’in¶ation sémantique,1 mais qui servait à l’origine à saluer un argument, exprimé par l’interlocuteur, qui confortait ou illustrait la thèse qu’on défendait soi-même. Comme sa vertu persuasive était jugée d’autant plus forte qu’elle émanait, non du défenseur de la thèse, mais d’une instance indépendante (mieux, parfois de la partie adverse!), il était bon de souligner que l’appui venait spontanément d’un vis-à-vis, sans la moindre suggestion du locuteur. Voilà à coup sûr une expression qu’un Socrate n’eût pu utiliser sérieusement, lui qui s’évertuait à diriger le débat pour “accoucher les esprits”, pour “faire dire” à son interlocuteur, après lui avoir évité apories, faux-fuyants et impasses, la seule conclusion imposée par la cohérence rationnelle.
2. Une déªnition normande Une déªnition plus savante de l’ironie, devenue classique depuis Aristote et reprise, à quelques variantes près, par les traités de rhétorique anciens et modernes (Fontanier 1830/1968, Perelman & Olbrechts-Tyteca 1954, Muecke 1970, Booth 1974, Le Guern 1975), est l’action de moquer quelqu’un (ce peut être soi-même) “en disant le contraire de ce qu’on veut faire entendre”. Il s’agit donc là encore d’un comportement verbal marqué par un décalage entre l’être et le paraître, le sincère et l’insincère, mais qui est poussé à son paroxysme par l’introduction du terme contraire. Tantôt on oppose ce que le locuteur dit et ce qu’il “veut faire entendre”, tantôt ce qu’il dit et ce qu’il “pense réellement”.2 L’une et l’autre version méritent qu’on s’y arrête, ne fût-ce que pour en souligner les faiblesses. A commencer par la dernière, qui tire l’ironie vers le faux semblant et le mensonge. Pour ceux qui aiment en souligner l’aspect de dissimulation trompeuse, d’un dire apparent qui occulte une intention contraire, la question se pose de savoir si ces facultés humaines ne dérivent pas d’un instinct animal, que la pression de la sélection aurait façonné et a¹né au long de l’ère néolithique. Certaines observations en éthologie ou paléontologie (Dewaal 1986, Lestel 1995, Picq & Coppens 2001) tendent en eŸet à présenter la dissimulation chez les grands singes et les primates comme une préªguration
Quand dire, c’est “faire rire aux dépens”
du jeu de rôle, de la tromperie, voire du mensonge chez l’homme. A l’évidence, ce rapprochement ne peut se faire avec quelque espoir de succès, qu’en mettant entre parenthèses l’existence, voire la nature, des “intentions” sous-tendant ces comportements comparés.3 On retient alors comme trait commun la signalisation d’un état de choses non conforme à la réalité. Dans cette optique, le simple fait de se cacher d’un prédateur par un leurre ou de dissimuler une réserve de nourriture, équivalent à l’attitude qui consiste à donner de la réalité une image fausse ou déformée. En outre et surtout, la comparaison ne vaut que si l’on oublie ou feint d’ignorer, parmi d’autres traits essentiels, que le mensonge nécessite, au sens fort du terme, l’utilisation du langage, plus précisément celle de propositions que l’énonciateur sait ou croit fausses, et qu’il profère dans le but de les faire accepter comme vraies par l’interlocuteur. Or, ce n’est apparemment pas ce qui se passe dans le monde animal, si évolué soit-il. Ce n’est pas davantage le cas de la contradiction ironique,4 où le but n’est pas de tromper le public ou de l’induire en erreur en disant le contraire de ce qu’on sait ou croit vrai, mais bien plutôt de lui faire comprendre, par un choix approprié d’intonations, de mimiques et de gestes,5 voire par une évidence contraire qui se lit dans la situation d’énonciation, qu’on ne croit pas soimême à la vérité ou au sérieux de l’énoncé, voire qu’on est convaincu du contraire. Si donc mensonge et ironie partagent une velléité de dissimulation, on ne peut pour autant les classer sur le même rayon du faux-semblant. Car le propre du propos mensonger, qui se veut e¹cace, est qu’il ne doit comporter aucun indice laissant à penser que l’énonciateur pourrait être insincère. En revanche, l’énoncé ironique n’a de chance d’être décodé comme tel, que s’il comporte un signe non équivoque permettant de l’entendre, comme on dit couramment, “au second degré”, c’est-à-dire comme n’étant pas destiné à être pris pour la pensée, le sentiment, la conviction de l’énonciateur, mais pour son contraire, qu’il entend ainsi dénoncer et combattre en le ridiculisant.
3.
“Faire entendre” et l’illusion descriptive
Une autre version de la déªnition classique insiste sur la contradiction entre ce que l’ironiste dit et ce qu’il veut “faire entendre”. Cela n’est guère plus satisfaisant, dans la mesure où les termes de l’opposition sont alors présentés comme si les deux étaient directement et également accessibles. Or, ce qu’un locuteur “veut dire” ou “veut faire entendre” se déduit en général de ce qu’il dit réellement, étant donné le contexte dans lequel son énoncé s’insère. S’il dit, par
99
100 Maurits Van Overbeke
exemple, il fait un temps magniªque, alors qu’il pleut à verse, il produit soit un mensonge, si son intention est de nous convaincre d’un état de l’univers qu’il sait (ou croit) faux; soit une antiphrase, s’il nous sait (ou croit) les spectateurs simultanés de cette déplorable météo. Dans ce dernier cas, il compte sur notre sagacité pour apprécier le contraste entre les deux, pour qu’on ne se méprenne pas sur ce qu’il veut “faire entendre”, en l’occurrence, pour qu’on estime à sa juste valeur l’optimisme feint de son énoncé. Or, l’intention du locuteur paraît facile à cerner tant que l’opposition est évidente entre le sens référentiel d’un énoncé et l’état dont il est, ou prétend être, la description. Il existe néanmoins d’autres formes d’ironie, et elles sont probablement plus nombreuses ou fréquentes, qui échappent à ce rapport dit “vériconditionnel” du langage. En eŸet, on peut craindre que la comparaison avec le monde animal évoquée précédemment ne conforte l’idée que la communication, animale ou humaine, n’entretient avec la réalité que des rapports descriptifs ou référentiels, et qui seraient de ce fait susceptibles de vériªcation. Or, un courant important de la philosophie analytique, celui qui est notamment à l’origine de la théorie des “actes de langage”, s’inscrit en faux contre cette idée reçue, l’appelant même, à la suite d’Austin ([1962]1970), “l’illusion descriptive”. Et de fait, bon nombre de phénomènes puisés dans les échanges quotidiens militent en faveur de cette position de refus. Car si le langage servait simplement à représenter l’univers, en ce compris les états mentaux, on verrait mal ce que pourraient bien décrire des locuteurs utilisant des phrases telles que Asseyez-vous! Je vous promets de venir! Je lève mon verre à votre réussite! Pour se borner à ce dernier exemple, on sait qu’une convention sociale veut que l’énoncé soit accompagné d’un geste équivalent, qui consiste à “porter un toast” (ce qui implique entre autres de se lever si ce n’est déjà fait, de lever de la main droite un verre contenant une boisson idoine et de prononcer la formule convenue). Dans ce cas, la tentation est grande d’arguer de la simultanéité entre parole et geste, pour conclure que l’une est la “description” de l’autre. Un examen plus poussé fait néanmoins apparaître qu’en l’occurrence, l’hypothèse descriptive non seulement n’épuise pas le sens de l’énoncé, mais qu’elle passe à côté de l’essentiel, à savoir un acte de langage tel que “saluer ou souhaiter la réussite d’une entreprise”. De ce qui précède on déduira que l’intention ironique ne se laisse pas toujours valablement apprécier par la contradiction entre l’énoncé descriptif et l’état de l’univers correspondant. Bon nombre de ªgures de l’ironie dépassent en eŸet ce niveau référentiel et se servent des énoncés pour accomplir des actes de langage.6 Ainsi la locution merci bien, dont le sens habituel consiste à exprimer sa gratitude à l’adresse d’un interlocuteur. L’utiliser envers quelqu’un qui vient
Quand dire, c’est “faire rire aux dépens” 101
de vous brûler la politesse, vous in¶iger une critique acerbe, voire vous spolier, ne peut à l’évidence relever de la même intention. Est-ce à dire que l’énoncé exprime le contraire de la gratitude et que c’est là le sentiment que l’énonciateur veut “faire entendre”? Certainement pas: dans sa version ironique, l’énoncé merci bien n’exprime pas, comme contenu mental du locuteur, l’ingratitude, la non-reconnaissance, voire l’oubli coupable, qui sont des antonymes de la gratitude, mais il constitue à son tour un acte de langage par lequel celui-ci entend dénoncer l’oŸense subie.
4.
De l’unique sujet parlant au concert de voix
Pour mieux cerner le statut particulier de l’ironie, des travaux déjà datés mais qui stimulent toujours la ré¶exion7 ont ouvert des perspectives nouvelles, tablant plus sur les lois argumentatives de l’énonciation que sur l’opposition traditionnelle entre ce que le locuteur dit et ce qu’il pense ou veut faire entendre. Au-delà de leurs divergences, ils s’accordent à montrer qu’une approche plus féconde de l’ironie est conditionnée par la remise en cause de l’unicité du sujet parlant, d’une part, et par la distinction tranchée entre la “signiªcation de la phrase” et le “sens de l’énoncé”, d’autre part. Pour commencer par la dernière, une approche plus pragmatique mène à la conclusion que l’ironie appartient à ces ªgures de style qui rendent obligatoire la double dichotomie “phrase/énoncé” et “signiªcation/sens”. Si en eŸet une phrase grammaticale est une entité abstraite dont la forme véhicule une signiªcation déªnie par la composante sémantique d’une grammaire G, son utilisation concrète dans un énoncé doit être su¹samment souple pour permettre la communication de sens autres que celui prévu par G. C’est ainsi que, si la phrase il pleut pouvait servir exclusivement à la communication de cette météo bien connue, et non pas également à l’incitation ‘n’oublie pas ton parapluie’, voire au reproche ‘tu avais pourtant prévu qu’il ferait beau’, on ne voit pas comment le sens d’un énoncé pourrait constituer le “contraire” de la signiªcation de la phrase utilisée. Autrement dit: comment pourrait-on atteindre l’opposé dans l’ordre sémantique, si même la simple diŸérence était hors de portée ou exclue d’avance? Les conséquences de l’autre innovation théorique s’avèrent encore plus déterminantes: si dans la production d’une phrase, le sujet parlant est unique, artiªciellement séparé, pour les besoins de l’analyse, de son public et de l’ensemble des conditions d’énonciation hic et nunc, en d’autres termes, s’il ne s’insère pas dans une interaction verbale projetée sur l’arrière-fond d’un savoir partagé, ou de ce que des locuteurs diŸérents ou
102 Maurits Van Overbeke
ªctifs ont déjà dit ou fait dans le passé, ou pourraient dire ou faire dans l’avenir, comment arriver à la nécessaire “négociation du sens” de la phrase produite? Car tout énoncé appelle ou présuppose, à part la reconnaissance de sa signiªcation habituelle ou première, le décodage concerté d’un sens pluriel, la pluralité allant parfois jusqu’à impliquer le sens opposé. Pour illustrer ce point de vue, partons d’une phrase tirée de l’arithmétique de base, telle que deux et deux font quatre. Lorsque celle-ci devient un énoncé dans une situation donnée, la première question qu’elle suscite n’est assurément pas celle de sa vérité mathématique, mais bien plutôt la pertinence ou les raisons su¹santes de sa production. On peut dès lors imaginer un échange verbal tel que (1). Bien entendu, la plupart du temps, la négociation du sens, qui découle en l’occurrence d’un échange explicite, n’a pas besoin du va-et-vient des questions et réponses, les interlocuteurs familiers étant à même de se comprendre à demi-mot. (1) – – – – –
Deux et deux font quatre. Pourquoi tu dis ça? Parce qu’à part nous deux, il y aura aussi Paul et Jeanne. Oui, et alors? Tu as dit toi-même que pour garder un secret, on ne doit pas jamais être plus de trois.
En eŸet, l’interaction verbale (1) aurait tout aussi bien pu tourner court, sans autre perte de lisibilité d’ailleurs, si à l’évocation de la fameuse addition, l’interlocuteur avait réagi tout de suite par: Oui, je sais, il y en aura un de trop, mais on fera avec! L’exemple est éloquent en ce qu’il illustre bien le cas d’une phrase qui ne peut être produite dans son acception première (hormis le contexte de l’enseignement élémentaire, et encore!) sans ouvrir sur un foisonnement de sens divers, qui dérivent tous, non de la signiªcation de la proposition arithmétique, mais des raisons de sa production, des intentions de l’énonciateur, voire des actes de langage que l’énoncé permet d’accomplir. Un de ceux-ci sera, par exemple, l’ironie blessante: proférer cette équation rudimentaire à l’adresse d’un auditeur adulte et normalement constitué ne manquera pas d’être perçu comme ayant le sens: “quel ignorant tu fais, pour qu’on doive te rappeler des savoirs aussi élémentaires!” Notons en passant que dans les deux cas, il s’agit bien d’une voix “pirate”, réelle ou ªctive, qu’on est amené à supputer aªn d’y projeter l’énoncé et d’en tirer un proªl faisant sens. Dans la version complète, c’est le dit antérieur de l’interlocuteur (tu as dit toi-même que…) qu’on rappelle; dans la version abrégée, c’est le dit impersonnel, l’opinion, la doxa, à savoir une connaissance de base en calcul, que tout le monde est censé avoir assimilée. Or,
Quand dire, c’est “faire rire aux dépens” 103
c’est ce “concert de voix” qu’il s’agit désormais d’interroger pour pénétrer plus avant le caractère particulier de l’ironie. Certes, le rapport des sens contraires ou opposés ne doit pas être abandonné, mais il se proªle sur une situation d’énonciation plus complexe, où le sens ultime dérive d’un enchevêtrement de variables communicationnelles, dont la phrase isolée, produite par un sujet parlant unique, n’est qu’un élément parmi beaucoup d’autres.
5.
Contradiction ou ambiguïté argumentative?
A la déªnition classique ou traditionnelle de l’ironie, Berrendonner (1981) a reproché à bon droit qu’elle n’explicite pas ce que la contradiction qui la fonde a de particulier, comparée à celle présente dans d’autres ªgures rhétoriques, comme la métaphore, l’énoncé performatif ou la litote.8 Ce qui l’amène à situer la spéciªcité de la contradiction ironique dans sa pertinence argumentative. Il est vrai que “l’évidence contraire fournie par la situation d’énonciation” que nous avons plaidée ailleurs,9 n’explique qu’un cas de ªgure de l’ironie, fréquent certes, mais qui exclut toute perplexité concernant le degré — premier ou second, sérieux ou ironique — auquel il convient d’entendre l’énoncé. Ainsi, à moins d’émaner d’un locuteur dont la préférence pour ce type de climat serait de notoriété publique, l’exclamation quel temps superbe! alors qu’il tombe des hallebardes, ne laisse guère de doute sur l’intention ironique du propos. Il n’en va pas de même lorsque l’énoncé présente une double “pente argumentative”, selon l’expression de Ducrot (1980a), et qu’aucune évidence référentielle ne permet de choisir entre les deux. Dans l’énoncé de la phrase Pierre est un petit malin (c’est l’exemple de Berrendonner), qu’est-ce qui permet en eŸet de conclure qu’il s’agit d’un argument en faveur de la pente positive (`Pierre est un homme avisé qu’il faut écouter’), ou de la pente négative (`Pierre manque de perspicacité’)? Une loi de cohérence discursive veut qu’un même énoncé ne peut servir au même moment à argumenter en faveur d’une conclusion et de son contraire. Or, il semble bien que l’ironie soit exactement l’exception à cette règle: l’intersection entre deux classes d’arguments incompatibles, et donc notoirement disjointes, cesse d’être vide. Dans ce cas, la perception de l’ambiguïté argumentative et le choix de la pente ironique ne dépendent pas d’un état de fait, mais d’un ensemble de signes comportementaux qui peuvent être directement liés à l’énonciation: intonations, choix de registre stylistique, imitation de voix, reprise d’erreurs ou de fautes de grammaire10) ou indirectement (mimiques, gestuelle, maintien, dégaine…)
104 Maurits Van Overbeke
6.
Polyphonie ou dissonance?
Une autre piste de ré¶exion inaugurée par Sperber & Wilson (1978) et revisitée par Ducrot (1980b) dans le cadre de sa théorie de la polyphonie, concerne l’ironie comme mention. Sans doute faut-il prévoir là aussi toute une gamme de doubles énonciations, allant de la voix le plus implicite11 à celle qui se lit ou s’entend dans l’énoncé ironique comme un rappel, voire un écho. On peut aussi, à l’instar de Kerbrat-Orecchioni (1980), se pencher sur le décodage du trope ironique et analyser en détail les indices cotextuels et contextuels qui permettent de refuser le sens premier de l’énoncé pour aboutir à la lecture tropique ou “tournée”, auquel cas les voix parallèles pourront résonner entre autres dans un savoir partagé ou supposé tel. Pour en illustrer la variété, dont la liste ne se veut nullement exhaustive, imaginons une situation d’énonciation unique, celle d’un mari rentrant très tard le soir, qui se voit accueilli par une épouse inquiète ou excédée par la longue attente. Si l’on projette sur cet invariant situationnel les énoncés ªgurant en (2), qu’obtient-on? (2) a. b. c. d. e. f.
Tu rentres trop tard! Te voilà enªn, c’est pas trop tôt! Ponctuel, comme toujours! Pour quelqu’un qui devait rentrer tôt, c’est réussi! Encore des embouteillages, sans doute? L’avenir appartient à qui se couche tôt!
Déjà dans (2a), qu’on s’accordera à considérer comme dénué de tout trope rhétorique, en particulier de toute ambiguïté ironique, l’indication de l’excès peut s’interpréter comme renvoyant à une norme convenue. En eŸet, aucun moment de la journée n’équivalant objectivement à trop tard, cette expression ne prend sens que sur l’arrière-fond de ce que les interlocuteurs (voire une communauté plus large, les “convenances”, le “sens commun”) estiment normal. Dans (2b) il s’agit clairement d’une litote, ou le même excès est dénoncé par l’exclusion de son inverse extrême, mais ce n’est pas une ironie au sens voulu ici: d’une part, la négation de la valeur polaire tôt, qui selon un mécanisme bien connu renforce la valeur contraire, supprime la contradiction entre l’énoncé et la situation, ou entre ce que le sujet parlant dit et veut faire entendre; d’autre part, il lui manque l’ambiguïté argumentative permettant de l’exploiter en faveur de deux conclusions diamétralement opposées. En revanche, à partir de (2c), nous avons aŸaire à des énoncés répondant à cette déªnition et qui en outre font résonner la voix d’un ou plusieurs énonciateurs
Quand dire, c’est “faire rire aux dépens” 105
théoriques, absents, voire ªctifs. Dans (2c) non seulement la ponctualité évoquée vise l’absence de cette vertu, mais elle rappelle une prétention formulée par l’interlocuteur, ou par sa renommée apparemment indue, et qui contraste avec le comportement dénoncé. La résonance d’une voix parallèle est encore plus claire dans (2d), où la contradiction est soulignée entre l’intention, voire la promesse, exprimées par mention indirecte (tu avais projeté ou promis de rentrer tôt) et la rentrée tardive. Par ailleurs, dans (2e), on perçoit l’énoncé, ªctif ou antérieur, d’une excuse habituelle, qui est rappelée pour faire comprendre qu’on n’y croira pas, le refus de l’excuse alléguée aboutissant à présenter le retard comme inexcusable. L’ambiguïté argumentative réside ici dans la suggestion, soit d’une excuse plausible, qui pourrait faire pencher la balance dans le sens de l’indulgence ou de la décharge, soit d’une excuse prétexte, mais qui renforce alors la condamnation du retard. Dans (2f) enªn, on a aŸaire à la citation d’un énoncé exprimant une sagesse populaire ou impersonnelle (Berrendonner parlerait dans ce cas d’une “ON-vérité”) et que l’interlocuteur est invité à appliquer par inférence à la situation d’énonciation, soit: l’avenir n’appartient pas à moi, qui me couche (trop) tard. Remarquons que selon le savoir partagé des interlocuteurs, la mention, ici tout à fait explicite, pourra prendre diverses formes, mais elle ne sera ironique que dans la mesure où elle permettra de percevoir une ambiguïté entre elle-même et l’argument pertinent de l’échange, en l’occurrence le retard.12
7.
Dire, faire dire et pincer sans rire
L’ironie étant un fait d’énonciation, la question se pose également du type d’actes de langage qu’elle permet d’accomplir et des conditions de réussite que ceux-ci requièrent. Vu sous l’angle de l’e¹cacité communicationnelle, la question est d’autant plus cruciale que l’ironie contient par déªnition une ambiguïté ouvrant le champ à des interprétations, dont certaines risquent d’être à l’opposé de ce que l’énonciateur a voulu faire entendre. On s’est déjà penché sur le type d’actes que permettent d’accomplir des énoncés performatifs tournés en litote par une négation feinte.13 Nous avons alors constaté que la négation de verbes ayant un sens favorable ou mélioratif (approuver, applaudir, complimenter, congratuler) a pour eŸet non pas de suspendre ou de réduire à néant l’acte qu’ils signiªent, mais qu’ils autorisent la réalisation de l’acte opposé, en l’occurrence défavorable. Par exemple, si (3a) est un acte de langage par lequel l’énonciateur assure son interlocuteur de la part qu’il prend à son
106 Maurits Van Overbeke
succès, sa réussite, sa performance, bref, ce qui lui arrive d’heureux, la négation de (3a) en (3b) équivaut non pas au non-accomplissement de cet acte mais, par un eŸet inhérent à la litote, à son contraire, à savoir à la désapprobation, voire au blâme. (3) a. Je te félicite. b. Je ne te félicite pas. c. Je te félicite2.
Si on se demande ensuite ce qui s’accomplit dans l’énonciation de (3c), où le chiŸre 2 en exposant indique un “second degré”, un détournement ironique de (3a), on s’interrogera d’abord sur la situation qui motive l’utilisation de ce performatif. Si sa raison est non une circonstance heureuse, mais une contreperformance, une catastrophe, voire un délit, force sera de conclure que les félicitations apparentes représentent une désapprobation, voire une condamnation. On voit par là que deux ªgures rhétoriques apparentées mais formellement opposées, — la litote basée sur la négation feinte, l’ironie sur la reprise de la version a¹rmative, mais pervertie “au second degré” — permettent d’obtenir des eŸets similaires. Similaires mais non identiques. Car s’il partagent une part de dissimulation, ils se distinguent en ce que seule la deuxième assure l’eŸet de raillerie, de “rire aux dépens de”, qui présuppose dans le chef de l’énonciateur un jugement supérieur conªnant au mépris, à la volonté de blesser en moquant, sans que la victime puisse se retourner contre le railleur, puisque son énoncé lui est formellement favorable.
8.
Ironie et conversation
On a eu raison de souligner (Berrendonner 1981, Ducrot 1980b, KerbratOrecchioni 1980, entre autres) que l’ironie peut générer des expressions toutes faites, telles c’est malin! Une fois idiomatisées, celles-ci perdent en eŸet leur potentiel d’ambiguïté argumentative, ce qui implique qu’à la rigueur, elles ªnissent par ne plus exprimer que leur contraire. Cette évolution a atteint son terme lorsqu’un élément grammatical vient diŸérencier l’idiomatisme ironique de son équivalent sincère; ainsi, c’est du joli! s’oppose à c’est joli! et c’est du propre! à c’est propre! A part ce phénomène désormais connu, on n’a pas toujours accordé l’attention qu’elle mérite à une tendance marquée, surtout en langage parlé et utilisé de préférence par la couche la plus populaire ou la plus jeune, de tournures qui exploitent le vieux ªlon de l’antiphrase pour ajouter du
Quand dire, c’est “faire rire aux dépens” 107
piment rhétorique à l’interaction verbale. On en trouvera quelques exemples, parmi les plus fréquents, dans (4). (4) a. T’inquiète! b. Je vais me gêner! c. Tu m’étonnes!
Ainsi (4a) se dit couramment à un interlocuteur qui semble se faire des soucis à propos d’une action projetée ou de la tournure que pourraient prendre les événements. Sachant que la tendance jeune consiste à parler “allégro”, voire à réduire les mots et expressions à un minimum de syllabes, on pourrait également invoquer ici l’apocope, en l’occurrence l’élision du seul élément restant de la négation, à savoir pas dans (ne) t’inquiète pas. Il resterait néanmoins à comprendre pourquoi une syllabe aussi importante, puisqu’elle peut transformer le sens d’un énoncé en son exact contraire, peut être supprimée sans provoquer le moindre malentendu.14 Dans (4b) on a l’illustration du fait que la plupart de ces tournures plus ou moins idiomatisées demandent un contexte conversationnel, sans lequel leur piment rhétorique ne serait pas apprécié. Il faut bien en eŸet qu’au préalable un interlocuteur ait dissuadé, voire interdit, une action ou un comportement, pour que cette gêne proclamée se tourne en un sans-gêne revendiqué. Dans le mode également ironique on peut citer l’expression apparentée, te gêne surtout pas, qui sert à dénoncer par antiphrase une nonchalance coupable. Le cas de (4c) est plus complexe dans la mesure où il équivaut à l’expression d’un accord avec ce qui vient d’être dit, un peu comme Bien entendu! A qui le dis-tu! C’est évident! Ça va de soi!, sauf que l’évidence du propos enregistré est accueillie par un étonnement feint.
9.
De l’ironie au persi¶age
Comme on l’a vu précédemment, l’ironie se présente souvent comme un chant à plusieurs voix, dont celle du sujet parlant est la plus manifeste, sans être pour autant la plus signiªcative. Une autre façon d’exploiter ce ªlon “polyphonique” est de ne pas se limiter à la mention d’une voix seconde ou tierce, mais de reprendre, pour exprimer le contraire de sa pensée et le dénoncer, le discours même qui y est associé, à savoir celui du camp adverse, avec ses particularités lexicales ou rhétoriques. On feint alors de donner la parole à l’adversaire, à l’objecteur ou à l’ennemi, allant même parfois jusqu’à inventer un personnage ªctif qui tient leur langage. L’exercice était devenu un genre très prisé chez les
108 Maurits Van Overbeke
“philosophes” du XVIIIe, avec Voltaire comme ªgure de proue. On a d’ailleurs observé qu’il prospère volontiers à la faveur de régimes absolus ou autoritaires, qui ne reconnaissent pas la liberté d’expression et sanctionnent durement les voix qui dérangent. Or, cette stratégie pamphlétaire est en réalité une communication oblique avec les avantages et inconvénients propres à celle-ci. Car d’une part, elle permet à l’auteur de se dégager de la responsabilité des idées qu’il fait semblant de propager, laissant au lecteur le choix de trancher l’ambiguïté du propos. Mais d’autre part, rien ne permet d’être sûr que le public visé percevra cette voix, non pas comme celle de l’énonciateur, mais comme celle imitée, parodiée et donc ridiculisée de l’adversaire. En d’autres termes, comment est-on certain que des propos insincères ne seront pas décodés comme sincères? Un exemple illustre en est le chapitre sur l’esclavagisme dans L’esprit des lois de Montesquieu. A première vue, étant donné la position idéologique défendue par l’auteur, les arguments qu’il déploie en sa faveur et les particularités langagières qu’il emprunte à ses adversaires, rien ne permet de distinguer son texte d’un quelconque pamphlet esclavagiste de l’époque. Sauf que subrepticement (c’est la dissimilation qui se dévoile au bon entendeur) il le truŸe d’hyperboles, d’expressions outrancières, voire d’incohérences manifestes, qu’on ne peut prendre au premier degré que si on est dénué de tout ¶air pour le second. Bien entendu, le décodage correct de cette communication oblique gagnera à être projeté sur l’arrière-fond de ce qu’on sait de l’auteur, en particulier de ses convictions, ses inclinations, voire de ses “chevaux de bataille”. Il n’empêche que le principe même de son “autorité” reste paradoxal: car d’une part, l’auteur d’un discours parodique mettant en cause une thèse ennemie souhaite que le lecteur reconnaisse la voix apparente comme insincère ou pervertie, aªn qu’il puisse se délecter de la dérision qui en résulte. Mais d’autre part, il désire aussi être reconnu comme la source cachée de cette imitation vocale et de la délectation qu’elle procure au lecteur perspicace. Autant dire que l’auteur conªe en sourdine à qui veut ou peut l’entendre: ce n’est pas moi qui défend des idées aussi ignobles ou ridicules, mais c’est tout de même moi qui les tourne en dérision. Voici donc un autre stratagème découlant de l’ironie polyphonique: elle ne se borne pas à évoquer par mention plus ou moins explicite une autre voix, celle de l’objecteur ou de la partie adverse, mais elle l’imite, elle la contrefait. Elle va jusqu’à emprunter à l’adversaire son langage, jusqu’à présenter et défendre sa position, mais elle s’applique à le faire d’une façon tellement maladroite, à force d’incongruités et d’outrances, que la thèse en devient risible, voire ridicule. L’exercice est comparable, tant pour ses avantages que pour ses dangers, à l’incursion de l’éclaireur ou de l’espion en
Quand dire, c’est “faire rire aux dépens” 109
pays ennemi: pour ne pas se faire repérer, il adopte l’uniforme et la langue de l’adversaire et parvient à y accomplir sa mission sous ce déguisement. L’entreprise reste néanmoins périlleuse dans la mesure où elle implique à terme le risque de se faire abattre comme ennemi par les agents de son propre camp. C’est donc par excellence une arme à double tranchant.
10. Pour conclure sans clore La diversité des ªgures de l’ironie, dont nous n’avons tenté d’éclairer ici que les plus saillantes, fait surgir la question de leur dénominateur commun. Contrairement au mensonge qui consiste à cacher, déformer ou nier ce qu’on sait ou croit vrai, elle oŸre à l’interlocuteur les moyens de percevoir l’insincérité du propos. Elle part donc sans conteste d’une volonté de faire percevoir un sens précis, mais sans s’exposer aux conséquences normales d’un acte de langage, dont la plus importante est le risque de se voir interpellé, voire contredit, sur la base de ce que l’autre aura perçu, après avoir tranché l’ambiguïté argumentative. Dès lors, conçue pour déjouer les contraintes du discours, elle permet de “dire sans preuve formelle d’avoir dit”, et surtout d’éviter la mise en cause qui pourrait s’ensuivre. En cela, elle ouvre certes un espace de liberté propice, voire nécessaire, au jeu de la feinte et du “rire aux dépens de”. Mais comme elle constitue en même temps une injure cinglante à la cohérence discursive, elle invite à chercher ailleurs son vrai dessein, qui est celui d’exclure un interlocuteur du champ de l’argumentation rationnelle et de le condamner ainsi aux rigueurs du mépris. On voit donc que, pour approximative qu’elle soit, la déªnition classique de l’ironie, basée sur le contraste entre ce qu’on dit et ce qu’on veut faire entendre, n’en comporte pas moins un trait qui l’apparente à celle de Socrate, et qu’on retrouve dans l’étymologie du terme grec eironeia, dérivé de eiron (‘qui fait semblant d’interroger’), à savoir la dissimulation. Dans l’ironie socratique, nous l’avons vu, c’est l’ignorance qui est feinte. D’où la tension entre un simulacre de non-savoir et une connaissance caché, non pas normale ou moyenne, mais supérieure, au point de dicter la voie vers l’expression de la vérité ou de la vertu. Or, cette conception ¶atteuse de l’ironie socratique, qui doit beaucoup à sa traduction latine et aux commentaires de Cicéron et Quintilien, n’était apparemment pas partagée par certains de ses contemporains, et pas des moindres. Une interprétation moins gratiªante de cette souplesse dialectique se lit non seulement dans Les nuées d’Aristophane, auteur plus proche de l’ironie au sens moderne, mais également chez Platon
110 Maurits Van Overbeke
lui-même, entre autres dans sa mise en scène de l’entretien avec Alcibiade dans Le banquet. Là le beau jeune homme qui fait l’éloge du vieux maître, ne peut s’empêcher de lui reprocher en même temps son hubris (pour une traduction sensible au contexte, on peut hésiter entre ‘orgueil’, ‘dédain’ et ‘insolence’) qu’il assimile directement à l’eironeia.15 Dans ces conditions, le verbe eironeuesthai (‘faire de l’ironie, ironiser’), plutôt que d’exprimer le laborieux accouchement de la vérité, vise un stratagème discursif permettant de répondre à toute question épineuse par une autre question, de vaticiner sans prendre partie, et d’ainsi se dérober sans perdre la face. Et après tout, c’est peut-être là un des noyaux intemporels de l’ironie, celui qui relie celle de Socrate à la nôtre: une manipulation discursive qui n’ose pas dire son nom. Elle consiste à prétendre participer à armes égales à un entretien franc et ouvert, mais garde pardevers soi cette arme non conventionnelle: laisser à son interlocuteur le soin de dénouer l’ambiguïté du propos, quitte à se tourner contre lui dès lors qu’il en fait la lecture péjorative, autrement dit: dès lors qu’il prouve avoir saisi l’ironie. Bien sûr, l’ironiste doit alors se récriminer: Ne me faites pas dire ce que je n’ai pas dit! Car en eŸet, il ne l’a pas dit, mais il l’a “fait entendre”. Or, pour cerner exactement cela même qu’il a “fait entendre”, l’interlocuteur est bien obligé de lui faire un procès d’intention, procédé inique s’il en est, et qui le met irrémédiablement dans son tort. Et c’est ainsi que l’ironiste non seulement parvient à parer à tout reproche en mettant les rieurs de son côté, mais encore réussit à mettre en cause sa victime, convaincue de malveillance. Cependant, la même stratégie ironique peut encore servir un autre dessein. Car dans des structures politiques ou sociales rigides, où la délibération égalitaire se voit sinon exclue du moins freinée ou contrecarrée, elle est souvent la seule forme de communication permettant de dénoncer la restriction de l’espace dialogal, de faire résonner une voix discordante tout en paraissant rendre hommage aux seules voix autorisées. Pour en trouver des illustrations convaincantes, il n’est pas besoin d’évoquer les régimes totalitaires ou autoritaires. La communauté de croyance, le milieu de travail, voire le simple cadre familial y su¹sent. Tant il est vrai que toute organisation sociale, par un besoin impérieux de prospérer et de survivre, tend à sécréter une idéologie naturelle, une pensée non pas unique mais diŸuse, qui nourrit un discours conventionnel et convenu. Dans ces conditions, la polyphonie o¹cielle, tout en se prévalant de l’harmonie, ne peut s’empêcher de se faire unisson. A la voix monocorde de la majorité dite silencieuse, la voix ironique fait alors semblant d’apporter sa petite pierre. Mais c’est une pierre dans le jardin de l’Eden consensuel.
Quand dire, c’est “faire rire aux dépens”
Notes 1. Dans sa conception plus récente et presque idiomatisée, la formule exprime simplement l’approbation de ce qui vient d’être dit. 2. Certains dictionnaires spécialisés invoquent le contraire ou la négation pour définir l’antiphrase. Ainsi Mounin (1974: 31): “antiphrase: façon expressive de traduire un jugement par l’expression contraire, affectée d’un ton adéquat: Évidemment, tu as toujours raison (= Tu as tort).” Mazaleyrat & Molinié (1989: 23): “antiphrase: trope qui sert éventuellement de support à l’ironie, lorsque celle-ci est saisissable sur telle ou telle lexie isolable. Le signifié tropique est la négation du signifié correspondant au signifiant occurrent. Si […] elle est belle est à entendre [comme] ‘elle est vraiment moche’, il y a antiphrase sur belle…” Une encyclopédie en ligne la définit même comme “dire le contraire de ce qu’on pense, mais de façon qu’il n’y ait aucune ambiguïté possible” (cf. http://fr.encyclopedia.yahoo.com). On s’interrogera plus loin sur cette prétendue absence d’ambiguïté. 3. Des pans entiers de la pragmatique s’effondrerait si on faisait l’impasse sur les “intentions”; pensons entre autres au rôle (controversé) qu’elle joue dans l’interprétation de la promesse ou dans celle de la fiction comme “feindre sans intention de tromper”. Sans doute la feinte de l’ironie appelle-t-elle elle aussi le blocage de cette intention. (Cf. Bratman 1990 et Gibbs 1999.) 4. L’ironie n’est pas réservée au domaine langagier. Ainsi, il arrive fréquemment que dans un match de football, un joueur qui a reçu un carton jaune, se tourne vers l’arbitre en faisant le geste de l’applaudissement. Dans ce cas, l’arbitre peut lui infliger à la rigueur un deuxième carton jaune qui équivaut à l’exclusion. La règle non écrite de cette jurisprudence, est que des applaudissements à la suite d’une pénalité relèvent d’une gestuelle ironique et peuvent dès lors être interprétés comme une effronterie équivalant à “l’insulte à arbitre” et susceptible d’être sanctionnée comme telle. 5. Voir à ce propos Kreuz (2000). 6. Voir à ce propos Haverkate (1990). 7. Pour s’en tenir au domaine français, on pense entre autres à Berrendonner (1981), Ducrot (1980), Kerbrat-Orecchioni (1980) et Sperber & Wilson (1979, 1992). 8. On pourrait y ajouter les prétéritions du genre: On s’est régalé, je te dis pas! ou Dupont, pour ne pas le nommer. 9. Voir Van Overbeke (1986). C’est également ce cas de figure qui est retenu par Searle [1979]1982. 10. Dans la fameuse affaire Omar Haddad, la victime avait eu le temps, avant d’expirer, d’écrire avec son propre sang la phrase accusatrice Omar m’a tuer (avec l’infinitif au lieu du participe). Peu de temps après, une phrase-écho dans laquelle Omar avait été remplacé par Balladur, avec la même faute d’orthographe, fut ironiquement mise dans la bouche d’un PDG de l’audiovisuel français, qui se voyait victime des velléités de privatisation du Premier ministre de l’époque.
111
112
Maurits Van Overbeke
11. Pour Ducrot (1980), l’énonciation de toute phrase négative (ex. il ne pleut pas) implique déjà une voix autre que celle du sujet parlant. Pour une analyse critique de la négation dans la théorie polyphonique, voir Moeschler & Reboul (1994: 327–329). 12. Ainsi, pour des interlocuteurs qui connaissent bien le répertoire de Raymond Devos, l’énoncé se coucher tard… nuit! fera l’affaire. Et dans le cas de deux amateurs de Proust, le célèbre incipit longtemps je me suis couché de bonne heure représentera également une solution alternative. 13. Voir Van Overbeke (1986). 14. La même explication vaut sans doute pour l’expression désormais familière T’occupe! comme équivalent de Ne t’en occupe pas! C’est pas tes oignons!. 15. Telle n’est cependant pas l’analyse de Romilly (1995).
Références Austin, John L. [1964]1970 Quand dire c’est faire. Paris: Seuil. Berrendonner, Alain 1981 Eléments de pragmatique linguistique. Paris: Minuit. Booth, Wayne C. 1974 A Rhetoric of Irony. Chicago: University of Chicago Press. Bratman, Michael E. 1990 “What is intention”. In Intentions in Communication, Ph. R. Cohen, J. Morgan and M. R. Pollack (eds.), 15–32. Cambridge, MA: MIT Press. De Waal, Frans B. M. 1986 “Deception in the natural communication of chimpanzees”. In Deception: Perspectives on Human and Nonhuman Deceit, R. W. Mitchell and N. S. Thompson (eds.), 221–244. New York: SUNY Press. Ducrot, Oswald 1980a Les échelles argumentatives. Paris: Minuit. 1980b “Analyse de textes et linguistique de l’énonciation”. In Les mots du discours, 7–56. Paris: Minuit. Haverkate, Henk 1990 “A speech act analysis of irony”. Journal of Pragmatics 14: 77–109. Kerbrat-Orecchioni, Catherine 1980 “L’ironie comme trope”. Poétique 41: 108–127. Le Guern, Michel 1975 “Éléments pour une histoire de la notion d’ironie”. Linguistique et Sémiologie 2: 47–59. Lestel, Dominique 1995 Paroles de singes. Paris: Seuil. Mazaleyrat Jean and Molinié, Georges 1989 Vocabulaire de la stylistique. Paris: PUF.
Quand dire, c’est “faire rire aux dépens”
Moeschler, Jacques and Reboul, Anne 1994 Dictionnaire encyclopédique de pragmatique. Paris: Seuil. Mounin, Georges 1974 Dictionnaire de la linguistique. Paris: PUF. Muecke, Douglas C. 1970 Irony. London: Methuen. Perelman, Chaïm and Olbrechts-Tyteca, Lucie 1954 Traité de l’argumentation: La nouvelle rhétorique. Paris: PUF. Picq, Pascal and Coppens, Yves 2001 Aux origines de l’humanité. Paris: Fayard. Romilly, Jacqueline de 1995 Alicibiade. Paris: Fallois. Searle, John R. [1979]1982 Sens et expression. Paris: Minuit. Sperber, Dan and Wilson, Deirdre 1978 “Les ironies comme mentions”. Poétique 36: 399–412. 1992 “On verbal irony”. Lingua 87: 53–76. Van Overbeke, Maurits 1986 “L’avenir d’une négation feinte: Approche pragmatique de la négation”. In Etica e pragmatica, M. Olivetti (ed.), 283–308. Rome: Archivio di Filosoªa. 2001 “La politesse, l’ironie et la marque”. Pragmatique et Politesse 78: 155–172.
113
Part II … & beyond: Art, mind, and community
Pragmatics and evolution David Holdcroft University of Leeds
Diachronic pragmatics has shown that a speech act can have a history. For instance, goodbye, whose primary role now is to mark the end of a discourse, is derived from God be with you, which as well as having that discourse function also had the illocutionary force of a blessing (Arnovick 1999: 95–118). But if speech acts and other pragmatic phenomena can have a history, could the theory of evolution explain in principle some aspects of that history? In the light of contemporary discussions proposing evolutionary explanations of a wide range of cultural phenomena, this would not seem to be an unreasonable question (see, for example, Mellars & Stringer 1989, Barkow, Cosmides & Tooby 1992, and Wheeler, Ziman & Boden 2002). Nevertheless there are a number of strong prima facie objections to an a¹rmative answer. To begin with, there is a large, perhaps indeªnitely large variety of speech acts. Apart from workaday horses such as assertions, orders, and questions, these include ritual insults, dares, curses, and, in an appropriate institutional context, highly specialized acts such as baptizing, sentencing, and anointing, which can seem exotic or strange to those not familiar with the underlying institution. But is not such diversity the mark of local cultural innovations, rather than of behavior which is adaptive for the species? Indeed, if we supposed that it was adaptive, would we not be in danger of having to postulate an Ur-speech act from which the present variety has descended by modiªcations, which is evidently implausible? Moreover, many speech acts are not obviously adaptive at all. Daring, and its more formal variants such as challenging, can lead to injury or even death. A rather diŸerent sort of objection is that whilst evolutionary theory has a lot to say about the adaptive value of enhanced communication skills, this applies to much earlier stages of primate evolution. Writing to Max Muller, Darwin said that someone “fully convinced, as I am, that man is descended
118
David Holdcroft
from some lower animal is almost forced to believe a priori that articulated language has developed from inarticulate cries” (F. Darwin & Seward 1903: 2,45). A focused version of this argument would be that what distinguishes homo sapiens, and possibly some of its ancestors, from, e.g., vervet monkeys, is not the ability to communicate in rather complex situations, but the richness of the content of what is communicated — thanks to the fact that we have a language and they do not. In other words, communication is not a di¹cult trick that we acquired, rather what is novel is a way of vastly enriching the content of what is communicated. But are these objections individually or collectively decisive? The admitted diversity of speech acts rests in part on the existence of ‘institutional’ acts, which Austin’s description of performatives ªts perfectly. There must exist roles to which people have been appointed, rules for eŸecting things by uttering words, and the participants on a given occasion must be as prescribed (Austin 1976: 14). Since the diversity of such roles and rules does not seem to be in principle speciªable, it is perhaps not surprising both that the variety of ‘institutional’ acts is very great, and that they can appear strange to those not familiar with the institution. However, it has been argued that such institutions are a late development, before which it was possible to explain the patterns of behavior within communities “without any reference to the institutional rules and associated inducements and sanctions which deªne the modes of production, persuasion, and coercion of the societies of the Neolithic, and thereafter” (Runciman 2001: 230; see also Runciman 2002: 14). Claiming that such ‘institutional’ acts are of relatively recent provenance does not imply, of course, that there were no institutions, i.e., stable patterns of social behavior, before that. What is being claimed rather is that an institution such as patrilocality — when, at maturity, the males are likely to remain within the community and the females to disperse —, which has been claimed to be “the ancestral human condition [which] predominates through our evolution” (Foley 2001: 78), is diŸerent in kind from, e.g., a courts martial. The former does not require an elaborate system of rules and roles to maintain itself in the way in which the latter does. If, then, we set aside ‘institutional’ acts as late developers, which those who emphasize the role played by illocutionary acts in communication argue require a diŸerent treatment anyway (Bach & Harnish 1982: 108), the argument from diversity and exoticness loses much of its sting. What we are left with is a number of categories whose broad outlines have been recognized by most writers who have proposed a taxonomy, even if their account of illocutionary acts is diŸerent, e.g., Searle (1979) and Bach & Harnish (1982):
Pragmatics and evolution
Table 1. a. Constatives: these commit the speaker S to the truth of the content expressed — or at least to the possibility of its truth; central examples: stating, reporting. b. Directives: these are attempts by S to get the audience A to do something specified by the content expressed; central examples: ordering, warning, and questioning. c. Commissives: these commit S to a future course of action specified by the content expressed; central examples: threatening, promising. d. Expressives: these are formal expressions of S’s psychological state; central examples: thanking, welcoming.
The broad similarities between the proposals suggest that there is some underlying system which, however great the variety of acts falling within each category, seems to be in principle describable. So the fact that there is a great variety of illocutionary acts seems to be a less compelling objection, since we can concentrate on the system. Moreover, it is not, I submit, obvious that mastery of at least some of the central acts within these broad categories could not in principle be adaptive. It seems at least a priori plausible that creatures capable of so doing might be able as a consequence to do things they could not otherwise have done, and thus solve problems that they could not otherwise have solved. At the very least, it seems clear that performing such acts is a di¹cult trick. Just how di¹cult becomes clear, I would suggest, when one considers the third objection that the ability to communicate has a long evolutionary history, and that what is novel to us is the medium we use to communicate with, language, not the fact that we communicate. In a classic study, Cheney & Seyfarth (1990) have inter alia given a detailed and convincing account of the communication system of vervet monkeys. These live in relatively small matrilineal groups, that is, normally adult females remain within the group but adult males join another. The groups occupy quite small territories, which they defend from incursions by other groups of vervets. Cheney and Seyfarth produce abundant evidence that for vervets kin are important, e.g., kin are more likely to aid each other in a con¶ict; that dominance relations are a major feature of vervet matrilines; and that lower-ranked individuals have a preference for alliances with higher-ranked individuals but, at the same time, seek to improve their own status, e.g., by forming alliances against a higher-ranked individual. This, they argue,
119
120 David Holdcroft
seems to make evolutionary sense. Attraction to kin re¶ects the evolutionary beneªts to be gained through kin selection (Hamilton 1964); deference to those of higher rank, while simultaneously attempting to increase one’s status re¶ects the best “mixed strategy” (Maynard Smith 1974) for animals in groups where high rank is often correlated with greater reproductive success. (Cheney & Seyfarth 1990: 57)
Not only can the vervets’ behavior be described as if they recognize other individuals, distinguish kin from non-kin and high-ranking from low-ranking individuals, but Cheney and Seyfarth also argue convincingly that they do indeed have the requisite social knowledge. Hence they live in a complex social environment; indeed the very sort of environment which it has been argued provided one of the main drivers for the evolution of intelligence. For instance, Jolly (1988a: 31) has maintained that the social use of intelligence is of crucial importance to all social primates. As the young develop, they depend on the troop for protection and instruction in their role in life. Since their dependence on the troop both demands social learning and makes it possible, social integration and intelligence probably evolved together.1
In such a social environment, there could well be occasions on which an ability to communicate would be advantageous. Alerting other members of the group to the presence of predators, or of a rival group competing for territory, could increase the chances of survival of kin, for instance. Whilst a signal that the approach of a dominant individual does not constitute a threat, at least for the time being, makes it possible to have a temporary relationship advantageous to both — e.g., to form an alliance against a common enemy. Cheney and Seyfarth argue that vervets indeed have such abilities. There is scope for reservations about their detailed approach, which begins with a description of vervet ‘communications’ followed by a discussion of their meanings, in that the former at times comes perilously close to anticipating the conclusions of the latter — particularly in their attempts to show that vervet cries have reference. However, their conclusion in the latter that vervet communicative behavior exhibits level-one intentionality in Dennett’s (1988: 187) sense, i.e., that they have beliefs and desires, is, with one reservation, convincing. The vervets studied emit three diŸerent alarm calls in the presence of three diŸerent types of predator, each of which typically leading to a diŸerent sort of behavior which increases the chances of an individual’s survival:
Pragmatics and evolution
Table 2. Call
Predator
loud barking leopard & some other cat species double-syllable cough martial or crowned eagle chutter pythons, cobras, mambas
Typical response If on the ground, climb a tree. Look up, or run into bushes. Stand bipedally, and peer into grass.
Since this is a small ‘system’, it should be noted that there are three other alarm calls emitted in the presence of minor mammalian predators that rarely attack vervets. Moreover, since a system of proto-illocutionary force which has only one type of proto-force is evidently remote from the complexity of our own developed system, it is important to add that Cheney and Seyfarth argue that there is also a ‘system’ of grunts emitted, which are given in at least four diŸerent social situations. First, a monkey may grunt when she approaches a more dominant individual; second, a monkey may grunt as she approaches a subordinate. Third, monkeys often grunt as they watch another animal, or as they themselves initiate a group movement across an open plain. Fourth, grunts may be given when a monkey has apparently just spotted the members of another group. (Cheney & Sefarth 1990: 104)
However, at this point there is something of a problem from a human observer’s point of view, since it is apparently di¹cult to discern diŸerences between the grunts, and in most of the cases grunts do not evoke any salient behavioral responses. Nevertheless, Cheney and Seyfarth argue that although we cannot distinguish between the grunts, vervets themselves can. Not only do they distinguish between them, but each of the four described above also conveys speciªc information. However, this conclusion rests on the fact that consistent diŸerences in responding to diŸerent types of grunt occurred in many of their paired comparisons, so that grunts do typically evoke diŸerent types of behavior, contrary to what was suggested earlier. Indeed, if they did not, it would be hard to see how they could have been distinguished, given that they are acoustically indistinguishable from our own point of view. But just how diŸerent is the grunt ‘system’ from that of predator alarms? Certainly, not all of the grunts have a diŸerent proto-force from that of an alarm. Cheney and Seyfarth themselves describe the fourth as ‘an initial warning’, whilst the third might plausibly be held to be a kind of alert.2 As for the ªrst two grunts, from a dominant monkey to a subordinate and vice versa, which seem to signal a willingness to interact despite diŸerences in rank, they
121
122 David Holdcroft
could well have a function that could be adaptive, since they facilitate cooperation. But it would surely be stretching things to describe them as, e.g., invitations or oŸers, let alone statements, not least because if vervets can perform such acts, it is di¹cult to see why they do not do it more often. As Cheney & Seyfarth (1990: 120) shrewdly observe, “… we can easily explain why vervets have so many calls, but we can’t explain why they have so few”. So an important point is not only that their repertoire of proto-forces is very limited, but so is the number of calls they make. Even so, Cheney and Seyfarth’s conclusion that vervet communicative behavior exhibits ªrst- but not second-level intentionality — i.e., that they have beliefs and desires, but almost certainly not beliefs and desires about beliefs and desires of either themselves or others — is actually convincing. The hypothesis that vervets have zero-level intentionality, that is, that their cries are involuntary reactions occasioned by fear (as are the typical reactions to the cries), is ruled out by the ¶exibility of vervet behavior. Vervets do not always emit a cry in the presence of a predator; calls are modiªed depending on the characteristics of the ‘audience’; and the response to them is not always the typical one (Cheney & Seyfarth 1990: 165). Moreover, experiments suggest that if a particular monkey is an unreliable source of warnings of a speciªc kind, its warnings will come to be ignored. On the other hand, the hypothesis that vervets have second-level intentionality (i.e., beliefs and desires about beliefs and desires) is ruled out because “[m]any diŸerent examples, re¶ecting many diŸerent approaches to the same problem, all support the hypothesis that monkeys are unable to attribute mental states to others” (Cheney & Seyfarth 1990: 253). Moreover, “while monkeys are adept at recognizing their own position in a social network or dominance hierarchy, they show little selfawareness” (Cheney & Seyfarth 1990: 254), which is consistent with the view that they have no second-order knowledge or beliefs. Since we have ruled out the hypothesis of zero-level intentionality, we have to conclude that vervets have only level-one intentionality. Accordingly, they utter a leopard cry, for instance, not because they want other members of the group to believe that there is a leopard nearby, but because they want them to run into the trees. It is worth adding that this account strictly speaking only requires the attribution of wants and desires to vervets, but not of beliefs. This is perhaps not as odd as it seems at ªrst sight if Jolly’s suggestion that “desires are probably simpler than factual beliefs, and older both phylogenetically and ontogenetically” (Jolly 1988b: 367) is correct, and it is a point to which we will return.
Pragmatics and evolution 123
Whilst I think that Cheney and Sefarth have made a compelling case that vervets communicate, there is nevertheless a huge gulf between their protospeech acts and ours. First, there is a very limited range of types of proto-force; indeed, arguably there is only one kind that one would feel conªdent in ascribing, viz., that of alerting. Second, there is only a limited range of contrasts within a type. One could perhaps cite as an example the diŸerence in signiªcance between the cry uttered when another troop is ªrst spotted (a wrr), and that when con¶ict is imminent (a chutter), but this seems to be the only example (Cheney & Sefarth 1990: 151). Moreover, it is perhaps partially undermined by the evidence that if the wrr of a monkey is ignored because it has been unreliable, then so too is its chutter (Cheney & Seyfarth 1990: 154). These points, together with the fact that the repertoire of vervet calls is very limited anyway, account for the sort of di¹culty, discussed in Dennett (1988), in ascribing a proto-force with conªdence to a cry unless it is very obvious indeed, as with the leopard call. Further, even if one accepts Cheney and Seyfarth’s arguments that vervet cries have a reference and hence a semantics, which I would resist, then what is striking is that a leopard call, for example, only ever has one proto-force. Not only is this unlike canonical speech acts in our developed system, which have a content that can be used with diŸerent forces, but it also suggests that it is not really possible to draw a sharp contrast between force and content in the case of vervet cries, since these do not vary independently of each other. Finally, theoretical accounts of illocutionary force all assume that speakers and their audiences have higher levels of intentionality than those exhibited by vervets, and the classiªcation of diŸerent types of speech act depends on that assumption. A possible response to the existence of such a big diŸerence would be to argue that it is not plausible to suppose that an evolutionary account can be given of our developed system of speech acts; just as some have argued that, since human language is so diŸerent from animal communication systems, there is no point in trying to develop an evolutionary account of the former. But to do this would be premature. I would argue instead that what the diŸerences show, is that it is simply not true that the extensiveness of our communicative repertoire, compared with that of monkeys, depends only on the fact that language makes available an inexhaustible fund of contents, and that communication as such is not a special skill that we have acquired. On the contrary, communication using the core categories listed in Table A is as much of an innovation as language itself. So that if — and that is, of course, a big ‘if’ — an evolutionary account of the latter is plausible, so is one of the former.
124 David Holdcroft
One indication that this is so, is the fact that several of those who argue for an evolutionary account of the latter give one in which an ability to perform ‘novel’ kinds of speech act has important adaptive consequences. In his account of the origins of proto-language Bickerton (e.g., 1990), for example, argues that the selective pressure in early hominids for proto-language was not social intelligence. So he rejects the Machiavellian Intelligence Hypothesis at this stage of his account, for all that it plays an important role in his account of the evolution of language from proto-language. As a result of moving to the savannah there was, Bickerton argues, an increased risk from predators, whilst at the same time the sources of food became potentially more varied and dispersed, and thus harder to ªnd. So early hominids had not only to be constantly on the alert against predators, but also better at foraging and predation. But predator detection and predator avoidance depend not on social intelligence but on what we might call “pragmatic intelligence”: the noting and interpretation of clues in the environment (footprints, crushed vegetation, and so on), something the apes don’t seem to have. (Bickerton & Calvin 2001: 111)
Since food resources were very varied and widely dispersed, what became more complex for early hominids was not interaction with each other, but with their environment. When foraging over a wide area, an ability to tell others of sources of food, or of diŸerent sources between which it might be necessary to choose, would be adaptive — provided that it was reciprocal. So too would an ability to pass on to the young information that a certain mark was a leopard’s footprint. In either case, the use of a proto-language supplemented by gestures would do the trick, even if it had relatively few terms. But from the point of view of an evolutionary account of speech acts, it is interesting to note that what makes the introduction of proto-language adaptive on Bickerton’s account is its potential to perform proto-speech acts whose function is similar to that of a constative or a directive, e.g., ‘dead mammoth follow’. This lends plausibility to the idea that proto-language and proto-speech acts might have coevolved, without of course reducing the former to the latter. By contrast, Deacon gives a central role to commissives. He makes a fundamental distinction between indexical and symbolic relationships between signs and their ‘referents’. The former (e.g., vervet alarm calls) pick out objects one by one directly but unsystematically. The latter, if they ‘refer’ at all, only do so indirectly but systematically in virtue of logical relations between
Pragmatics and evolution
them (Deacon 1998: 86). Language is, of course, a symbolic system, but there are other such systems, and Deacon argues that the origins of them in early hominids lay in a ritual which has the skeleton of what we recognise as a marriage agreement. … Unlike what is found in the animal world, it is a symbolic relationship. But it is also not just a reciprocal set of promises between two individuals, regarding sexual access and economics. As the French anthropologist Levi-Strauss and many others have emphasized it is also the establishment of alliances: promises and obligations that link a reproductive pair to the social group of which they are part. (Deacon 1998: 400)
The ritual which gives a central role to commissives is adaptive, because it is a solution to the problem that males must hunt cooperatively, but reproductive females cannot join in. The latter need guaranteed access to food, whilst the former need a guarantee that they are in fact providing the food for their own oŸspring. The marriage agreement provides a symbolic solution to this problem. Each account is, of course, speculative, and provided that neither claims exclusivity they could be complementary. But if I had to choose, I would opt for Bickerton’s. First, there is an impressive argument in Milton (1988) that the intellectual di¹culties of widely dispersed locations and varieties of food were crucial for the evolution of intelligence in some primates, which lend support to Bickerton’s argument about the importance of food gathering. Second, constatives have a special position in the identiªcation of speech acts. Typically we do this with conªdence only if we can attribute certain beliefs (and other states) to the speaker and her audience, and also have reason to think that the speaker thinks her audience capable of having beliefs. Thus, not only do we need to have a theory of mind, but we must suppose that the speaker does, too.3 Now we could be conªdent that this was indeed so if the speaker was able to express her beliefs using some symbolic medium, even one as rudimentary as proto-language supplemented by gestures. Moreover, without relying on knock down arguments that only creatures with language can have beliefs, Premack (1988) shows just how di¹cult it is to attribute simple, let alone sophisticated, beliefs otherwise. So I think that, to say the least, it would be surprising if one could conªdently attribute sophisticated promises and undertakings to hominids, but not any constatives.
125
126 David Holdcroft
Notes 1. See also Humphrey’s (1988: 18) suggestion that extensive knowledge of practical techniques and of the habitat can “only be acquired in the context of a social community — a community which provides both a medium for the cultural transmission of information and a protective environment in which individual learning can occur”. 2. For an account of the problems in interpreting this call as part of an insightful review of Cheney and Sefyarth’s work, see Dennett (1988: 198). 3. This is another reason why, if I am right that we do not need to attribute beliefs to vervets, it is di¹cult to attribute a wide range of proto-speech acts to them.
References Arnovick, Leslie K. 1999 Diachronic Pragmatics. Amsterdam: John Benjamins. Austin, John L. 1976 How to Do Things with Words. Oxford: Oxford University Press. Bach, Kent and Harnish, Robert M. 1982 Linguistic Communication and Speech Acts. Cambridge, MA: MIT Press. Barkow, Jerome H., Cosmides, Leda and Tooby, John (eds.) 1992 The Adapted Mind: Evolutionary Psychology and the Generation of Culture. New York: Oxford University Press. Bickerton, Derek 1990 Language and Species. Chicago: University of Chicago Press. Bickerton, Derek and Calvin, William H. 2001 Lingua Ex Machina: Reconciling Darwin and Chomsky with the Human Brain. Cambridge, MA: MIT Press. Cheyney, Dorothy L. and Seyfarth, Robert M. 1990 How Monkeys See the World: Inside the Mind of Another Species. Chicago: University of Chicago Press. Darwin, Francis and Seward, A. C. (eds.) 1903 More Letters of Charles Darwin. Appleton. Deacon, Terence 1998 The Symbolic Species: The Co-Evolution of Language and the Human Brain. Harmondsworth: Penguin. Dennett, Daniel C. 1988 “The intentional stance in theory and practice”. In Machiavellian Intelligence: Social Expertise and the Evolution of Intellect in Monkeys, Apes and Humans, R. Byrne and A. Whitten (eds.), 180–202. Oxford: Clarendon. Foley, Robert A. 2001 “Evolutionary perspectives on the origins of human social institutions”. In The Origin of Human Social Institutions, W. G. Runciman (ed.), 171–195. Oxford: Oxford University Press.
Pragmatics and evolution 127
Hamilton, William D. 1964 “The genetical evolution of social behaviour”. Journal of Theoretical Biology 7: 1–51. Humphrey, Nicholas K. 1988 “The social function of intellect”. In Machiavellian Intelligence: Social Expertise and the Evolution of Intellect in Monkeys, Apes and Humans, R. Byrne and A. Whitten (eds.), 13–26. Oxford: Clarendon. Jolly, Alison 1988a “Lemur social behaviour and primate intelligence”. In Machiavellian Intelligence: Social Expertise and the Evolution of Intellect in Monkeys, Apes and Humans, R. Byrne and A. Whitten (eds.), 27–33. Oxford: Clarendon. 1988b “The evolution of purpose”. In Machiavellian Intelligence: Social Expertise and the Evolution of Intellect in Monkeys, Apes and Humans, R. Byrne and A. Whitten (eds.), 363–378. Oxford: Clarendon. Maynard Smith, John 1974 “The theory of games and the evolution of animal con¶ict”. Journal of Theoretical Biology 47: 209–221. Milton, Katharine 1988 “Foraging behaviour and the evolution of primate intelligence”. In Machiavellian Intelligence: Social Expertise and the Evolution of Intellect in Monkeys, Apes and Humans, R. Byrne and A. Whitten (eds.), 285–305. Oxford: Clarendon. Mellars, Paul and Stringer, Chris 1989 The Human Revolution: Behavioural and Biological Perspectives on the Origins of Modern Humans. Edinburgh: Edinburgh University Press. Premack, David 1988 “ ‘Does the chimpanzee have a theory of mind’ revisited”. In Machiavellian Intelligence: Social Expertise and the Evolution of Intellect in Monkeys, Apes and Humans, R. Byrne and A. Whitten (eds.), 160–179. Oxford: Clarendon. Runciman, Walter G. 2001 “From nature to culture, from culture to society”. In The Origin of Human Social Institutions, W. G. Runciman (ed.), 235–254. Oxford: Oxford University Press. 2002 “Heritable variation and competitive selection as the mechanism of sociocultural evolution”. In The Evolution of Cultural Entities, M. Wheeler, J. Ziman and M. Boden (eds.), 9–25. Oxford: Oxford University Press. Searle, John R. 1979 “A taxonomy of illocutionary acts”. In Expression and Meaning. J. R. Searle, 1–29. Cambridge: Cambridge University Press. Wheeler, Michael, Ziman, John and Boden, Margaret (eds.) 2002 The Evolution of Cultural Entities [Proceeding of the British Academy 112]. Oxford: Oxford University Press.
Expressive language games Antonio J. Marques Universidade Nova de Lisboa
Human language may be seen as an organic complex of language games, in Wittgenstein’s sense of the term. However, a ªner comprehension of that complexity will not avoid the essential diŸerence between the various games. Especially, the distinction between expressive and descriptive language games (curiously not very central in the recent literature) is of crucial signiªcance for a global comprehension of linguistic human behavior. In the expressive use of language there is no-one that speaks except the 1st person and, because of that, there is no possibility of observing myself there, of hearing or describing myself in any way. I tremble is a description of behavior that I observe in myself, and that fully retains the same meaning in the 3rd person’s perspective: he trembles. Here we cannot ªnd any asymmetry. But with I believe that…, for instance, this is not a sentence resulting from an observation of myself. It is, rather, an expressive form, whose meaning can only come from the 1st person of the indicative present. Is the observation of myself (of my own belief) at all possible? Can I translate it into words? And what kind of linguistic behavior would bring about that dislocation in point of view? Wittgenstein presents this situation as follows: “Judging from what I say, this is what I believe”. Now, it is possible to think out circumstances in which these words would make sense. And then it would also be possible for someone to say “It is raining and I don’t believe it”, or “It seems to me that my ego believes this, but it isn’t true”. One would have to ªll out the picture with behaviour indicating that two people were speaking through my mouth. (Wittgenstein 1953: 192e)
The rediscovery made by the pragmatic study of language of the importance of language’s expressive function is still an important philosophical topic, even though it has not been su¹ciently noticed.1 It was deªnitely the second Wittgenstein who brought to the fore the valorization of expressive language
130 Antonio J. Marques
games, establishing a clear demarcation relative to descriptive uses. Notice, anyway, that if the uses are distinct, this does not mean that there is no great variety of quick and frequent phases of transition from the expressive to the descriptive. Wittgenstein acknowledges the importance of that transition: “But here is the problem: a cry, which cannot be called a description, which is more primitive than any description, for all that serves as a description of the inner life (Seelenzustand).” (Wittgenstein 1953: 189b) Or: “A cry is not a description. But there are transitions. And the words ‘I am afraid’ may approximate more, or less, to being a cry. They may come quite close to this and also be far removed from it.” (Wittgenstein 1953: 189c) The lesson we can extract from the acknowledgement of these transitions is that we are confronted with the reversible and binary function of a substantial part of linguistic activity: what we designate as expressive may turn into descriptive, and vice versa. That duplicity of levels is a feature of great relevance, as well as man’s innate possibility — yet always applied in context — of recognizing transitions between the descriptive and the expressive. To that purpose, Wittgenstein points out that this capacity does not depend on any previous inspection of the speaker’s inner life. The truth is that, to the listener, the transformation of an expressive language game into a descriptive one is naturally identiªed within a context: The phrase ‘description of a state of mind’ characterizes a certain game. And if I just hear the words ‘I am afraid’ I might be able to guess which game is being played here (say on the basis of the tone), but I won’t really know it until I am aware of the context. (Wittgenstein 1953: 50)
This is the situation in relation to the listener. But in relation to the speaker, something diŸerent happens and, once again, an asymmetry rebounds. In fact, I know (not just presume!) when I express or describe a mental state or an inner experience. But does the asymmetry arise here, namely, in the fact that I know, with a unique authority of the 1st person? The other also knows whether my linguistic behavior (for instance, when I say I am afraid) is a description or an expression, if he is situated in the context. Where, then, does the asymmetry arise? Without making this question clearer, we shall say that it starts in the authority of the 1st person, with respect to its expressive forms. In fact, only I can a¹rm, in an expressive tone, that I am afraid and on the basis of that expression I will a¹rm, now in a descriptive tone, that I’ve been afraid all morning. For that description I have certainly used introspection, i.e., the capacity to reactivate my inner experience by means of the language I have at
Expressive language games
my disposal. We shall not ignore this concept of introspection, as several important misunderstandings have been generated about it in the literature and the commentary of Wittgenstein’s philosophy. Let us therefore start by verifying one of the most frequent senses of the term, in which it is identiªed with Cartesianism and, in its turn, with the description of an inner life inhabited by mental entities with singular features and identities. Introspective activity, which consists in considering the mind’s scenery inhabited by entities that we can describe, corresponds to a philosophical image of the inner life to which, in fact, Wittgenstein objects himself. The description of behaviors, events, or observable physical processes is somehow equivalent to an anti-Cartesian attitude, with respect to self-knowledge. Is this the fundamental intention of Wittgenstein? That would mean that Wittgenstein eliminates the inner life and substitutes it by the knowledge of what is observable in the exterior, which does not correspond to his thinking. But in what we call the Cartesian attitude another philosophy converges, identiªable by a diŸerent name. That name is Platonism: the language of the introspectionist, just as current language, is impregnated with Platonism, since it presumes a set of entities that are considered independent of the way they are used in our linguistic practices. Now, it is this Platonism working in the context of the Cartesian conception of the mind that Wittgenstein intends to deconstruct. The existence of asymmetries and of the authority of the 1st person challenges a so-called behaviorist component of Wittgenstein’s philosophy in the 30s and 40s. On the other hand, it also corrects an overly structuralist image of Wittgenstein, according to which subjects are speakers made equal by the language games in which they take part and were taught: it is not the speaker who means, he only says what the language game already allows him to say. If the whole technique of the use of terms and expressions that the language game allows is already established from the beginning, the margins of meaning and the role of the speaker are delimited by the variety of uses allowed by the grammar or set of uses of such terms and expressions. What we must take into account is that the behaviorist and structuralist approach, in the case of Wittgenstein, is simplistic or even mistaken, precisely because of the relevance he attributes to the existence of asymmetries and to the authority of the 1st person. The asymmetry appears in the simplest language games, such as in the paradigmatic example of pain, as in I’m in pain, I doubt that he is in pain, etc. A simple grammatical analysis shows how “it makes sense to say about other people that they doubt whether I am in pain; but not to say it about myself” (Wittgenstein 1953: 246).
131
132 Antonio J. Marques
What we call asymmetry arises in the diŸerence between the expressive and nonexpressive uses of language and rebounds in the form of certain 1stperson expressions of the indicative present, possessing a meaning that is not equivalent to that of the corresponding forms in the 3rd person, even though their propositional contents are identical. A simple consideration of the contents of sentences like I believe that it’s raining and he believes that it’s raining does not reveal any signiªcant diŸerences between them. However, Moore’s famous paradox shows how the meaning of sentences, apparently with the same content, is not the same in the 1st and in the 3rd person of the indicative present. While, for instance, I believe that it’s raining, but it’s not raining is not possible, the same content presented in the hypothetical form, suppose that I believe that it’s raining, but it’s not raining, is. In this case, it is as if I had observed myself and then provided information about the disparity between my mental state of belief and some fact. Exactly the same would happen if the 3rd-person perspective were assumed: he believes that it’s raining, but it’s not raining is also evidently possible. But in fact, the 1st person’s form does not allow the conjunction of the expression of a mental state with the negation of a fact (in this case, ‘be raining’). We are now referring to the expressive use of ‘believe’, always in the 1st person of the indicative present, but we shall admit that the same phenomenon occurs with practically all the so-called psychological verbs, i.e., those that express and are used as signs of mental states: fear, desire, intend, want, hope, etc. Thus, psychological verbs introduce a real asymmetry, immediate in the uses of the 1st and 3rd persons of the indicative present, which should be attributed to the singularity of expressive linguistic practices, in contrast to the practices of informing and observing. What is important is the fact that a linguistic practice in which I do not describe myself, neither internally nor externally, is found here, in contrast to what happens in the remaining signiªcant linguistic practices about myself as a practical-linguistic agent. It may seem strange that the opposing Cartesian view also defends such an asymmetry. However, what is truly interesting is that the asymmetry of the Cartesian is an inverted asymmetry relative to the one we have been studying. In the Cartesian view, the epistemological privilege belongs to the speaker, while in the Wittgensteinian one, it belongs to the listener. In other words, in Wittgenstein’s case, it is the 3rd person who assumes a cognitive perspective, not the 1st. If we observe carefully, it makes sense to say I know what the other thinks, feels, etc., but not to say I know what I think, feel, etc. Thus, the grammar shows how, with respect to myself, no epistemological privilege is vindicated.
Expressive language games
This grammatical and anti-Cartesian note from Wittgenstein has important consequences in the philosophy of mind. The Cartesian asymmetry is epistemological, while the Wittgensteinian is expressive. 1. Cartesian asymmetry: epistemological privilege of the 1st person; only the subject has access to his mental states and can describe with clarity the relevant objects. 2. The inverse of this asymmetry: expressive privilege of the 1st person; only the subject can express an experience, and others can only know through inference my own mental states. In fact, I do not say, for instance, I know I’m believing, but others can say I know he believes (because he has this or that behavior). The asymmetry really exists in both views but, in the Cartesian, with the costs inherent to the presence of a subject observing himself, the scene of his mind is something to which no-one else has access and that allows him to describe with enough clarity what the others will never be able to describe. But the Cartesian image of the epistemological privilege of the 1st person has crucial consequences for the form of self-knowledge and how it is applied in the knowledge of the others and of the world. The Cartesian position does not admit as essential the distinction between expression and description, as all of them pass via the theater of the mind, which is in principle describable. What we ask, then, is whether the later Wittgenstein’s philosophy represents, in its deepest motivations, an attempt to free ourselves of what is called Cartesianism, or whether it is applying in it, even more profoundly, a positive and constructive vis. Under the guise of Cartesianism, let us retain two features that may represent a common denominator: (1) true self-conscience depends on a selfknowledge directed to the inner ‘world’, which it is possible to describe in natural language for a subject having privileged access to that world of his; (2) the description of an inner experience enables the conscience of the external experience. If we think of Descartes’ 2nd Meditation, this seems a formulation quite close to his program of founding, in the cogito, knowledge, as well as the ontology itself, of both the internal and the external world. However, it is problematic that the rejection of Cartesianism, thus formulated, implies a radical devaluation of introspective activity on the part of Wittgenstein, pace the great majority of commentators that sees in his philosophy a strict rejection of the value of introspection.2 Hacker (1996), for instance, interprets what Wittgenstein says in his Philosophical Investigations (§ 412) about “turning my
133
134 Antonio J. Marques
attention on to my own consciousness” as a completely negative judgment about introspection. This is an illusory operation, as the introspectionist, at least according to Wittgenstein in Hacker’s interpretation, wants something impossible: to isolate the mind in itself. Now, in this section of the Investigations the reference to the “incredible” character of that operation comes from the imaginary interlocutor with whom Wittgenstein is debating and who has the function of introducing the philosophical suggestions subject to his therapeutic analysis. For now, we shall only refer to Sections 412–413 of the Investigations, where the valorization of introspective activity is clariªed, as a development of the 3rd-person perspective on the expressive activity of the 1st: an expression like I’m in pain is the linguistic (expressive) basis for subsequently describing that sensation to, say, a doctor, including a description of its intensity, where it hurts, etc. It is a simple operation of turning my attention to an inner experience, expressed or expressible (I may not have expressed it, but it could be expressed), and thus transforming it into an object of description. Note that, even though Descartes defends an epistemological privilege of the subject in relation to what occurs in his mind, or to his immediate inner experiences, there is nothing that tells us that this knowledge is not always linguistically mediated. Does the image (perhaps more solipsist than simply introspectionist) of a subject exclusively describing his own mental states correspond to the true image of the Cartesian cogito? What this truly accomplishes is an operation on self-knowledge, from the 3rd person’s perspective: nothing of what occurs in me presents itself to the mind, unless it is thinkable and describable precisely from that perspective. Yet, the Cartesian objects, when we talk about the cogito, is it not precisely the 1st person who describes, as no-one else (no other perspective) would do, the inner life? And are we not, then, confronted with an asymmetry? But notice what the Cartesian cogito does: it thinks what occurs in me as a will, a pain, a desire, etc. In this sense, the cogito is always self-objectiªable and a self-objectiªable position is, after all, characteristic of the 3rd person. A primary feature of some of our expressive language games is that they are not inferred from any observation of my behavior or of the behavior of the others, while the opposite happens with other language games. Expressive language games correspond, in general, to the linguistic exteriorization of experiences and to what we have already veriªed to be expressive forms of our linguistic activity.
Expressive language games
It is easy to verify how expressive verbs, in the above-mentioned sense, do not simultaneously fulªll functions of observation and description, in addition to their speciªc expressive function. If this were not so, there would be no incompatibility between the a¹rmation of my state of belief and the concurrent negation of the belief’s object, as in the examples above. If belief is an observable state of mind, why shouldn’t you observe that it doesn’t rain and at the same time observe a belief that it is raining? This only shows that treating belief as an observable state of mind is a complete mistake. (Wittgenstein 1989: 65)
Notice how I am writing means ‘I observe myself writing’, while I am believing is not possible, at least in the meaning of ‘I observe myself believing’. The same holds for the expression I am in pain, which in the 1st person of the indicative present cannot mean ‘I observe myself being in pain’. That is why Wittgenstein makes the observation that it is not possible (grammatically speaking, of course) to say I know I’m in pain, while it is possible to a¹rm that I know he is in pain.3 The asymmetry reveals itself in the grammatical analysis, and we can say that we are thus ªnding the real 1st person, i.e., one that manifests itself in the expressive use of certain psychological verbs. Wittgenstein’s point is essentially that there is in fact a 1st person, and all of his argumentation is oriented to show that it does not hide itself behind linguistic practice. A persistent and systematic grammatical analysis can reveal that real 1st person. Thus, the reality of an experience is admitted, which may approximate the notion of a private inner life we are looking for in the 1st person. We shall not reject this approach if we maintain the notion of an impossibility of an observation simultaneous to its expression. Remember Wittgenstein’s remark that “that queer relation of ªrst and third person — typical of psychological verbs — characterizes experience. Experience is always expressed.” (Wittgenstein 1989: 154) Obviously, what is used as an expression can also come to fulªll descriptive functions, and it is notable that in linguistic practice the transition from one function to another is not always easy to recognize. The expressive function is more primitive and it is in their primitive use that it is possible to say that expressive forms do not sanction a simultaneous observation from a 3rd-person perspective.4 Remember how an expressive linguistic form is a substitute of an original manifestation of pain, fear, pleasure, desire, etc. (Wittgenstein 1953: § 244). The child is taught to replace those manifestations by ever more complex language games and doubtlessly, the constant need of communication brings the child to develop those games into description in general, which includes a multiplicity of functions:
135
136 Antonio J. Marques
For one or another of a class of features goes with what we call “describing”. Observing, considering, remembering behaviour, a striving for accuracy, the ability to correct oneself, comparing. A cry is not a description. But there are transitions. And the words “I am afraid” may approximate more, or less, to being a cry. They may come quite close to this and also be far removed from it. (Wittgenstein 1987: 51)
However, the possibility of the transition of any expressive form into a description does not put its originality and primacy at risk: the child ªrst learns to express a desire and only then to describe the object of desire. To what extent does this descriptive activity then depend on the primitive expression? This is not only a linguistic curiosity without philosophical consequences. It happens that the primacy of the expression proves the existence of a real 1st person, in accordance to what emerges from a grammatical analysis. As a matter of fact, it would be the use of the verbal form of the 1st person of the indicative itself that would be at risk if, so to speak, an original illocutionary force of the expression was not ªrst instilled in us. That is clearly what Wittgenstein suggests in the following lines: Imagine a language in which “I believe that p” is expressed only by means of the tone of the assertion “p”. They say, not “He believes” but “He is inclined to say…” and there exists also the hypothetical “Suppose I were inclined to say…”, but not the expression “I am inclined to say…”. (Wittgenstein 1987: 83)
In other words, the absence of a linguistic sign or form corresponding to the 1st-person expression makes room for the mere a¹rmation of propositional contents (p), which is, because of that, symmetrical to the same contents a¹rmed from the 3rd person’s perspective. Of another person I shall say she is inclined to say that p instead of she believes. In that case, the 1st person’s correspondent form is not I’m inclined to a¹rm p but suppose that I am inclined to say p. With this Wittgenstein means that the symmetry cannot be constructed, unless with observable verb forms and not with expressive ones. In suppose that I am inclined to say…, a 1st person observing himself is constructed; this does not happen in I am inclined to say that…, which is, in turn, an expressive form. The symmetrical construction reveals a crucial feature of expressive forms, which is their nonobservation or, more precisely, their selfobservation. We have seen that one feature of the indicative’s 1st person for psychological verbs is that they are not based on observation, contrary to what happens in the 3rd-person paradigm. This feature, far from being exposed by some introspective methodology, is instead revealed by grammatical analysis. We may say that the subjectivity is demonstrated through a panoramic view
Expressive language games 137
(Übersicht) of expressive and nonexpressive language games, and through the study of conªgurations of asymmetry. It is our conviction that the goal of philosophical peace was pursued by Wittgenstein, literally, until his last day, and that it was on an attempt to elucidate what constitutes the inner self, in all its structural opacity and capacity of dissimulation, that he ultimately focused. That peace may have not been reached but, on the other hand, Wittgenstein did propose another picture, more real, of human inner nature, as ªrm and spontaneous as the expressivity ascribed to human beings.
Notes 1. An important exception is, doubtlessly, John Searle. In “A taxonomy of illocutionary acts”, Searle (1996: 143, 224) reminds us that “in general, in the performance of any illocutionary act with a propositional content, the speaker expresses some attitude, state, etc., to that propositional content. Notice that this holds even if he is insincere, even if he does not have the belief, desire, intention, regret, or pleasure which he expresses, he nonetheless expresses a belief, desire, intention, regret, or pleasure in the performance of the speech act. This fact is marked linguistically by the fact that it is linguistically unacceptable (though not self-contradictory) to conjoin the explicit performative verb with the denial of the expressed psychological state. Thus one cannot say ‘I state that p but do not believe that p’, ‘I promise that p but I do not intend that p’, etc.” 2. Wittgenstein’s attitude towards introspection is doubtlessly critical, if we understand it as an activity of representing objects that exist internally and that I can designate privately. The value that he attributes to introspection does not so much lie in the capacity to know the inner self through designating entities that inhabit it, but more in a capacity of re¶ection about inner experience that is linguistically expressed or expressible. This is clearly shown in the following passage about William James: “And James’ introspection showed, not the meaning of the world ‘self’ (so far as it means something like ‘person’, ‘human being’, ‘he himself’, ‘I myself’), nor any analysis of such a thing, but the state of a philosopher’s attention when he says the word ‘self’ to himself and tries to analyse its meaning. (And a good deal could be learned from this.)” (Wittgenstein 1953: § 413) Recognizing a more favorable position with respect to introspection, McGinn (1997) still cannot specify its precise positive role in Wittgenstein’s work. 3. See Wittgenstein (1953: II, xi): “I can know what someone else is thinking, not what I am thinking. It is correct to say ‘I know what you are thinking’, and wrong to say ‘I know what I am thinking’. (A whole cloud of philosophy condensed into a drop of grammar.)” 4. The observer’s perspective is always the third person’s, even if the observer is me observing myself.
138 Antonio J. Marques
References Hacker, Peter M. S. 1996 Wittgenstein, Mind and Will. Oxford: Blackwell. McGinn, Marie 1997 Wittgenstein and the Philosophical Investigations. London: Routledge. Searle, John R. 1996 “A taxonomy of illocutionary acts”. In The Philosophy of Language, J. R. Searle, 141–156. New York: Oxford University Press. Wittgenstein, Ludwig 1953 Philosophical Investigations. Oxford: Blackwell. 1987 Last Writings on the Philosophy of Psychology, Vol. 1 [G. H. von Wright and H. Nyman (eds.)]. Oxford: Blackwell. 1989 Wittgenstein’s Lectures on Philosophical Psychology 1946–47. Chicago: Chicago University Press.
Language as pragmatics Studying meaning with simulated language games Domenico Parisi Institute of Cognitive Sciences and Technologies National Research Council, Rome
1.
Meanings represent a “terrible” problem for science
Unlike philosophers, who rely only on conceptual analysis, reasoning, and discussion with other philosophers, both living and dead, scientists must ultimately judge the validity of their theories and hypotheses by comparing these theories and hypotheses with systematically and objectively observed empirical phenomena. Hence, the possibility of observing, systematically and objectively, reality, not just talking or thinking about reality, is a critical requirement for doing science. Language is both linguistic forms and the meanings of these forms. Linguistic forms by themselves are just sounds that are heard or produced in speaking, or visual images that are read or produced in writing. To become linguistic signs these forms must have a meaning for both speakers/ writers and hearers/readers. It is their meaning that allows linguistic forms to play such an important role in the mental and social life of human beings. But meanings are very di¹cult to observe. Linguistic forms are reasonably well-deªned entities, easy to perceive, locate, identify, compare, and measure. The meanings of linguistic forms are nothing like that. As argued by Wittgenstein (1953), the meaning of a linguistic form is the role the form plays in the behavior and social interactions, or, more generally, in the lives of the individuals who use it. But human behavior is di¹cult to observe systematically and to describe and measure and, consequently, the meanings of linguistic forms are di¹cult, if not impossible, to perceive, locate, identify, compare, and measure. It is not even clear where to begin to look. In fact, as objects of scientiªc inquiry meanings are almost
140 Domenico Parisi
hopeless. A science of language must be a science of both forms and meanings. But while linguistic forms can be systematically and objectively observed, their meanings cannot. Hence, meaning represents a ‘terrible’ problem for a science of language. Since the meanings of linguistic forms are di¹cult to observe and identify, scientists of language take recourse to clearly inadequate substitutes. Meanings are introduced into discourse by simply producing the linguistic form which designates them: “the meaning of dog”. Nothing is supplied except the form. Alternatively, what is supplied is a deªnition of the form, i.e., a set of other forms that are collectively assumed to have the same meaning as the deªned form: “a dog is a mammal with a tail”. But one never sees the meaning, either of the deªned form or of the deªning forms. If one is studying a foreign language, the meaning of a form belonging to the language is identiªed simply by oŸering the form that translates the foreign form into the native language. The meaning of Hund in German is ‘dog’. Again, one sees only forms, never meanings. Historical linguistics dealing with dead languages is an even more desperate case, since only linguistic forms survive. Another strategy is ostensive deªnition: one points out some particular object in reality to which the linguistic form is used to refer. One identiªes the meaning of dog by pointing to a dog or to a picture of a dog. Aside from the fact that this strategy can be used only for some linguistic forms but not for all, ostensive deªnitions have well-known intrinsic limitations. When I point to a dog to explain the meaning of dog, how does the hearer know that I am pointing to the entire animal, and not, for example, to the color of its skin? It is clear that all these methods function only because they implicitly assume that who is listening to a lecture or reading a scientiªc article reporting a piece of research on language already knows the meanings of the linguistic forms that are studied. But in this way the researcher is renouncing the study of one half of language, in that he or she does not provide or report any systematic observations on the meanings of the linguistic forms. Linguists only deal with linguistic forms and they do not have available to them any methodology for observing and describing meanings. Psycholinguists try to remedy this situation and capture the meanings of linguistic signs for young children by systematically observing and audio- or video-recording the behavior of children while they are using language. However, clearly most speakers/ hearers are not young children. Psycholinguists also run experiments on the linguistic behavior of adults. But the context of use of linguistic forms in laboratory tasks is always very artiªcial. Participants may be asked to decide
Language as pragmatics
whether a succession of sounds or letters is or is not a word, to choose the picture to which a given word applies among a set of diŸerent pictures, or to say whether a sentence is true or false with respect to a given picture. Clearly, in these tasks the meaning of the linguistic forms is presupposed, not observed. The fact that the science of language does not possess an empirical methodology for studying the meanings of linguistic forms, does not only make the science of language a half-science — a science of linguistic forms but not of their meanings — but is at the origin of a series of distortions and mistaken views about the nature of language and of linguistic meanings. By practically identifying the meanings of linguistic forms with these forms, meanings appear as entities that are well-deªned, clearly identiªable, ªxed, invariant from individual to individual, with a well-deªned internal structure, whereas meanings as components or aspects of behavior are entities with ill-deªned boundaries, di¹cult to identify, always changing, diŸerent from individual to individual, very sensitive to the context of use, with an internal structure which is complex and diŸerent from case to case. In this way, some of the most important properties of language elude the very science that should capture and explain them.
2.
How to solve the problem
Can the ‘terrible’ problem that meanings represent for a science of language be solved? In this section, we argue that it can be solved by using a methodology which uses Wittgenstein’s notion of language games and realizes language games as computer simulations. As already noted, for Wittgenstein the meaning of a linguistic form is the role the linguistic form plays in the behavioral interactions of the user of the form with the world, including his or her co-speciªcs. The linguistic form is part of these interactions, is part of the individual’s life. To study language as “a form of life” Wittgenstein introduces a simpliªed situation in which a user deploys a simple language in his or her interactions with a simpliªed world. By creating and analyzing these simpliªed scenarios one can observe and study how language is actually used and, therefore, determine what is the actual meaning of a linguistic form. Wittgenstein’s justiªcation for introducing language games was philosophical. He thought that, by observing how linguistic forms are actually used, one can dissolve philosophical problems which, for him, were pseudo-problems that arise from misusing language. But we believe that the idea of language games can be adapted for the purposes of
141
142 Domenico Parisi
science, that is, for studying language as an empirical phenomenon to be inserted in the uniªed picture of reality that science is progressively building up. The main adaptation which is required to use languages games for the purposes of science is to realize language games as computer simulations. What are computer simulations? Computer simulations are theories or models expressed in a novel medium. Traditionally, scientiªc theories and models are expressed by using either ordinary language or mathematical symbols. Simulations are theories or models expressed as computer programs. When it runs in the computer, the program generates results which are the empirical predictions that are derived from the theory or model incorporated in the simulation. If the results match the observed phenomena, the theory or model is conªrmed. Otherwise, it has to be modiªed or abandoned. Computer simulations are a new research methodology which is being increasingly adopted in many scientiªc disciplines, because it oŸers important advantages. The ªrst advantage is that simulations force the researcher to formulate his or her theories or models, that is, his or her ideas and hypotheses on what is behind the observed phenomena, and explain them in ways that must be precise, explicit, detailed, and complete. The reason is that theories or models which are not su¹ciently precise, explicit, detailed, and complete cannot be incorporated in a program which actually runs in a computer to produce the expected results. Another advantage is that by manipulating the various conditions and variables of a simulation, the researcher can obtain a large number of detailed results which, as we have already said, are the empirical predictions derived from the theory incorporated in the simulation. These can then be compared with the actual phenomena. Thus, not only can the dialogue between theories and empirical data, which is critical for science, be enlarged and strengthened, but the researcher and, especially, his or her colleagues can have a clear idea of the actual empirical content of any given interpretation and explanation that is being proposed. A third advantage of simulations is that they can reproduce complex situations in which there are many causes and factors, and the observed phenomena are the results of the interactions among many diŸerent entities which cannot be predicted or deduced from even a perfect knowledge of the entities and of the rules that govern their interactions. The researcher puts into the computer a large number of entities together with the rules of their interactions, and he or she observes what comes out of these interactions, including changes in the interaction rules themselves. Furthermore, computers
Language as pragmatics 143
have greater memory and processing capacities than those of human researchers. They can store large knowledge bases that belong to various disciplines and they can run interdisciplinary theories and models that invoke all these diŸerent knowledge bases. In this way simulations function as a common language, a lingua franca, among disciplines that tend to use diŸerent theoretical frameworks and methods. All these advantages are clearly important for studying such a complex phenomenon as human language, but they are critical for addressing the problem of how to study the meanings of linguistic forms (Knight, StuddertKennedy & Hurford 2000, Parisi 1997, Parisi & Cangelosi 2002). If the meanings of linguistic forms are the roles that these forms play in the behavior and interactions of human beings with each other and with the environment, one can simulate a certain number of individuals who live in an environment which contains various types of object, including other individuals of the same species (co-speciªcs). Each individual interacts with the diŸerent objects and with his or her co-speciªcs. To interact means that the environment causes inputs in the sensory receptors of an individual, and the individual responds by generating movements of his or her body that modify the environment. Linguistic forms are sensory inputs for hearers produced by the phono-articulatory movements of speakers. Each individual’s behavior is controlled by a neural network which simulates the individual’s nervous system. A neural network is a set of units (neurons) linked by unidirectional connections or synapses (Rumelhart & McClelland 1986). Connections can be excitatory or inhibitory, and each connection has a “weight” which simulates the e¹cacy of the synapse (number of synaptic sites). A neural network includes three layers of units: one layer of input units that encode states and events in the external or internal (bodily) environment; the input units are connected to one layer of internal neurons that encode transformations of the input encodings; the internal units are connected to one layer of output units that produce either physical movements of the individual’s body and, as a result, changes in the external environment or changes inside the body. The neural networks of human beings include many recurrent circuits that produce self-generated inputs which constitute “mental life”, that is, mental images, memories, thoughts, etc. The regularities that are present in the environment and in the interactions of the individual with the environment are captured and represented in the particular synaptic weights of the connections linking the network’s units. It is these weights that cause an individual’s neural network to respond to each
144 Domenico Parisi
input with the appropriate output, that is, with an output which tends to increase the individual’s survival and reproductive chances and his or her ‘happiness’. Synaptic weights are generally learned, i.e., they change during the individual’s interactions with the environment in such a way that the individual tends to respond to the input with more appropriate output. Linguistic forms are sounds produced by the phono-articulatory movements of an individual (speaker) and encoded in the acoustic receptors of another individual (hearer). What gives meaning to these sounds is that in the individual’s experience (i.e., in the interactions of the individual with the environment) speciªc sounds tends to co-vary with speciªc aspects of the individual’s experience. Speciªc sounds co-vary with speciªc objects that are seen, touched, or manipulated, with speciªc properties of these objects, with speciªc events or processes, with speciªc actions executed by the individual or seen by the individual as they are executed by another individual. These regularities and co-variations are captured and represented in the individual’s synaptic weights, and it is these synaptic weights that give to each sound its meaning for the individual. In computer simulations, one can study (observe, analyze, measure, manipulate) both the linguistic forms produced or heard by individuals and the covariations of speciªc forms with speciªc aspects of the individuals’ experiences in the environment. In this way, simulations make it possible to overcome the asymmetry between linguistic forms and their meanings from the point of view of the possibility of observing and measuring the latter, which, as we have argued, is the main problem for a science of language. Both can be simulated.
3.
Some concrete examples
An individual lives in an environment that contains both edible and poisonous mushrooms (Cangelosi & Parisi 1998). To survive and reproduce, the individual must eat the edible mushrooms and avoid the poisonous ones. Since each individual mushroom is perceptually diŸerent from all other mushrooms, the individual’s task is to recognize each encountered mushroom as edible or poisonous, and to respond to edible mushrooms by approaching and eating them and to poisonous mushrooms by going away from them. When it encounters a mushroom, the individual can trust its personal capacity to discriminate the two categories of mushrooms, or there may be a more expert co-speciªc who wants to help the individual and tells the individual what the
Language as pragmatics 145
category is of the mushroom. The co-speciªc produces one particular sound if the mushroom is edible, and another, diŸerent sound if the mushroom is poisonous. In another scenario, the individual may see a mushroom, but the mushroom is too distant for the individual to be able to perceive su¹ciently well the mushroom’s perceptual characteristics and recognize its category. Hence, for the individual the only available strategy is to approach all distant mushrooms until it is su¹ciently close to a mushroom to be able to recognize its category — which would not be a very energy-e¹cient strategy. In these circumstances, if a co-speciªc is closer to the mushroom and informs the individual of the mushroom’s category, the individual would be able to approach only edible mushrooms — with a clear advantage in terms of the energy consumed for moving in the environment. By realizing these simple scenarios as computer simulations, one can study in which conditions a simple language like this can emerge given various environmental constraints, how the language can come to possess such desired characteristics as social ‘sharedness’ and limited homonymy and synonymy, what induces an individual to help another individual by sending out informative signals, and how the neural network of both speakers and hearers is internally organized compared to the neural network of individuals without language. Are the two signals that are exchanged in the preceding (very simple) scenario verbs or nouns? Should they be translated by the two English nouns (or, more precisely, noun phrases), edible mushroom and poisonous mushroom, or by two English verbs, approach and avoid? The scenario appears to be too simple for deciding the matter one way or the other. In the experiences of the speaker and the hearer, the two sounds co-vary with the type of mushroom and the type of action at the same time. Hence, the two sounds cannot be classiªed as nouns or as verbs. But consider a somewhat more complex scenario (Parisi, Cangelosi & Falcetta 2002). An individual is endowed with a retina on which one or more diŸerent objects can appear at the same time, and with an arm with which the individual can reach and manipulate the objects. When there is a single object on the individual’s retina, another individual produces one sound to which the ªrst individual responds by pushing the object away from itself, and another sound to which the ªrst individual responds by pulling the object towards itself. If two diŸerently shaped objects are present at the same time on the individual’s retina, the co-speciªc produces one sound in response to which the ªrst individual executes some particular action on a speciªc one of the two objects, and another sound in response to which the
146 Domenico Parisi
same action is executed on the other object. Since in the ªrst case the two sounds co-vary with the two diŸerent actions, the two sounds can be interpreted as proto-verbs; in the second case the two sounds co-vary with the objects on which an action is executed and they can be interpreted as protonouns. If both a proto-verb and a proto-noun are produced in a sequence by the co-speciªc, so that the hearer knows which action to execute and on which object to execute the action, we can talk of a proto-sentence. In this case we can study how proto-nouns and proto-verbs emerge in the simulated scenario, which class of words emerges ªrst, what happens, and where, in the neural network of both speakers and hearers when they produce or hear proto-nouns and proto-verbs. For example, it turns out that in the neural network of our artiªcial organisms, which includes successive layers of internal units, proto-nouns are represented in internal units that are closer to the sensory input units whereas proto-verbs are represented in internal units closer to the motor output units. The simulated scenarios can be appropriately modiªed and multiplied to capture other classes of words and meanings. For example, if on the individual’s retina there are many diŸerent objects that can vary in both shape and color, to allow the hearer to identify the object on which he or she must execute some particular action the speaker must not just produce a proto-noun but both a proto-noun and a proto-adjective, i.e., a proto-noun phrase. (In actual languages nouns tend to co-vary with the shape of objects, adjectives with their other properties; cf. Landau, Smith & Jones 1988, 1992.) Or an object can be identiªed by its spatial relation to another object. In these circumstances, the speaker must provide the hearer with both a sound that co-varies with the other object, and a sound that co-varies with the particular spatial relation of the object to be identiªed with the other object, e.g., under. In all the simulations that we have described, speakers and hearers use linguistic forms that co-vary with aspects of public experience. An object or an action is experienced by both the speaker and the hearer. But language can also be used to talk about private experiences. Mothers can teach their children to produce some speciªc linguistic form when they have external, i.e., public, evidence that their children are having some speciªc private experience, such as pain. In this way individuals can learn to use language to communicate about their mental life with others. In our simulations individuals have a physical body and they live in a physical environment, but these physical aspects are very simpliªed and simulated in a computer. In other researches the physical aspects are still
Language as pragmatics 147
simulated but they can be more realistic (Marocco, Cangelosi & Nolª 2003). The emergence and self-organization of language can even be studied in artiªcial organisms which have a real physical body and live in a real physical environment, that is, in robots. In these conditions one can study the emergence of a language in a collection of robots or between robots and humans (Steels 2002).
4.
Conclusion
Computer simulations of how linguistic forms are used by a population of individuals interacting with the environment and among themselves can be a useful tool to study the meanings of linguistic forms. Human behavior remains a di¹cult phenomenon to observe, analyze, and measure and therefore the meaning of linguistic forms, being part of such behavior, remains a di¹cult topic for science. However, computer simulations are an important help for formulating theories and models of meaning in explicit, detailed, complete, and consistent ways, for publicly demonstrating the detailed empirical predictions derived from such theories and models, and for generating new empirical predictions to be compared with the actual observed phenomena. Of course using computer simulations to study how language is used has its costs. Simulations may seem to be appropriate for studying simpliªed languages in simpliªed environments, but language is very complex and it can be used to talk about very complex objects and events. Our simulations resemble Wittgenstein’s language games, which are simpliªed models of the complex roles that language plays in our everyday life and are similar to the games “by means of which children learn their native language” and to a “primitive language” (Wittgenstein 1953: 5). However, there are two arguments in favor of the method proposed for studying the meanings of linguistic forms. Simulations are theories and as theories they simplify, like all theories in science. Scientiªc theories and models allow us to go beyond descriptions of the variety of observed phenomena and to discover what lies behind these phenomena and explain them, precisely because they simplify and capture what is essential in the observed phenomena. Therefore, the fact that simulations simplify is not in itself an objection. Of course, like all theories, simulations must make the correct simpliªcations. What can be objected in speciªc cases is that some particular simulation makes the wrong simpliªcations, that is, that it leaves out aspects
148 Domenico Parisi
of the actual phenomena which are critical for explaining and understanding the phenomena of interest. But simulations, like all theories, must simplify. The second argument in favor of our linguistic simulations is that, although they resemble Wittgenstein’s language games, they are diŸerent in one important respect. The simulations that we have described are not just language games, they are physically realized language games. They actually run on a computer or control physical robots. As we have observed, computers have memory and processing resources that go well beyond those possessed by humans. Given the great memory and processing capacities of computers, we can expect that in the future the kinds of linguistic simulation that we have described will simplify less and will make it possible to study more complex languages in more complex environments. The approach to the study of language which underlies our simulations is based on the following assumption of “genetic epistemology”: if you want to understand X, try to reconstruct how X has become what it is now. Hence, we are forced to start from simple languages for simple organisms in simple environments. Then, we can move on and study how language changes and possibly becomes more complex in evolution, ontogeny, and language change (Parisi & Cangelosi 2002). On a more general level, our simulations lead to a view of language as entirely constituted by pragmatics. Traditionally, one distinguishes three aspects of language: semantics, which is concerned with the meaning of linguistic forms, syntax, which is concerned with the rules for creating linguistic forms made up of more elementary forms, and pragmatics, which studies the purposes of speakers and how language is actually used in context. Wittgenstein’s idea of language games and our implementation of this idea in computer simulations imply a pragmatic view of language. Everything in language is language use. Language is the use of linguistic forms in the behavior and social interactions of a particular species of organisms. Semantics and syntax are only abstractions or idealizations from language use. This does not mean that semantics and syntax are nonexistent or unreal. But, as abstractions from language use, semantics and syntax cannot be really understood unless we ªrst understand the uses of language from which they are abstracted.
Language as pragmatics 149
References Cangelosi, Angelo and Parisi, Domenico 1998 “The emergence of a ‘language’ in an evolving population of neural networks”. Connection Science 10: 83–97. Knight, Chris, Studdert-Kennedy, Michael and Hurford, James (eds.) 2000 The Evolutionary Emergence of Language: Social Function and the Origins of Linguistic Form. Cambridge: Cambridge University Press. Landau, Barbara, Smith, Linda B. and Jones, Susan S. 1988 “The importance of shape in early lexical learning”. Cognitive Development 2: 291–321. 1992 “Syntactic context and the shape bias in children’s and adults’ lexical learning”. Journal of Memory and Language 31: 807–825. Marocco, Davide A., Cangelosi, Angelo and Nolª, Stefano 2003 “The emergence of communication in evolutionary robots”. Philosophical Transactions of the Royal Society London A 361: 2397–2421. Parisi, Domenico 1997 “An Artiªcial Life approach to language”. Brain and Language 59: 121–146. Parisi, Domenico and Cangelosi, Angelo 2002 “A uniªed simulation scenario for language development, evolution, and historical change”. In Simulating the Evolution of Language, A. Cangelosi and D. Parisi (eds.), 255–275. New York: Springer. Parisi, Domenico, Cangelosi, Angelo and Falcetta, Ilaria 2002 “Nouns, verbs and language games”. Italian Journal of Linguistics 14: 99– 114. Rumelhart, David E. and McClelland, James L. 1986 Parallel Distributed Processing: Explorations in the Microstructure of Cognition. Cambridge, MA: MIT Press. Steels, Luc 2002 “Grounding symbols through evolutionary language games”. In Simulating the Evolution of Language, A. Cangelosi and D. Parisi (eds.), 211–226. New York: Springer. Wittgenstein, Ludwig 1953 Philosophical Investigations. Oxford: Blackwell.
Sharing… But why a language or world?* Bart Vandenabeele Ghent University
Herman Parret taught me the vital philosophical importance of the concept of sharing. I owe much to Herman for that, since it has occupied my thinking ever since and my discussions with him helped me greatly in shaping my own thinking. In this essay I want to explore the relevance of this concept in the domain of (intercultural) communication, confronting thoughts by Donald Davidson and Hans-Georg Gadamer on the subject. At ªrst sight, it may seem strange to confront two such diŸerent thinkers as Gadamer and Davidson in one and the same essay. Do they not belong to what became known as two very diŸerent philosophical traditions? They have a lot in common, though: they both thoroughly studied Plato in their younger years and shared the same interest in language, truth, and interpretation. It is not without good reason that two people who belong to diŸerent traditions are played out against each other here, since the mainspring of this essay is that communication between people, no matter whether they belong to the same tradition or not, is possible, and that the requirement for communication between people is intimately connected with their active participation in traditions. Tradition is obviously one of the basic terms in Gadamer’s philosophical hermeneutics. Although neither Gadamer nor Davidson thought or * The larger part of this text was written during my stay as a visiting professor at the University of Stellenbosch, South Africa, from 18 February till 17 April 2003. Previous versions of this paper were read at the University of Cape Town, the University of South Africa (Pretoria), and the University of Stellenbosch. It was also delivered at a conference on Interculturalism in Milan, Italy, on 31 October 2003. I wish to thank all those who asked questions or made interesting suggestions on those occasions. I am greatly indebted to Jaap van Brakel for raising my interest in the subject of intercultural communication in the first place, and for his many useful comments and suggestions. I also wish to thank Arnold Burms for his comments on a previous version of the text.
152 Bart Vandenabeele
published extensively about intercultural communication, it is my conviction that a confrontation between them is deªnitely important and interesting from an intercultural point of view. And since a motive behind this paper is to go beyond schisms, I will use a variety of approaches to get across. The idea that to understand one another (best, ideally), one needs to share a language is widespread: “only on rare occasions there is a hint that there might be something ªshy about the single language approach to negotiation or communicative interaction” (van Brakel 2003: 152).1 Davidson is one of the very few philosophers who explicitly uttered qualms about the idea and necessity of a common language. I shall argue that Davidson is right to claim that a common language is not needed for successful communication. Furthermore, Davidson (1984: 190) rightly rejects the conception of a world as “something neutral and common that lies outside all schemes”, and this chimes well with Gadamer. But Davidson has been unable to develop this insight fruitfully and underestimates the importance of the thought that language and communication are ªrmly embedded in what Gadamer calls a “tradition” (cf. Gadamer 2001: 160, 174–175, 217, 275–277, 280, 294–296, 300, 335–336, 357–361, 389– 391, 395, 441, 463, 533, 535, 537–538, 541, 562, 571). Some synthesis is possible, although these thinkers draw on quite diŸerent philosophical traditions. However, as we shall see, neither comes out of the confrontation unscathed. I will start by sketching their respective views on language, communication, and interpretation, and then outline my major disagreements with both.
1.
Gadamer’s common language “syndrome”
The presupposition that a single language warrants communication is linked with the ideal-language approach, which pervades much of analytical philosophy and can also be found in some “Continental” philosophers (e.g., Derrida 1991; Merleau-Ponty 1964: 120; Lyotard 1984: 66; Taylor 1994: 125; and Tully 1995: 133f.). Strictly speaking, an ideal language is objective, i.e., all idiosyncratic and poetic aspects have been eliminated, and compositional, in that the reference and sense of each expression can be “calculated” from the reference and sense of its constituents, and each sentence is either true or false.2 Nothing could be further from Gadamer’s view. Yet I shall use “ideal language” in a broader sense, following van Brakel (2003: 153–154): it also covers any stance or theory assuming that for any sort of understanding or communication to work, one must (ideally) speak the same language. (From this perspective both
Sharing… But why a language or world?
the universalist and the relativist suŸer from the ideal-language syndrome.) In this broader sense, the ideal language “syndrome” also pervades Gadamer’s work. Gadamer contends that: [e]very conversation presupposes a common language, or better, creates a common language. Something is placed in the center, as the Greeks say, which the partners in dialogue both share, and concerning which they can exchange ideas with one another. Hence reaching an understanding on the subject matter of a conversation means that a common language must ªrst be worked out in the conversation. This is not an external matter of simply adjusting our tools; nor is it even right to say that the partners adapt themselves to one another but, rather, in a successful conversation they both come under the in¶uence of the truth of the object and are thus bound to one another in a new community. To reach an understanding in a dialogue is… being transformed into a communion, in which we do not remain what we were. (Gadamer 2001: 379; italics added)3
Gadamer contends that “there is nothing that is fundamentally excluded from being said, to the extent that our act of meaning extends it. Our capacity for saying keeps pace untiringly with the universality of reason.” (Gadamer 1966: 67). In Gadamer’s writing, there is a strong “longing” for a common ground and/or common language: “Language always presupposes a common world — even if it is only a play world.” (Gadamer 2001: 406). But what does he really mean by the necessity of a common language? According to Gadamer, successful hermeneutic understanding (Verstehen) is coming to a dialogic consensus (Verständigung). He writes that “there can be no speaking that does not bind the speaker and the person spoken to” (Gadamer 2001: 397). His conviction that “nothing is more di¹cult than a dialogue in two diŸerent languages in which one person speaks one and the other person the other, each understanding the other’s language but not speaking it” only accentuates this point (Gadamer 2001: 384). What Gadamer wants to stress is not that the successful outcome of Verstehen favors either the initial claims of the interpreter or those of the other. Rather, just as the conclusion of a genuine conversation is not the sole property of either one of the dialogue partners, the outcome of “understanding” (Verstehen, Verständigung) is a unity or agreement that goes beyond the original positions of the various participants, and represents a new stage in the development of the tradition. Every understanding is embedded in what Gadamer calls a horizon. What happens in interpretation is a “fusion of horizons” (Horizontverschmelzung). More accurately, for a horizon of interpretation to occur, it is necessary that a fusion of horizons comes about (Gadamer 2001: 306–307, 374–375, 397, 576). Interpretation and understanding are essentially connected to one’s own
153
154 Bart Vandenabeele
horizon. In other words, when one understands a text or person, one does not give up one’s previous horizon in the fusion of horizons. On the contrary, an expansion of one’s own horizon takes place in this fusion, which is the process of understanding, where the new view is incorporated into one’s former horizon. The exercise of this hermeneutical capacity means that what is said “has a claim over one, and this is impossible if one’s own ‘world-view and languageview’ is not also involved” (Gadamer 2001: 442). There is some kind of agreement which is more “originary” than the mutual agreement or consensus Apel and Habermas talk about: “in a real community of language… we do not ªrst decide to agree, but are always already in agreement, as Aristotle showed” (Gadamer 2001: 446; see also 431f.), and it is “a life process in which a community of life is lived out” (ibid.). Again, this is closely related to the idea of a horizon in Gadamer: “each one potentially contains every other one within it — i.e., each worldview can be extended into every other” (Gadamer 2001: 448). This implies that, although Gadamer never explicitly deals with understanding or interpretation across linguistic and or community boundaries, the idea of horizons being fused easily transposes to any occurrence of understanding which involves overcoming some initial strangeness. The horizon constituted by a speciªc “situatedness” is not closed; it is always open to “fusion” with other horizons (see also McDowell 2002: 176). Regardless of how intimately related the theme of the common language is to the idea of a fusion of horizons, it remains unclear what Gadamer exactly means with a common language. Up to a certain point, language is always something that is shared, according to Gadamer. (Thus, Gadamer concurs with Schleiermacher that everything presupposed in hermeneutics is only language.) This stress on the necessity of a common language is even more problematic when we see that, as Hegel does, Gadamer stresses the ongoing continuity of the process of interpretation through which both parties become entangled in a universal history of reason and truth. The views of a tradition are not simply adopted but modiªed in accordance with changed historical circumstances, and understanding “is always… guided by the transcendental expectations of meaning that stem from the relation of what is meant to the truth” (Warnke 1987: 97). Gadamer seems to distinguish between two diŸerent senses of agreement, one of which entails a concrete unity of judgment, and the other critical and re¶ective integration (Warnke 1987: 106). Warnke also argues that by reducing the second to the ªrst, Gadamer remains caught in a conservative and Hegelian, euro-centric idea of evolvement and the history of truth: if what someone says, coheres with our own understanding, then it appears rational or plausible to us.
Sharing… But why a language or world?
Yet Gadamer deserves more credit, since — unlike, e.g., Habermas and Apel — he does not feel any need at all to appeal to an “ideal speech situation” or an “unlimited transcendental community of interpretation”, which purportedly oŸer criterions for determining whether or not a traditional consensus betrays the eŸects of force and coercion.4 Gadamer is surely right to deny that there is any need for an appeal to an ideal speech situation — counterfactual or not. He also rightly denies that there should ever be one correct or absolutely exhaustive way of understanding oneself or one’s way of life. Thus, he would agree with Catherine Elgin’s contention that it is obviously true that either there is one right answer or there is not, but that “the error arises when we interpret the second disjunct as ‘or there is none’. One way there can fail to be one right answer is that there is none. Another is that there are several. To say that a problem does not admit of a unique solution is not to say that it is unsolvable or that all proposed solutions are equally good.” (Elgin 1997: 194) If we substitute “interpretation” for “solution” in Elgin’s words, we end up with Gadamerian hermeneutics. If, on Gadamer’s account, language, world, horizon, and tradition are always intertwined, then the notion of a common language should also be read in the light of this intertwining. A common language does not so much refer to a shared grammar or shared rules, or to some learnable common core of consistent speech, but rather to a shared practice. But is it really necessary to presuppose a shared (linguistic) practice for communication to work? Davidson thinks it is not. In the following sections, I shall argue that Davidson is right to deny that a shared ability to produce linguistic behavior in conformity to speciªable rules could su¹ce for mutual understanding to take place, and that there is no portable interpreting machine set to grind out the meaning of an arbitrary utterance. Yet Davidson is wrong in completely denying the relevance of a shared practice or tradition, as if the philosophical notions of tradition or practice could be interesting only “if it were true that [tradition] su¹ces, of itself, for mutual understanding between parties” (McDowell 2002: 184). I shall discuss this in the third part of the essay. Let us have a look ªrst at Davidson’s arguments against the concept of a common language.
2.
Davidson: No need to speak the same language
Davidson is one of the very few philosophers to cavil at the idea that it is necessary to speak the same language in order to communicate. Davidson explicitly criticizes Gadamer’s idea of a common language. After quoting Gadamer extensively, Davidson says:
155
156 Bart Vandenabeele
Where I diŸer [from Gadamer] (and this may merely show I have not fully understood Gadamer) is that I would not say a conversation presupposes a common language, nor even that it requires one. Understanding, to my mind, is always a matter not only of interpretation but also of translation, since we can never assume we mean the same thing by our words that our partners in discussion mean. What is created in dialogue is not a common language but understanding: each partner comes to understand the other. And it also seems wrong to me to say that agreement concerning an object demands that a common language ªrst be worked out. (Davidson 1997: 431–432)
Although this quotation posits the awkward interdependence of translation and interpretation — awkward, since Davidson clearly separates translation and interpretation in “The social aspect of language”5 — it does reveal one of the important diŸerences between Gadamer and Davidson. In Truth and Method, Gadamer stresses the importance of a shared language, whereas Davidson (1986) sees clear reasons to doubt that sharing a language is essential. The same doubts apply to the notion of following a rule, engaging in a practice, or conforming to conventions, if these are taken to imply such sharing.6 What must be shared for communication to succeed is not a conventional system, such as a natural language. Sharing a language may, of course, facilitate linguistic understanding, but such sharing is not essential for understanding to be possible — only a passing theory is (Davidson 1986: 442). What Davidson calls a passing theory is the theory “the interpreter actually uses to interpret an utterance, and it is the theory the speaker intends the interpreter to use. Only if these coincide is understanding complete.” (ibid.) A passing theory is distinguished from a prior theory in Davidson’s terminology. The hearer’s prior theory indicates how he is prepared in advance to interpret an utterance of the speaker, while the passing theory is how he does interpret the utterance. Not only is it unnecessary to presuppose a shared language for communication to succeed, it is also the case that most of the time prior theories will not be shared, and “there is no reason why they should be”, since (as Davidson argues) it “is not a condition of successful communication that prior theories be shared” (1986: 443). Neither passing nor prior theories are what we would normally call a language, and neither one of them characterizes a speaker’s or interpreter’s linguistic competence. Understanding and agreement do not arise when a common language is worked out but when passing theories coincide. Mastery of an actual language is even useless, Davidson claims, “since knowing a passing theory is only knowing how to interpret a particular utterance on a particular occasion” (1986: 443). Hence, linguistic ability is the ability “to converge on a passing
Sharing… But why a language or world?
theory from time to time”, and consequently “we have abandoned… the ordinary notion of a language” and have to “conclude that there is no such thing as a language, not if language is anything like what many philosophers and linguists have supposed. There is therefore no such thing to be learned, mastered, or born with.” (Davidson 1986: 445–446) Although Davidson would disagree, I believe that this drastic conclusion implies a radical rupture with his earlier attempts to establish a theory of interpretation modeled after a Tarskian theory of truth. If the idea of a language is considered super¶uous for a theory of interpretation, and if sharing passing theories is enough for successful communication, then at least one of the early important claims of Davidson’s has been shaken, viz., that “speakers belong to the same speech community if the same theories of interpretation work for them” (Davidson 1984: 135), and that: central to interpretation… is a theory of truth that satisªes Tarski’s Convention T (modiªed in certain ways to apply to a natural language). Such a theory may be taken as giving an interpretation of each sentence a speaker might utter. To belong to a speech community — to be an interpreter of the speech of others — one needs, in eŸect, to know such a theory, and to know that it is a theory of the right kind. (Davidson 1984: 161; italics added)
The rejection of the ordinary notion of a language has as an important but highly problematic consequence that the idea of a linguistic community, i.e., the concept of people sharing a common linguistic practice by using the same theories of interpretation, is discarded. Davidson’s contention that sharing passing theories, and not sharing languages, is su¹cient for understanding is a radical rupture with his own theoretical endeavor: instead of defending the possibility of a Tarskian theory for a language (community), which would be applicable to all speakers belonging to the same speech community — or, more precisely, which could oŸer a good criterion for identifying speech communities — , Davidson is now only left with the theoretical importance of idiolects and passing theories. Thus the aim of developing a theory of interpretation for a linguistic community, i.e., a Tarski-style theory which should be shared (in principle) by speakers belonging to the same linguistic community, can no longer be defended by Davidson, since the whole notion of a linguistic community has been discredited. A second, and perhaps even more damaging, consequence of Davidson’s radical conclusion that there is no such thing as a language, is that it is not only true — as Davidson admits — that “an interpreter must be expected to have quite diŸerent prior theories for diŸerent speakers” (1986: 443). It also follows
157
158 Bart Vandenabeele
from his repudiation of the idea of a (common) language, and from his contention that only passing theories ought to coincide in communication, that an interpreter should have diŸerent theories for the same speaker. If Davidson is right, and I think he is, in claiming that linguistic ability is the ability to converge on a passing theory from time to time, then it does not only follow that the traditional notion of a language has been abandoned, as he justiªably concedes, but also that the traditional notion of an idiolect has been discredited — perhaps I should add: if an idiolect is anything like what many philosophers and linguists have supposed. If we accept the image of sharing passing theories as su¹cient for communication (and I have not yet seen any convincing argument not to), then the following should be accepted as correct, too. What we end up with is the image of a speaker speaking many passing idiolects for which an interpreter develops diverse, ever-changing passing theories. The notion of an idiolect as the language spoken by only one speaker does not exist or, put more carefully, does not need to be presupposed for communication to succeed. We must give up the idea of a language user’s idiolect, as what we are left with are only the speaker’s utterances at a given moment and in a particular situation, or a speaker’s passing idiolects in the plural. Davidson is right that a passing theory “is not a theory of what anyone… would call an actual natural language. ‘Mastery’ of such a language would be useless, since knowing a passing theory is knowing how to interpret a particular utterance on a particular occasion.” (Davidson 1986: 443) However, it seems to me that this implies that the notion of an idiolect has become as spurious as that of a language. In other words, for communication to succeed it is not only unnecessary to share a language, it is equally ¶awed to presuppose the existence of a speaker speaking only one idiolect. If this is right, we should instead say that a speaker speaks many idiolects, or — more precisely — that there is no need to appeal to (the unity of) a speaker’s idiolect, since its theoretical relevance for semantics and interpretation is as insigniªcant as that of the notion of a language. One may wonder whether this unforeseen consequence of the enterprise undertaken by the later Davidson is not completely damaging for building a theory that intends to map the conditions of communication. Moreover, as McDowell remarks, “Davidson’s claims commit him to denying that one needs to learn to speak as others do, in the ordinary sense, in order to become a potential party to an encounter with another that leads to mutual understanding at all” (McDowell 2002: 183). His conclusions are generally disparaging to shared practice; so is this not just one step too far? This question takes us back
Sharing… But why a language or world? 159
to Gadamer, since the notion of a linguistic community or tradition lies at the core of his philosophy.
3.
No need to share a world?
Would it not be possible to consistently argue that (i) it is not necessary to speak the same language — not only for people belonging to diŸerent linguistic communities, but also for those belonging to the “same” —, and (ii) that Davidson’s idiolectical point of view7 is ¶awed, unless it is supplemented with an account of the necessary embedding of language in a common world or praxis? To answer this question, it is important to compare Davidson’s and Gadamer’s notions of world, and of what sharing a world means in their respective theories. The idea of a fusion of horizons and the importance Gadamer attaches to tradition are closely related to the intertwinement of language and world. The linguistic world in which we live is not a barrier to understanding, but “fundamentally embraces everything in which our insight can be enlarged and deepened” (Gadamer 2001: 405). Gadamer argues that the use of the expression “world-in-itself” is problematical. A world is always, and necessarily, linguistically constituted, and “always open, of itself, to every possible insight and hence for every expansion of its own world-picture, and accordingly available to others” (ibid.).8 Still, it is interesting to note that Gadamer often seems to use the concept of a “world-in-itself” in two rather diŸerent ways. In a ªrst sense, he speaks of man “having a world”, referring to man’s attitude toward the world and his capacity to have a language. Learning to speak means acquiring familiarity and acquaintance with the world and how it confronts us (Gadamer 1966: 63). It is in this sense that it is impossible to speak of (having) a world-in-itself, for our relationship is ªnite and will never be absolute. There is an inªnite gap that can never be bridged. Just as there will never be one and only one “right” translation (Gadamer 2001: 349), which makes all other translations super¶uous (since all translation is “highlighting” and not a comprehension in the literal sense of the term), there will equally be no conversation, interpretation, or dialogue that has an absolute ªnality or deªnitive ending. In the footsteps of Heidegger, Gadamer contends that in a dialogue, there is a constant referring forward and backward to “what is unsaid”, never “having said everything” (Gadamer 1966: 67). The horizons of understanding cannot be limited.
160 Bart Vandenabeele
In a second sense, “world-in-itself” alludes to a world separate and distinct from our thoughts, concepts, and language. A world-in-itself gives rise to our language: “From the relation of language to the world follows its unique factualness [Sachlichkeit]. It is matters of fact [Sachverhalte] that come into language. That a thing behaves [eine Sache verhalt sich] in various ways permits one to recognize its independent otherness, which presupposes a real distance between the speaker and the thing.” (Gadamer 2001: 445; italics added) Even if communication is not in need of any tools — language is not a tool or a means of communication, Gadamer says (2001: 443 and passim) — it does need some common, original reference point: “The world is the common ground, trodden by none and recognized by all, uniting all who talk to one another.” (Gadamer 2001: 446) Although Gadamer acknowledges the importance of the diversity of linguistic communities (ibid.), he holds that we all share a common world in the living process of communication: in language, the world presents itself (Gadamer 2001: 447). As the world that it is, “it is not relative to a particular language either” (Gadamer 2001: 452). To live in a linguistic world does not mean that one is placed within an environment, as animals are in theirs. Sharing a language is always at the same time sharing our common humanity, according to Gadamer, i.e., sharing with everybody who has a language and, hence, “has” a world (Gadamer 2001: 453). As Gadamer writes: “In every worldview, the existence of the world-in-itself is intended.” (Gadamer 2001: 447) The worldin-itself is not a neutral world outside all viewpoints, schemes, or positions. It is rather the familiar world with which Davidson says “we re-establish unmediated contact with the familiar objects whose antics make our sentences and opinions true or false” (Davidson 1984: 198). From “within” our horizon, which is open to every other horizon, we talk about and live in the world-itself, and not in a world constituted by whatever we imagine or think: “the closed horizon that is supposed to enclose a culture is an abstraction” (Gadamer 2001: 304). On this point, Gadamer and Davidson are — at least partly — in agreement. However, Davidson does not insist that one needs to share a tradition and have a language in order to become a human subject at all. On Davidson’s account, and this is a blind spot in his theory, sharing a tradition, a world, or a way of life with someone are just aids for arriving at an understanding, but certainly not for becoming potential subjects of understanding (see McDowell 2002: 184–185). Gadamer claims that our worldview is always open to correctness and that in a conversation, in the course of coming to understand other people, we can be “bound to one another in a new community… transformed into a communion in which we do not remain what we were” (Gadamer
Sharing… But why a language or world?
2001: 379). When we understand another person, the horizon within which we see the world is no longer in the same position, but “what is in view, now that the horizon is in its new position, is still the world”, i.e., “everything that seems to be the case from within a horizon so positioned. There is no devaluing of reality’s independence from thinking here.” (McDowell 2002: 180) So there does not seem to be too much diŸerence between Gadamer’s fusion of horizons and Davidson’s conception of world and his ideas on radical interpretation. Yet there remain some important diŸerences in both authors’ conception of world, which are often overlooked in the literature. (McDowell, for one, clearly underestimates these diŸerences.) Moreover, according to Davidson, “unless two of us are built in such a way as to generalize in the same way in some basic cases, we couldn’t come to communicate”, and “without a built-in inventory of similar modes of response, we couldn’t get started” (Davidson 1995: 275).9 He also talks about “objects or events we naturally ªnd similar”, the mutually salient causes in the world, and argues that “the innate similarity responses of child and teacher — what they naturally group together — must be much alike” for communication to work (Davidson 1989: 199; 1992: 263). This is a view that is so common in philosophy, psychology, and (cognitive) linguistics that it has become natural indeed. Yet I wonder whether it is really a necessary precondition for communication. It is one thing to say that, without a social environment, nothing could count as misapplying words in speech and that interpretation and communication are a social matter, but quite another to claim that it is necessary to share a world in the way Davidson describes it, i.e., to ªnd particular basic objects and events naturally similar, in order to think and communicate. Davidson’s contention that some basic saliencies are necessary, is connected with his famous theory of triangulation. Triangulation is explained as follows: There is a prelinguistic, precognitive situation which seems to me to constitute a necessary condition for thought and language, a condition that can exist independent of thought, and can therefore precede it. Both in cases of nonhuman animals and in the case of small children, it is a condition that can be observed to obtain. The basic situation is one that involves two or more creatures simultaneously in interaction with each other and with the world they share; it is what I call triangulation. It is the result of threefold interaction, an interaction which is twofold from the point of view of each of the two agents: each is interacting simultaneously with the world and with the other agent. To put this in a slightly diŸerent way, each creature learns to correlate the reactions of other creatures with changes or objects in the world to which it also reacts. (Davidson 2001: 128)
161
162 Bart Vandenabeele
This is an interesting sketch of the simplest interpersonal relation: two creatures each correlate their own reactions to external phenomena with the reactions of the other. Davidson is correct in assuming that “the pure Robinson Crusoe case”, involving a Robinson Crusoe who has never communicated with others, is impossible (Davidson 2001: 115). Yet, pace Davidson, it does not follow that Robinson and Friday need to share the same particular natural categories, objects, and events so as to be able to communicate, interpret, and understand each other. That is not the only problem with Davidson’s account. In his famous article “On the very idea of a conceptual scheme”,10 Davidson claimed that there is no globally independent reality accessible by means of privileged sensations organized by a scheme (of metricization, measurement, or laws). Instead of a distinct ontological, globally mind-independent reality, which we visually “discover” and “use” (however skillfully), there is local “experienced reality”. Davidson adds that communication does not demand that languages be shared, “that Kripke’s criterion for speaking a language cannot be right”, and that (successful) communication “does not depend on two or more speakers speaking in the same way; it merely requires that each speaker intentionally make himself interpretable to the other (the speaker must ‘go on’ more or less as the other expects, or at least is equipped to interpret)” (Davidson 2001: 115). So far, so good. Yet if Davidson sticks to his picture of triangulation, and if there is no social practice with which to compare the speaker’s performance, will whatever the speaker says not be (as Wittgenstein remarks) in accord with some rule(s)? If the speech behavior of others does not provide the norm for the speaker, then what can? Davidson answers that it is the intention of the speaker which provides the “norm”: “the speaker falls short of his intention if he fails to speak in such a way as to be understood as he intended” (Davidson 2001: 116). So Davidson thinks that any tie with a linguistic community is not necessary to account for the possibility of communication. Such a tie comes into play only when the speaker intends to be interpreted as (certain) others would be. But is this not always — and necessarily — the case? How can a speaker ever be interpreted as communicating something, unless she intends to be interpreted as others would be? Although this does not threaten the argument that it is not necessary to share a language to communicate in a meaningful way, it seems to me that this reference to the speaker’s intention is highly contentious. Is the intention of the speaker to be interpreted in a certain way alone really su¹cient to explain successful communication? If so, where does this intention come from?
Sharing… But why a language or world? 163
The answer is fairly simple: the possibility of intending to be interpretable in a certain way originates in the (often unconscious) habit of interpreting and being interpreted by other people with whom I am engaged in a common linguistic practice or community. An intention to be interpreted in a certain way cannot be the ultimate norm. This intention is itself immersed or embedded in a shared way of life and speech community. As Davidson rightly remarks at one point: “there would still have to be a society (or at least another person) legitimately to judge that Robinson Crusoe meant something by his noises” (Davidson 2001: 115n). One may wonder what could be meant by Davidson’s subsequent claim that a failed contention to speak “correctly”, unless it foils the intention to be interpreted in a certain way, does not matter to what the speaker means. I do not see how an intention to speak “correctly” and an intention to be interpreted in a certain way can be distinguished as sharply as Davidson suggests. Someone who really wants to be understood will try and speak as correctly as she can. (Naturally, this does not mean that she will always speak “correctly”, nor that her making mistakes will inevitably block communicative interaction.) The ability to intend some (speech) act to be understood has to be learnt, just as the ability to perform a lay-up in a basketball game. What is the upshot of all this? Davidson’s position will have to take recourse to a more fundamental preoccupation with what it means to do as others do, since it is impossible to intend to be interpreted in a certain way without taking part in (or having grown into) a community of people with similar communicative intentions. A linguistic practice with speakers who intend to be interpreted in a certain way and who intend to speak “correctly” or “meaningfully” is thus necessarily presupposed in explaining the very possibility of linguistic behavior and mutual understanding. We cannot separate our ability to interpret and understand from tradition(s), for we are part of tradition(s), and it is through being part of tradition(s) that we interpret and understand. It is in Gadamer’s writings that we can ªnd a more detailed account of this common practice or tradition. Like Davidson, Gadamer argues that it is only in interpersonal communication that there can be thought, the grasping of an objective, that is, a shared world. The aim of conversation is “shared understanding”: all understanding is interpretation and takes place in the medium of language, which would allow the object to come into words. But the concept of a world in Gadamer is diŸerent from Davidson’s. Whereas in Gadamer the concept of world is intimately related to that of tradition, with Davidson “world” merely refers to the triangle in his conception of triangulation. This triangle, Davidson argues, is “essential to the existence, and hence to the
164 Bart Vandenabeele
emergence of thought” (Davidson 2001: 129). Without the triangle, we cannot account for the objectivity of thought and the empirical content of thoughts about the external world. I do not believe, though, that Davidson’s triangle provides a su¹cient ground for interpersonal communication. Wittgenstein has suggested, or I take him to have suggested, that we would not have the concept of getting things wrong or right if it were not for our interactions with other people. Davidson then argues: Once these correlations are set up, each creature is in a position to expect the external phenomenon when it perceives the associated reaction of the other. What introduces the possibility of error is the occasional failure of the expectation; the reactions do not correlate. […] Social interaction, triangulation also gives us the only account of how experience gives a speciªc content to our thoughts. Without other people with whom to share responses to a mutual environment, there is no answer to the question what it is in the world to which we are responding. (Davidson 2001: 129)
Davidson insists that thought and language are social, and that “there must be an interacting group for meaning — even propositional thought […] to emerge” (Davidson 1994: 10), but this “social” in Davidson is a rather thin concept; Davidson merely oŸers a “duetist” account of communication (cf. Hacking 1986: 458). Anthropological literature abounds with excellent examples of contacts between people from diŸerent backgrounds who are able to engage in communication at a considerable level of sophistication (see Vandenabeele 2002, 2003), and so far Davidson’s point seems proven. However, if these people do not mutually presuppose that the other belongs to a speech community, that is, that the other is — as Wittgenstein would say — doing as others would do, communication would not even start. Of course, the “others” may sometimes be wrong and we do not have to presuppose any ªxed conventions. The point is, however, that it is not enough to share responses to a common environment with another creature. This other creature needs to be recognized as a linguistic being, i.e., I have to recognize the other as — at least roughly — sharing speech with others. This does not mean that we have to presuppose some elitist group of standard-language speakers (e.g., all the speakers of standard Bengali), but that the notion of semantic correctness only arises within broader linguistic communities or practices. There is no answer to the question what makes communication possible, unless some recourse to (loosely connected) communities, practices, or traditions is recognized. Davidson is right that thought and language are necessarily social, but his account of this social aspect is both too minimalist and too strict. It is too
Sharing… But why a language or world? 165
minimalist, since it fails to explain how an intention to be interpreted in a certain way may provide the ground for mutual understanding. It is also too strict, since it unjustiªably presupposes that people have to share standards of similarity to be able to engage in communicative practice.
4.
Conclusion
The con¶ict with Davidson and Gadamer is one that I cannot pursue further here. I hope to have made clear why I believe that Davidson’s criticism of Gadamer’s idea of a common language is valid. However, as an explanation of what is required for (intercultural) communication, Davidson’s “duetist” picture of triangulation needs to be supplemented with an account of the impact of a linguistic community or tradition — which, I should add, does not imply that we should replace Davidson’s account by the idea that language is a purely conventional system, as, e.g., Dummett holds. Davidson’s notion of idiolect must be de-reiªed and de-essentialized, and so should Gadamer’s notion of tradition, if these are to be used to describe and explain what happens when people (whatever their backgrounds) communicate with each other. Davidson’s theory must be further “enriched” by a historical and tradition-constituted approach, otherwise it cannot explain how the intention to understand someone and to be understood comes about. Yet tradition itself must be neither homogenized nor reiªed, and communication cannot be restricted to a fusion of horizons, openness, or tolerance: one of the perils of this is that in the name of tradition and openness in theory, “the procedural rules redeªne that openness on so narrow a base that the status quo changes not one jot” (Saunders 2003: 139). It is of the utmost importance nowadays, in a world in which a complex network of “Euro-colonial worlds have set the scene for a permanent tra¹c in peoplehoods, selfhoods and identities” (Saunders 2003: 132), to reestablish contact with traditions or practices, in all their vagueness and historicity. By this I do not just mean tradition or practice in the abstract, but I understand them concretely and both in the plural and the singular at the same time: there is a tradition (practice, life-world,…) which all human beings share, and there are many diŸerent traditions (practices, life-worlds,…) of human beings. Yet is Gadamer right when he restricts communication and interpretation to the creation of a common language and horizon? I do not think so. People may, of course, act together, and they often do — that is what relatedness and
166 Bart Vandenabeele
communication is all about. But, strictly speaking, people never act in a common language — this is what Davidson rightly argued, and never completely in a common world either — this is where I disagree with him. In and through such relatedness, there is incommensurability: this cannot and need not be overcome in a universal theory or conventional system of meaning, nor in a successful fusion of horizons, let alone in rational consensus. There is, of course, always a temporarily shared scene or inter-space, but this can be maintained only if the plurality of perspectives that constitute it is preserved (Young 1997: 359). Utterances are always contested. Communicative agents rarely, if ever, have a “common will”, but they do share a local, changing environment or public space of contestation. These split sites of enunciation may well be the conditio sine qua non for (intercultural) communication to take place.
Notes 1. See also, for instance, Hoy (1991: 173): “The tacit assumption in Taylor’s and Habermas’s account that I find and question is that the different languages would converge ideally into a single language… I suspect that the tendency to assume convergence results from a vestigial commitment to monism.” And MacIntyre (1988: 384): “This distortion by translation out of context — from the standpoint of those who inhabit the traditions from which the distorted texts are taken — is of course apt to be invisible to those whose first language is one of the internationalised languages of modernity.” 2. Following Frege (1892), on whose work the formal-language approach is based. See also van Brakel (2003). 3. Other related quotations are: “Every language can be learned so perfectly that using it no longer means translating from or into one’s native tongue, but thinking in the foreign language. Mastering the language is a necessary precondition for coming to an understanding. Every conversation obviously presupposes that the two speakers speak the same language.” (Gadamer 2001: 385) “But this means that the interpreter’s own thoughts too have gone into re-awakening the text’s meaning. In this the interpreter’s own horizon is decisive, yet not as a personal standpoint that he maintains or enforces, but more as an opinion and a possibility that one brings into play and puts at risk, and that helps one truly to make one’s own what the text says. I have this described above as a ‘fusion of horizons’. We can now see that this is what takes place in conversation, in which something is expressed that is not only mine or my author’s, but common.” (Gadamer 2001: 388). See also Gadamer (2001: 384, 386, 397; and 1966: 65). 4. According to Habermas and Apel, ideological distortions in the self-understanding of a society are to be uncovered by moving to a critical theory of society, based on a model of communication in which all parties are aŸected to examine and discuss disputed claims on an equal basis and without fear of force or reprisal (Warnke 1987: 130).
Sharing… But why a language or world? 167
5. Here, Davidson claims that “translation is no part of the transaction between speaker and hearer that I call interpretation. Where translation of a sort may be involved is in the description the philosopher gives in his language of what the hearer makes of the speaker’s utterances.” (Davidson 1994: 4) 6. I disagree with JeŸ Malpas’s contention that Gadamer is not committed “to any strong claim concerning the necessity of a common language for the possibility of understanding” (Malpas 2002: 210). It is true that, according to Gadamer, understanding is possible across linguistic boundaries and that commonality is — as with Davidson — (at least partly) constituted by linguistic interaction itself, but Gadamer insists that a common linguistic practice is necessary for any understanding to be possible, whereas Davidson does not. 7. As Michael Dummett (1986: 469) suggests, it may not be quite accurate to describe Davidson’s later position as giving priority to idiolects, but Davidson does not dissent from this characterization, and it is a convenient way to bring out the flavor of his thinking. (See McDowell 2002: 192.) 8. There is no language-independent world. Rather, as Wilhelm von Humboldt said, whatever language we use, we never achieve “anything but an ever more extended aspect, a ‘view’ of the world” (ibid.). This does not mean, however, that by entering into foreign linguistic worlds, we discard our own prejudices (or pre-judgments, Vor-Urteile) and limitations. In a Hegelian vein, Gadamer says, “like travelers, we return home with new experiences” (Gadamer 2001: 448). Although one may contest this Hegelian idea, Gadamer’s contention that the world is not “given” is very important. 9. Davidson borrows many views on the signiªcance of innate similarity spaces from Quine (1969, 1974). For an excellent critique of this view, see van Brakel (2000: 81–99). Many “naturalist” philosophers go much further and make the connection between innate standards of similarity and natural kinds, or “carving nature at its joints”. Here are some examples (cited in van Brakel 2000: 89): “I am inclined to think that evolutionary considerations might oŸer some modest reassurance, for example, that our innate dispositions to classify certain things as of a kind by and large approximately pick out real kinds, which could support the idea that we have some minimal explanatory competence, on which, by revising and correcting our beliefs in the face of further experience, we can build; on which, indeed, we have built science” (Haack 1993: 220); “As a consequence of our genotypes and our early developmental environments, human beings come initially to categorize the world in a particular way, to view certain kinds of things as dependent on others, to generalize from single instances of especially salient types” (Kitcher 1993: 241); “For at least some perceptual judgments… the fact of widespread agreement does not require a social explanation. The explanation of evolutionary biology and physiology are su¹cient” (Giere 1988: 109); and “When our inductive inferences are guided by our intuitive grasp of the real kinds in nature, and when we project those properties which we intuitively recognize to be essential to those kinds, our… conceptual and inferential tendencies jointly conspire, at least roughly to carve nature at its joints and project the features of a kind which are essential to it… This pre-established harmony between the causal structure of the world and the conceptual and inferential structure of our minds produces reliable inductive inference.” (Kornblith 1993: 94)
168 Bart Vandenabeele
10. This was first published in the Proceedings and Addresses of the American Philosophical Association in 1974; cf. Davidson (2001).
References Davidson, Donald 1984 Inquiries into Truth and Interpretation. Oxford: Clarendon. 1986 “A nice derangement of epitaphs”. In Truth and Interpretation, E. LePore (ed.), 433–446. Oxford: Blackwell. 1989 “The conditions of thought”. Grazer Philosophische Studien 36: 193–200. 1992 “The second person”. Midwest Studies in Philosophy 17: 255–267. 1994 “The social aspect of language”. In The Philosophy of Michael Dummett, B. McGuinness and G. Oliveri (eds.), 1–16. Dordrecht: Kluwer. 1995 “Laws and causes”. Dialectica 49: 263–279. 1997 “Gadamer and Plato’s Philebus”. In The Philosophy of Hans-Georg Gadamer, L. E. Hahn (ed.), 431–432. [Library of the Living Philosophers 24.] 2001 Subjective, Intersubjective, Objective. Oxford: Clarendon. Derrida, Jacques 1991 “Letter to a Japanese friend”. In A Derrida Reader: Between the Blinds, P. Kamuf (ed.), 269–276. New York: Harvester Wheatsheaf. Dummett, Michael 1986 “A nice derangement of epitaphs: Some comments on Davidson and Hacking”. In Truth and Interpretation, E. LePore (ed.), 459–476. Oxford: Blackwell. Elgin, Catherine Z. 1997 Between the Absolute and the Arbitrary. Ithaca: Cornell University Press. Frege, Gottlob 1892 “Über Sinn und Bedeutung”. Zeitschrift für Philosophie und Philosophische Kritik 100: 25–50. Gadamer, Hans-Georg 1966 Philosophical Hermeneutics. Berkeley: University of California Press. 2001 Truth and Method. London: Sheed & Ward. Giere, Ronald N. 1988 Explaining Science: A Cognitive Approach. Chicago: University of Chicago Press. Haack, Susan 1993 Evidence and Inquiry. Oxford: Blackwell. Hacking, Ian 1986 “The parody of conversation”. In Truth and Interpretation, E. LePore (ed.), 447–458. Oxford: Blackwell. Hoy, David C. 1991 “Is hermeneutics ethnocentric?”. In The Interpretive Turn: Philosophy, Science, Culture, D. R. Hiley, J. F. Bohman and R. Shusterman (eds.), 155–175. Ithaca: Cornell University Press.
Sharing… But why a language or world? 169
Kitcher, Philip 1993 The Advancement of Science: Science without Legend, Objectivity without Illusion. Oxford: Oxford University Press. Kornblith, Hilary 1993 Inductive Inference and Its Natural Ground. Cambridge, MA: MIT Press. Lyotard, Jean-François 1984 The Postmodern Condition. Manchester: Manchester University Press. MacIntyre, Alasdair 1988 Whose Justice? Whose Rationality? Notre Dame: University of Notre Dame Press. McDowell, John 2002 “Gadamer and Davidson on understanding and relativism”. In Gadamer’s Century: Essays in Honor of Hans-Georg Gadamer, J. E. Malpas, U. Arnswald and J. Kertscher (eds.). Cambridge, MA: MIT Press. Merleau-Ponty, Maurice 1964 Le visible et l’invisible. Paris: Gallimard. Quine, Willard van Orman 1969 “Natural kinds”. In Ontological Relativity and Other Essays, 114–138. New York: Columbia University Press. 1974 The Roots of Reference. LaSalle: Open Court. Saunders, Barbara 2003 “Arendt’s challenge to multiculturalism”. In Whither Multiculturalism? A Politics of Dissensus, B. Saunders and D. Haljan (eds.), 131–143. Leuven: Leuven University Press. Taylor, Charles 1994 Multiculturalism. Princeton: Princeton University Press. Tully, James 1995 Strange Multiplicity: Constitutionalism in an Age of Diversity. Cambridge: Cambridge University Press. van Brakel, Jaap 2000 “Quine and innate similarity spaces”. In Quine: Naturalized Epistemology, Perceptual Knowledge and Ontology, L. Decock and L. Horsten (eds.), 81–99. Amsterdam: Rodopi. 2003 “Varieties of multiculturalism: No need for a shared language”. In Whither Multiculturalism? A Politics of Dissensus, B. Saunders and D. Haljan (eds.), 145–163. Leuven: Leuven University Press. Vandenabeele, Bart 2002 “Interculturele communicatie en incommensurabiliteit: ‘Het’ subject in eerste contacten tussen verschillende culturen”. In De terugkeer van het subject? Recente ontwikkelingen in de ªlosoªe, R. Devos, A. Braeckman and B. Verdonck (eds.), 203–211. Leuven: Universitaire Pers. 2003 “Lyotard, intercultural communication, and the quest for ‘real people’”. Language and Intercultural Communication 3: 20–35.
170 Bart Vandenabeele
Warnke, Georgia 1987 Gadamer: Hermeneutics, Tradition and Reason. Cambridge: Polity Press. Young, Iris M. 1997 “Feminism and the public sphere: Asymmetrical reciprocity”. Constellations 3: 340–363.
Identity as denial of diversity* Jef Verschueren University of Antwerp
1.
Identity in the service of diversity?
Annoyed by the emerging rhetoric which the political establishment, the mainstream media, and some well-intentioned social scientists were developing to counter — most unsuccessfully — the surge of extreme-right intolerance, Jan Blommaert and the present author decided in the early 1990s to address the question of the Flemish attitude to diversity.1 To that end we examined the publicly accessible discourse produced by what we called the “tolerant majority”, paying attention to patterns of implicit meaning (as re¶ected in wording strategies, presupposition- and implication-carrying constructions, patterns of interaction, and forms of argumentation) that might deviate from the openly professed explicit attitudes of openness and tolerance. The conclusion was that ingredients of an ideology we called “homogeneism” were running wild: a clearly persistent view of the ideal society as being as homogeneous as possible, re¶ected in a tendency to abnormalize the foreigner, to normalize negative reactions (xenophobia and racism) in the face of abnormal deviance from the norm of homogeneity, and pleas for a form of re-homogenization to be achieved by means of a demonstrably discriminatory and repressive notion of integration.2 We ended our analysis with the recommendation that the acceptance of true diversity should be the very starting point of any fruitful perspective on society. Little has changed since the period covered by Blommaert & Verschueren (1998). In Flemish political parlance there has been a shift away from “integration”, but only to replace the same notion terminologically. Every now and * This paper was first presented in Madrid, October 9–10, 2003, during the conference Identidad y diferencias / Identity and diversity: Reflexiones filológicas y filosóficas. I would like to thank the organizers, Beatriz Penas Ibáñez and Carmen López Sáenz, for their invitation.
172 Jef Verschueren
then we even hear politicians and other public ªgures voicing our own recommendation to take diversity as a starting point (without, of course, acknowledging the source). But at the same time we see mounting frustration among minority groups, based on the minorizing eŸects of mechanisms, ranging from the subtleties of paternalistic goodwill to rationally framed resistance to voting rights and a shamelessly open treatment of the myth of monolingualism as both a fact and a natural goal.3 Clearly, therefore, acceptance of the recommendation itself still allows for violations of the principles behind it. What I would like to explore in this contribution is the role that the notion of “identity” plays in this process. The reason for this focus is that identity is often invoked as the protector of diversity: of course we are open to diversity, because we respect everyone’s identity, just as we expect everyone to respect ours. Going beyond this individual stance, institutions devoted to the cause of diversity can also be observed to provide a lot of space for a rhetoric of identity. Thus we can ask: is identity rhetoric compatible with a demand to take diversity seriously as a starting point? I will investigate this question with reference to one speciªc case: a respected and highly respectable initiative, the so-called Metropolis Project. The initiative is Canadian-based, linked to Citizenship and Immigration Canada, a government service, but with branches spreading geographically through the eŸorts of Canadian embassies.4 If there is one thing that deªnes the project, it is its emphasis on diversity. Explicit attempts are made not to look at migration as a problem, but as an asset. Research is stimulated on both diversity and migration, and the results are popularized (through a newsletter called Metropolis, the Metropolis World Bulletin, and through sponsoring special issues of magazines such as Canadian Issues / Thèmes Canadiens, and Canadian DIVERSITÉY Canadienne [sic]). In this context, there is a lot of space for the study of identity and identity-related issues. My questions will then be: How does identity relate to diversity in the context of this Metropolis Project? Does identity discourse really serve the cause of full respect for diversity? Do researchers distinguish between identity construction processes observed as social processes, and their own use of the discourse as an analytic tool? To that end, I will be looking closely at a few examples, paying attention to the wording of research themes and observations, and asking questions about the implications.
Identity as denial of diversity 173
2.
Collective identities
More often than not, identity is “located” at the level of collectivities, whether ethnic or national. Let us ªrst have a look at an example that focuses on the ethnic dimension: (1) The place of Canadian ethno-cultural festivals within the policy and tradition of multiculturalism has often been overlooked or misunderstood by Canadian journalists, academics, and policy makers interested in the complex processes of integration and ethnic identity formation. Many scholars and policy makers deride festivals such as Winnipeg’s Folklorama and Toronto’s Caravan as “kitschy” and reductive expressions of an outdated form of multiculturalism. However, I would suggest that these events merit much more serious scholarly attention. […] In this project, I will demonstrate both the unique dimensions of Folklorama and Caravan, and the ways both festivals represent similar, complex, and contemporary approaches to Canadian diversity. (From the abstract of a research project by Paul Bramadat on “Ethnic self-representation in multicultural festivals”, http://pcerii.metropolis.net/Virtual%20Library/ Abstracts/2001/bramadat01.htm)
The author clearly situates his theme intertextually (in relation to Canadian journalists, academics, and policy makers) in a debate over “multiculturalism”. He implicitly sides with the rejection of a simplistic (outdated) form of multiculturalism which allegedly juxtaposes clearly separable and separate cultural groups — a stance that no longer ªts scholarly discourse about culture. But at the same time he also makes a plea for certain cultural expressions (ethnocultural festivals) as a contribution to ethnic identity formation in interaction with, or in the context of, complex processes of integration. Festivals thus represent positively evaluated (complex and contemporary) approaches to Canadian diversity. There is no doubt that the project for which the abstract is formulated relies on knowledge of the relevant background and that it reaches a high level of sophistication. What strikes us, though, is that “ethnic identity” is taken very seriously here. A question that goes unanswered, because it is unasked, is how (ethnic) identity formation, as a process, may actually reduce diversity rather than give it its proper place (since diversity is much deeper and wider than ethnic or ethno-cultural parameters allow for). We will come back to this same example later (Section 4).
174 Jef Verschueren
An example focusing on the national dimension of collective identity is to be found in the following extract from the announcement of a conference: (2) Two major approaches have tended to frame the debate around citizenship. The ªrst approach revolves around the belief that a strong immutable core — often deªned as national identity — is essential for the successful integration of newcomers. The other approach revolves around the belief that the core can, and indeed must, evolve by including citizens and newcomers in ongoing discussion on the construction of the national identity. (From the announcement of a conference — http://www. canada.metropolis.net/main_e.html — on Shared Citizenship: Immutable Core or Dynamic Nucleus?, Tuesday, October 14, 2003, Ottawa, Canada)
Without saying so explicitly, the structuring of this text implies the rejection of an immutable core — a conclusion that becomes even clearer from the (longer) French version of the text, which is still not explicit on the issue but leads more directly to the suggested inference, because of the negative wording of the implications of the belief in an immutable core (qui ne satisfont pas … ne peuvent obtenir … ne se voient pas …, as opposed to Cette approche garantit …): (3) Deux principales approches s’opposent dans ce débat. La première approche repose sur la croyance qu’un noyau immuable solide, qui est souvent déªni comme étant l’identité nationale, est essentiel à l’intégration réussie des nouveaux arrivants. En eŸet, cette approche soutient que l’intégration est possible uniquement lorsque les nouveaux arrivants et les membres de la société d’accueil comprennent et adhèrent aux divers éléments qui composent le noyau. Les personnes qui ne satisfont pas à cette exigence ne peuvent obtenir le statut de citoyen et ne se voient pas accorder les occasions voulues de participer à la vie de la société. La deuxième approche repose quant à elle sur la croyance que le noyau peut, et doit en fait, évoluer grâce à la participation des citoyens et des nouveaux arrivants à une discussion continue sur la construction de l’identité nationale. Cette approche garantit, selon ses partisans, le soutien des responsabilités, des valeurs et des objectifs communs. (From the French announcement of the conference La citoyenneté commune: Noyau immuable ou dynamique?) ‘Two principal approaches are opposed in this debate. The ªrst approach is based on the belief that a solid immutable core, which is often deªned as national identity, is essential to the successful integration of newcomers. In fact, this approach maintains that integration is only possible if the newcomers and the members of the host society understand and adhere to the
Identity as denial of diversity
various elements that constitute the core. People who do not satisfy this demand cannot obtain citizenship and are not oŸered the desired occasions to participate in the life of the society. The second approach is in turn based on the belief that the core can and must evolve, thanks to the participation of citizens and newcomers in a continuous debate on the construction of national identity. This approach guarantees, according to its advocates, the maintenance of common responsibilities, values, and objectives.’
Yet, the author seems to accept the existence of a core, even if this core of national identity may have to evolve as a result of the in¶uence of newcomers. In other words, this text does not ask (i) whether a core is not just a myth reducing the ever-lasting though changing reality of diŸerence, and (ii) whether at the “national” level, any talk of unity beyond “shared citizenship” is not an unwarranted obliteration of diversity. Thus, the subtitle of the conference leads attention away from the potential of the title, which would imply the full acceptance of diversity amongst citizens who are fundamentally “equal”.5
3.
Between collective and individual identities
The following example (the opening lines of the abstract of the corresponding article) illustrates perfectly how identity formation at a group level may reduce multidimensional individual diversity to a single dimension: (4) In Canada for almost thirty years, the Shi’a Imami Isma’ili Muslims of Calgary are remarkably diverse in terms of countries of origin, races, ethnicities and languages. At issue is the basis of collective identity, either ethnicity, languages or religion, as well as the role of pluralism and policy in the construction of a liberal democratic society. Set within a larger study on contexts of identity formation of immigration youth which explores relationships between languages, faith, settlement patterns, community cohesion and collective identity, interviews with parents and community leaders indicate that it is religion that holds central deªning value. (From “Collectivized identity among Shi’a Imami Isma’ili Muslims of Calgary: Implications for pluralism and policy”, by Rani Murji & Yvonne M. Hébert — http://international.metropolis.net/events/rome/rome.html)
This reduction of ethnic, national, racial, and linguistic diversity to religious commonality is the central message of the article in general:
175
176 Jef Verschueren
(5) In the eleven interviews that were conducted with the Isma’ili community leaders and parents of adolescents, it became very clear that the community was composed of many cultures and spoke many diŸerent languages. The identity of the community, however, rests on the ethos provided by its faith: “I mean deªnitely the one thing that binds us is our religion, and that is what initially brings us together…” states one Council Member also responsible for the Women Development Portfolio.
Yet, the following conclusion is drawn: (6) Given the signiªcance of human rights, multiculturalism as well as the critical role of community leadership, and learning from the experience of this collectivity, it is recommended that pluralistic conceptions of citizenship and policies continue to balance individual human rights and group membership. The preservation and enhancement of collective identities as well as a commitment to democratic representative government within a communitarian understanding of liberal democratic citizenship are key to the successful integration of groups such as the Isma’ili Muslims. Although the Isma’ili community is more extensively organized than other Muslim sects as well as many other polyethnic groups […], a communitarian approach to integration and citizenship is equally relevant. This means legitimizing diŸerent ways of belonging to a common polity; preserving essential deªning elements of personal and collective identities such as language, gender, ethnicity, religion and race; protecting the nature of communities without unduly burdening them; and allowing for diŸerential identiªcations and permeable, movable boundaries; while supporting collectivities and emphasizing collective identities; as well as enhancing collaboration between socio-educational institutions and ethnocultural community groups.
Carefully worded as this may be, the authors end up emphasizing the need for a balance between individual human rights and group membership, thus accepting without questioning a reduction of the consequences of having to accommodate full diversity. Unasked and unanswered questions are whether democracy must not allow space for diversity ªrst and foremost at the level of individual human rights, or whether, while all individuals have the right to form groups, the resulting “group rights” can ever be allowed to override individual rights without violating the principle of respecting diversity.
Identity as denial of diversity 177
4.
Individual identities
Though the project started at the ethnic group level, the author of the abstract in (1) ends up speculating about the issue of individual identity in his later research report: (7) […] that for most groups of people, and for most of human history, identity was ascribed, given by others (families, clans, tribes, religions, etc.). One might understand the current debate about identity as a function of the decline or deconstruction of traditional closed systems of ascribed identiªcation. The on-going modern (or post-modern) struggle for each person to win the right (though not exactly the unconditional freedom) to deªne his or her own identity as he or she sees ªt represents the backdrop of this whole discussion. (From “Shows, selves, and solidarity: Ethnic identity and cultural spectacles in Canada”, by Paul A. Bramadat — www.metropolis.net)
Here the “con¶ict” between individual and group-imposed identities is erroneously described as unique to this (modern, post-modern) place and time. When looking at the issue from the point of view of philosophy, we ªnd Cassam’s (1997) views about a quite universal awareness of the self as a physical object: one cannot describe human reality without claiming that “persons” exist. In other words, whatever the time and place ([post-]modernism having nothing to contribute in this respect), groups have no existence prior to their creation by persons. From the point of view of anthropology, we refer to Cohen’s (1994) observation that the very idea that the “self” would not be important for so-called “collectivity-oriented cultures” is no more than a creation of western anthropologists. Not only is the idea of identity often (partly) group-based and thus an instrument in reducing diversity at the inter-individual level, also when located at the level of the individual can the notion of identity serve as a reduction of true diversity. Consider the following quote: (8) According to Erikson, healthy identity development entails achieving a coherent self-deªnition, which is stable across time and place, well-regarded by signiªcant others, and a source of purposefulness and direction in deªning personal goals and values. The task of identity formation, or that of gaining a clear and coherent sense of knowing oneself, and what one will be in life, is regarded as a normative developmental process in¶uenced by personal and contextual factors. (From “Discrimination and identity: An overview of theoretical and empirical research”, by Violet Kaspar & Samuel Noh — www.metropolis.net)
178 Jef Verschueren
Whether or not this makes sense depends on how one interprets coherence. If it is taken at face value, the notion of individual identity involved here is a reduction of the multiplicity of identities to be found in all individuals. Not only can people identify with diŸerent groups, the resulting diŸerent identities are often in con¶ict with each other. Cohen (1994) gives the example of adolescents who feel perfectly comfortable either with their parents or with their friends, but who do not know how to behave when the two groups are mixed. Maalouf (2000) provides an elaborate example of his own personal history as a Christian and an Arab, as a Frenchman and a Lebanese, etc. DiŸerent agents can be involved in the reduction of multiple identities to a single dimension. Often this is done by others. Particularly instructive in this respect is the example of Jean Muteba Rahier (in press).6 This son of a Belgian colonialist in the then-Belgian Congo and a Congolese ménagère7 moved to Belgium with his father (and his mother, who returned to Africa after a short while, though), where he soon found himself in a boarding school as a métis, a ‘mulatto’ occasionally able to take pride in his association with some highranking black peers (e.g., the sons of African ambassadors) yet usually regarded as an outsider, both by the white and by the black kids, but in highly variable ways. When there was no racial peace in the school, he belonged to the black group; when there was, he did not clearly belong with the Africans (since, having been educated exclusively in French, he was not able to participate in conversations in Lingala or Kiswahili, languages mastered by some of the white pupils who had grown up in the colony). When moving to Ecuador to study aspects of the African diaspora, he found a local mestizo culture that valued Europeanized Indians, but that had no space for blacks. The blacks he was studying regarded him as a real African (a view he did not have of himself), since he was born in Africa and had a mother still living there. The white or mestizo Ecuadorians, on the other hand, did not hesitate denigrating blacks in his presence, as they did not associate him at all with the local black communities, regarding him as a negro ªno ‘reªned black’. When moving to the U. S. he underwent yet another process of (re)identiªcation. At his ªrst barbecue he was approached by a black woman who asked him about his background. He concluded his life history with the statement that he obviously was a mulatto. The woman looked at him reproachfully and said: “Do you have a problem with your blackness?” The American one-drop rule had not yet penetrated Jean’s consciousness.
Identity as denial of diversity 179
5.
Identity: Diversity denied
Already in 1959, GoŸman oŸered a description of the social-psychological dynamics that are involved when groups form identities: When two teams present themselves to each other for purposes of interaction, the members of each team tend to maintain the line that they are what they claim to be; they tend to stay in character. Backstage familiarity is suppressed lest the interplay of poses collapse and all the participants ªnd themselves on the same team, as it were, with no one left to play to. […] At the same time, each team tends to suppress its candid view of itself and of the other team, projecting a conception of self and a conception of other that is relatively acceptable to the other. And to ensure that communication will follow established, narrow channels, each team is prepared to assist the other team, tacitly and tactfully, in maintaining the impression it is attempting to foster. (GoŸman 1959: 167)
GoŸman’s description bears speciªcally on institutional contexts, but the processes can be generalized, though the high level of cooperation is often lacking in other environments. Clearly, the processes GoŸman describes boil down to a reduction or even a denial of diversity. Analogous processes take place at a wider societal level. The question we are left with is why such processes are so prevalent. Habermas was no doubt right when claiming that a politics of redistribution needs to be supplemented with a politics of recognition. But in this context, we should heed Bauman’s (2001) pertinent observation that a politics of recognition has come to replace a policy of redistribution, the latter requiring true social justice, with full equality for all individuals. In his own words: The two developments — the collapse of collective redistribution claims (and more generally, the replacement of the criteria of social justice by those of respect for diŸerence reduced to cultural distinction) and the growth of inequality running wild — are intimately related. There is nothing incidental about this coincidence. Setting claims for recognition free from their redistributive content allows the growing supply of individual anxiety and fear generated by the precariousness of ‘liquid modern’ life to be channelled away from the political area — the sole territory where it could crystallize into redemptive action and could therefore be dealt with radically — by blocking its social sources. (Bauman 2001: 88; emphases in the original)
Recognizing groups, and devoting overpowering attention to the symbolization of this recognition, may be the political way out that will perpetuate inequality. It meets political support because it is, relatively speaking, cheap indeed. And it meets popular support because the rhetoric invokes comforting unity, culture, and community.
180 Jef Verschueren
The descriptions we have looked at do not just observe. They are not neutral. They may run the risk of incorporating the processes we have hinted at into their own perspective. Academics should carefully steer away from such tendencies. In the interest of respect for diversity, a critical questioning of identity should, therefore, be part of all investigations into processes of identity formation.
Notes 1. The research was originally reported in two books written in Dutch; an account in English is to be found in Blommaert & Verschueren (1998). 2. Note that there is nothing wrong with the notion of integration as such: in opposition to segregation (as in the North-American context), it is truly emancipatory. 3. Recent surveys (e.g., De Houwer 2003) show that in one out of six families in Flanders, another language is used than or in addition to (Flemish) Dutch. Yet school regulations and educational practice (as shown by Jaspers 2003) treat even the linguistically most mixed populations as if they were monolingual. Similarly, Flemish politicians claim the right — supported in this by the mainstream media — to ignore francophone population groups in o¹cial Flemish territory (see Lenaerts 2003). 4. The last two international Metropolis conferences were held in Oslo (September 9–13, 2002) and Vienna (September 15–19, 2003), respectively. 5. In this context it is useful to be reminded of the diŸerences between “civic” and “ethnic” nationalist discourses, as applied to Spain in Grad Fuchsel & Martín Rojo (2003). 6. For other examples, see Verschueren (2002). 7. The ménagère was a local woman who, with the consent of her family, went to live with a white Belgian colonizer, providing a variety of services, including sexual favors. When they became pregnant, they were usually sent back to their village. In the case of Jean Muteba Rahier, the father recognized his children and took care of them; he also took them with him when moving back to Belgium after the colony’s independence in 1960.
References Bauman, Zygmunt 2001 Commmunity. Cambridge: Polity. Blommaert, Jan and Verschueren, Jef 1998 Debating Diversity: Analysing the Discourse of Tolerance. New York: Routledge. Cassam, Quassim 1997 Self and World. Oxford: Oxford University Press.
Identity as denial of diversity
Cohen, Anthony P. 1994 Self Consciousness: An Alternative Anthropology of Identity. New York: Routledge. De Houwer, Annick 2003 “Home languages spoken in o¹cially monolingual Flanders: A survey”. In Con¶ict Linguistics, K. Bochmann, P. Nelde and W. Wölck (eds.), 71–87. St. Augustin: Asgard. GoŸman, Erving 1959 The Presentation of Self in Everyday Life. New York: Anchor Books. Grad Fuchsel, Hector and Martín Rojo, Luisa 2003 “‘Civic’ and ‘ethnic’ nationalist discourses in Spanish parliamentary debates.” Journal of Language and Politics 2: 31–70. Jaspers, Jürgen 2003 “Talige stileringen als identiteitsconstructie bij Marokkaanse jongens”. Handelingen van de Koninklijke Zuid-Nederlandse Maatschappij voor Taal- en Letterkunde en Geschiedenis 56: 151–171. Lenaerts, Gilberte 2003 The Columberg Report in the Flemish Media: Discourse and Identity Maintenance. University of Antwerp: Ph.D. dissertation. Maalouf, Amin 2000 On Identity. London: Harvill Press. Rahier, Jean Muteba in press “Métis/Mulâtre, Mulato, Mulatto, Negro, Moreno, Mundele Kaki, Black, …: The wanderings and meanderings of identities”. In Problematizing Blackness: Self-Ethnographies by Black Immigrants to the United States, P. C. Hintzen and J. M. Rahier (eds.). New York: Routledge. Verschueren, Jef 2002 “Pragmatic aspects of culture and cognition: Interactional and conceptual dimensions of diŸerence and equality”. In Cognitive Linguistics Today, B. Lewandowska-Tomaszczyk and K. Turewicz (eds.), 277–286. Frankfurt am Main: Lang.
181
Conceptual innovation in art Three case studies on Leonardo da Vinci, William Turner, and Henry Moore Wolfgang Wildgen University of Bremen
1.
Semiotic innovation in the art of Leonardo da Vinci
In his Trattato della pittura (‘Treatise on painting’), Leonardo states that the painter has as his primary aim the representation of two things: man and his mind (“l’uomo e la mente”; Pedretti 1995: §180). The nature of man becomes visible and, therefore, accessible to the eye in the diŸerent “accidents” and movements and in the proportions of his body parts (cf. Pedretti 1995: Terza parte). In order to represent man and his mind the artist must ªrst create a pictorial space, the stage for the topic of the painting. Secondly he must consider light and shadow in human bodies, the gestures of the hands, the postures of the head, and the facial expression, before he distributes the topics of the painting on the surface. Finally landscape, sky, objects, animals, and persons accompanying the topic of the painting (mostly individuals or groups of individuals) must be arranged in space, relative to light and shadow. Leonardo’s art goes beyond the mimesis of nature; it divulges the internal meaning of space, of light and shadows, to the viewer. The beauty of a scene is its recovered meaning in the mind of the viewer. With the foregrounding of movement and accident Leonardo requires interpretive activity of the viewer, who, based on his experience, must extrapolate movement and action from an instantaneous picture to a process which has caused it and which will bring it to a proper end later. In bodily movement, any movement of one body part has a counterpoise in another. The terms balance, weight, and counterpoise have the following meaning in the context of painting: A single body is in balance if the weight of the movement of one
184 Wolfgang Wildgen
part of the body, e.g., the head, has a counterpoise in another, e.g., in the movement of the shoulders or the trunk. The balance could be easily realized if all bodies were static. But this would make them “wooden”, i.e., unanimated. The painter who wants to show the mind of the persons in the scene must show them in movement, and the balance of a person or group of persons has to be a dynamic balance. In the following I shall analyze these semiotic aspects with reference to two groups of Leonardo’s paintings: The Last Supper and St. Anne with the virgin and the child. The painting of the Last Supper produces the illusion of a deep hall, with rectangular tapestry on the sides and three openings going into a landscape in the background. The linear perspective has its origin in the head (the right ear) of the Christ. The central point, Jesus’ ear, is signiªcantly related to the narrative content. Jesus has just uttered the prophecy that one of his disciples will betray him, and in the tense moment captured by the painting he is listening to their answers.1 His head is slightly displaced from the center in relation to the open door in the background, which puts his face into relief against the landscape. The upper border of the landscape deªnes a line on which the eyes of Christ are placed. Thus, the position of the central person in space is exactly calculated in order to match its geometrical and semantic centrality. The geometrical arrangement of the thirteen actors in the scene has a basic symmetry. Christ versus twelve apostles, six of them are sitting to his left, six to his right. This linear arrangement, which includes Judas, the traitor, is new. In most paintings which treated the same topic before, Judas is sitting on the other side of the table and turns his back to the viewer.
Figure 1. Leonardo da Vinci, The Last Supper (whole painting).
Conceptual innovation in art 185
In one of the preliminary sketches with John resting his head on the table, the grouping of the remaining apostles on Christ’s left equals two plus three. At his right it is one (Peter) plus two plus two. Leonardo restructures this scene to obtain geometrical symmetry, which may be expressed arithmetically: 13 = 3 + 3 + 1(Christ) + 3 + 3. In comparison with this order, we may observe that Giotto (1266–1336) painted his Last Supper with a central, nonhomogenous group of three: Christ, John, and Judas on the opposite side of the table, plus two groups of ªve apostles. The arithmetic order is: 13 = 5 + 3 + 5. Ghirlandajo (1449–1494) also adopts the same central group of three persons, but he further subdivides the ªve persons to their left. Leonardo’s proportions are new in relation to the tradition and they deªne a new cognitive or image-schematic model, which is able to restructure the reading of the biblical story. In an indirect way relevant for the broader public, the story of the Last Supper is a prototype for the interpretation of society and of relations in a human group (a family, a group of friends, a group of professionals, a political party with its leader, etc.). Therefore, the innovative organization of the topic in the painting creates a new frame of interpretation not only applicable to spiritual life, but also to political and everyday aŸairs. In this sense, it has the role of a discourse orientation and is able to serve as background knowledge for innumerable situations of communication. This leads to the human and social relevance of a piece of art, beyond its esthetic function. Such a semiotic view on art is, however, still in its infancy and I cannot elaborate on it in the present context. The narrative function of the painting is expressed by its dynamics. The utterance of Christ, “One of you will betray me”, constitutes a force, whose eŸect makes a visible and emotional impact on the apostles. Like a shockwave it most strongly hits the two groups sitting directly to the right and left of Christ, and to a lesser degree the exterior groups. If we consider the nearer groups, James is pushed back, whereas John, although displaced in relation to Christ, stays calm, and Judas seems to freeze in the moment as he is grasping at the bread. These two groups are more agitated than the calmer outer groups. Thus, the dynamic eŸect of the words of Christ is represented as a wave with repercussions and vortices. This is also true for the single groups of apostles. If we analyze their postures and gestures, we can further decompose all four groups of apostles into two plus one (center). The central person neutralizes, stops the movement issued from Jesus and thus brings it to rest, the natural locus of movement in Aristotle’s physics. It is as if the blow of the utterance dynamically shaped the four groups and their subgroups. As a result of this impact, the grouping could be a natural consequence of the underlying dynamics. Dynamically, Judas is clearly separated
186 Wolfgang Wildgen
from the other apostles, as he seems to be lost for any positive eŸect, holds his money in his right hand, and stops grasping for a moment at the bread. He shows a closed, sinister face, which is in full contrast to the face of John illuminated by the light from the left. In relation to the geometry of the painting, we can say that Leonardo tries to organize his composition as an instant in a process which shows the origin of the force, its immediate eŸects, and the multiple structures created by the percussions of the force, which is located in, and stems from, Jesus. As the emotional and intellectual eŸects of the central force are the main topic of the painting, Leonardo reorganizes the geometry of the scene, in order to arrive at an optimal representation of the percussions in body postures, gestures, and facial expressions. The semiotic revolution in Leonardo’s painting preªgures the scientiªc revolution announced in Copernicus’ De revolutionibus orbium (1543) and brought to the summit in the work of Galileo and Kepler. Together with other intellectual and religious changes they remodel common knowledge, the worldview of educated people in the sixteenth century and of Western civilization in the centuries to follow. Our second example also comes from the work of Leonardo, but it lies closer to linguistic semantics, mainly sentence frames and verbal valence: the thematic composition in Leonardo’s painting of St. Anne. In the painting of Leonardo three people (of diŸerent ages and sexes), St. Anne, Mary, and Jesus, are easily identiªable as the major topics. From the Bible we know that there is a kinship relation: Anne — mother of — Mary — mother of — Jesus (by transitivity we know that Anne is the grandmother of Jesus, by inversion that Jesus is the son of Mary and Mary the daughter of Anne). This knowledge level already constitutes a second space related to the (ªrst) visual space that we construe immediately as we look at the painting: Anne – background to – Mary – background to – Jesus Anne’s head – above – Mary’s head – above – Jesus’ head Anne – looks older than – Mary – looks older than – Jesus Anne – supports – Mary – supports – Jesus Anne and Mary form one group (by proximity), Jesus and the lamb another.2 The painting contains a rich geometric and dynamic structure (weights, bar centers, force lines, directions of gaze, etc.) which is used in many of Leonardo’s works. A purely static representation would be insu¹cient for both the pictorial and the narrative aims of the painting. Furthermore, this piece is typical of Leonardo’s art, which consistently exempliªes the concept of dynamic valence.3
Conceptual innovation in art 187
Figure 2. Leonardo da Vinci, St. Anne with Mary, Jesus, and the Lamb (Paris).
188 Wolfgang Wildgen
In this painting, we have on the surface a quaternary constellation: Anne – Mary – Jesus – lamb. If one considers the relevant force ªelds and actions, one notices that a basic interaction links Mary – Jesus – the lamb: – – – –
Mary pulls at Jesus Jesus pulls at the lamb the lamb resists Jesus resists being pulled away from the lamb
There is a con¶ict between Mary, who tries to prevent Jesus from seizing the lamb, and Jesus, who notices this (he looks back at her) but resists her action. This triad constitutes a force ªeld, which dominates the message of the painting. A ªrst schematic representation introduces two vector ªelds with attractors: Mary
Jesus
lamb
Jesus is in the meta-stable position between two attractors; the narrative (biblical) content of these attractors is: Mary: Jesus:
His mother; she cares for her baby. He feels the duty to sacriªce himself and leave Mary behind.
The cognitive dynamics lying at the heart of the two paintings by Leonardo may be described in the context of dynamical semiotics (cf. Wildgen 1994: 68– 72). The constellation of forces between Mary – Christ – the lamb corresponds to the basic archetype of transfer in Figure 3. Mary
main attractor 1
transient attractor
main attractor 2
Figure 3. The dynamical archetype of transfer (giving) and a ªber on it (with attributed contents).
Jesus
lamb
Conceptual innovation in art 189
As the archetype does not describe all the interactions in the pictorial composition, one has to add two complications: – –
Anne supports/anchors the whole event (physically and genealogically); she is a fourth attractor which sustains the event happening on her lap. The manner of “transfer” is further elaborated in the painting and could be described in a sentence like: Mary tries to prevent Jesus from seizing the lamb. This complex sentence goes beyond the elementary schema shown in Figure 3.
In the painting one sees Mary’s hands seizing Jesus and the hands (and feet) of Jesus seizing the lamb, and we also see that Jesus has a stronger grip on the lamb than Mary does on him. The turning of his head creates an opposition to the force direction of Mary’s hands. One could further distinguish two levels of content interpretation: – –
an interpretation based on everyday experience with human interaction; an interpretation linked to the Bible.
This distinction of two levels of analysis is natural, because in real-life situations, as when one observes a mother interacting with her child, the ªrst interpretation/blend occurs. In cultural traditions that have elaborated a collective visual or linguistic memory (e.g., of major biblical contents), the second interpretation is superimposed on the routines of the ªrst. It is typical of art to presuppose cultural knowledge and some such special “reading“ skills.
2.
The semiotics of landscape painting (William Turner)
Natural scenes and even actors may fade from sight, e.g., a landscape in fog or in the dusk, a person in similar conditions, or if the viewer is only allowed a short glimpse, as when passing by on a train. Reduced visual cues for an object or person portrayed in such a way are therefore not necessarily abstract but may be real. Another example of abstract perception occurring in real vision may be found in cases in which people half-perceive or only implicitly perceive objects which appear in a repetitive or monotonous context, such as the telephone poles at the side of the road on a long car trip, the contents of a room which one passes through everyday, and so on. Although people have the physical ability to perceive the whole situation in a precise manner, they largely ignore the multitude of details and pick up one or two speciªc features. In such situations
190 Wolfgang Wildgen
our memory is doing something one may call “abstraction”. In the same way, the artist may prefer to represent the mental (mnemonic) construal of a scene out of its selected features and neglect all foreseeable and general features. I shall analyze this process in a number of paintings by William Turner. William Turner (1775–1851) stands in the tradition of British landscape painters. Starting in 1729 there was a tradition among aristocratic families to send their young men on a “Grand Tour”, visiting Europe and especially Italy. Paintings of scenes encountered on the tour, mostly in the dominant Italian style, were brought home as souvenirs. In the course of the 18th century painters like Wilson, Gainsborough, Wright of Derby, Cotman, and Gertin, a. o., departed from this tradition, and Turner radicalized the move against topographical preciseness (cf. Meyer 1993). The loss of referential distinctiveness, a neglect of the identiªcation of speciªc objects, cities, and landscapes, may have been motivated by Turner’s preference for vague surfaces under speciªc (natural) conditions and by the ease with which he could work very long with his materials in the open, i.e., he preferred a more economic or even minimalist method. In a painting of the Castle of Chillon at the lake of Geneva from 1809 (British Museum), a precise representation with persons in the foreground, buildings, the lakeshore, and mountains is given; in a watercolor painting from 1841 the same lake of Geneva in depicted in a very vague fashion.
Figure 4. Geneva, The Mole, the Lake and the Savoy Hills, 1841 (cf. Wilton 1982: 63, Nr. 89).
Conceptual innovation in art
In the second painting (Figure 4) one can still recognize mountains, the shore of the lake, the water surface (blue), ships, and possibly people, but the symphony of colors, the transitions between surfaces, and, indirectly, the emotional values become dominant. Some of the later paintings re¶ect very speciªc, rare, and traditionally not represented phenomena like the “Snow storm — Steam boat oŸ a harbor’s mouth making signals in shallow water, and going by the lead.” Turner gave this precise description to avoid an interpretation of the picture at issue as a fantasy or caprice. In a certain sense, some of these paintings are even more realistic than those by Leonardo because they refer to concrete, personal perceptions of the painter. Their objects refer to geographically and historically precise entities which can be identiªed as parts of the context of the painter’s life, episodes of his journey, situations observed and remembered by the painter himself. This subjectivity, which refers to the life, the body of the author, is probably the new message which made Turner a precursor of the impressionistic style in the eyes of later generations. The next section will follow up on this line, which became dominant in the artistic schools of expressionism and cubism. The work of Picasso could be a good example, but I choose the sculptures (and sketches for sculptures) of Henry Moore.
3. Abstract analysis of human body postures in Henry Moore’s sculptures In order to restrict the scope of my analysis, I shall only consider the topic of the “Reclining Figure”, which is frequent in Moore’s oeuvre. In a crayon drawing (Reclining Nude, red and black crayon, ca. 1923; cf. Moore 1989: 96) Moore shows a realistic picture of such a reclining nude; see Figure 5. Another drawing from 1933 shows a series of projected sculptures on the same topic. Even if single parts of the sculptures may be recognized as belonging to a human body in a reclining position, these ªgures are like an exercise for drawing variable shapes of persons in the given pose. What is left is the horizontality, the partial suspension (and support) of the body typical of a reclining pose (cf. Moore 1989: 112).
191
192 Wolfgang Wildgen
Figure 5. Henry Moore, Reclining Nude, ca. 1923, red and black crayon (cf. Moore 1989: 96).
Figure 6. Henry Moore, Reclining Figure, Hand, 1979, bronze (cf. Moore 1989: 265).
Henry Moore became famous for his large sculptures, many of which resemble human bodies but some of which lack any referential support. Figure 6 presents one of many variants of Henry Moore’s treatment of this same topic in sculptures.
Conceptual innovation in art 193
The head is a mass in the normal position (above the shoulders), and the direction in which the head is facing (i.e., front, sides, back of the head) is distinguishable. Everything else, hair, ears, eyes, mouth, is absent. The sharp frontal line on this “head” could be a generalization of the nose. The big extremities (feet) may be associated with the shape of covered knees in a reclining ªgure, and the large mass at the end could be a long cloth covering the legs, which are open (cf. the drawings from 1948 in Moore’s Sketchbook; Moore 1989: 151). The result of this “abstraction” can be explained by a series of transformations, which are visually reconstructed by the viewer as soon as he knows the topic of the sculpture. What is the prototype of a draped lying woman? The simplest scheme is one involving two masses: mass 1: (head, trunk, arms) – mass 2: (legs) – transitory zone: back The transition zone is the most di¹cult part for this composition. It can be a sphere (maximally curved) — a bent cylinder (moderately curved) — or a regular solid, e.g., a pyramid, a cube, etc. All these solutions are at least implied in Moore’s sculptures. Figure 7 gives an idea of possible compositions. The narrative content of a “Reclining Figure” is concentrated in the verb: to lie, with its lexical neighbors lie down, sit down, and stand up, rise. The viewer of the sculpture may mentally ªll in masses for the head, the shoulders, the arms, the elbows, the knees, each leg, feet, hair, eyes, nose, mouth, etc. He thus goes back on a path of abstraction analogous to that taken by the sculptor himself (or even engages in a longer “story” of artistic development). This story may be understood as the trajectory from a prototype (or a set of prototypes) to our visual representation of a particular human body. The sculpture is thus a representation of this cognitive itinerary (cf. Leyton 1992).
Figure 7. Reducing the “Reclining ªgure” to cylinder, sphere, and (curved) pyramid.
194 Wolfgang Wildgen
A more radical visual analysis of the human body destroys its unitary gestalt, its topological coherence. Two particular trajectories may be observed in Moore’s œuvre: –
–
–
Instead of one body (from head to feet and hands), several sculptural entities are used which together “represent” the “Reclining Figure”. Moore distinguishes two-piece, three-piece, and four-piece reclining ªgures.4 The geometrical analysis shown in Figure 7 is not su¹cient to cover all the variants of the topic; but if one adds curved (and conical) cylinders, spheres with excavations, and bent cylinders, most of the forms may be classiªed on this basis. The bending of lines, the curves, and the concave or convex surfaces refer more speciªcally to the realm of living, biological entities. Beyond convex and concave limbs and surfaces, one can observe gaps. They may appear in a reclining ªgure, e.g., between arm and body, or between arms and legs and the support. These gaps may show up naturally.5 In other sculptures the title abandons any reference to human bodies, as in Three Rings, 1966–67 (Moore 1968: Nr. 74).
Instead of continuous (diŸerentiable) curves, there appear sharp edges (“knifeedge”), pikes, and strings attached to bodies. The combination of smooth surfaces, knife-edges, and pikes is illustrated in the sculpture Three Points 1939– 40, shown in Figure 8.
Figure 8. Henry Moore, Three Points 1939–40; cf. Moore (1968: 45, Nr. 37).
Conceptual innovation in art 195
4.
Conclusion
Innovations in art are more than temporary changes in taste or dominance shifts in the rivalry of schools. They contain at their heart a “problem” and its “solution” by the artist. This has been shown in the case of Leonardo, where the arrangement and the postures in the Last Supper contribute to the visualization of the dynamics inherent in the narrative episode represented in the fresco. In the series of works dedicated to the triad St. Anne, Mary, and Jesus, a dynamic conªguration with attractors, repellors, forces, weights, and counterpoises is visually organized together with a system of sight vectors. Case studies of Turner and Moore showed the natural process of symbolic abstraction in landscape painting (Turner) and ªgural sculpture (Moore), as well as the directions of their gradual development in the oeuvre of the artists.6 Visual semiotics should in its future development concentrate on these and similar “problems” of symbolic representation, and the ways of their “solution”. The treatment of dynamics seems to lie at the heart of such an endeavor.
Notes 1. Luca Pacioli, a friend of Leonardo’s, offered the interpretation that Christ has just said: “Unus vostrum me traditurus est” ‘One of you will betray me’, and that Christ in this moment accepts the beginning of his martyrdom. 2. The knowledge space is richer, as the colors help to complete the visual map. In Leonardo’s painting both mother and daughter have a green mantle and Mary can be seen to wear a red robe; both are subsumed under the color opposition red – green. The lamb is a symbol of Jesus’ sacriªce and an attribute of John (the Baptist). 3. Peirce (e.g., 1986) was the ªrst to propose a general scheme at this level of abstraction. His monads, dyads, and triads are considered as dynamical wholes (in Leibniz’ sense) not reducible by simple composition. However, they may be and often are degenerated, i.e., not complete, nonsaturated, with defects. 4. In Moore (1968), examples are shown in Nr. 10 (one piece), Nr. 11 (two piece), Nr. 40/41 (three piece); the four-piece composition (Nr. 86) is not called Reclining Figure but visibly belongs to the same family of sculptures. 5. Cf. Moore (1968: Nr. 68, Nr. 71 Reclining Figure 1939, Nr. 73 Reclining Figure (external form) 1953–54) 6. Further semiotic aspects of the art of Leonardo da Vinci have been treated in Wildgen (2004a; directions of gaze and gestures); the iconography of the Last Supper was the topic of a chapter in a book on the semiotics of art (in French; Wildgen 2004b). Innovation in art has been analyzed in the context of innovation in language and in science as a chapter of my book on the evolution of human language (Wildgen 2004c).
196 Wolfgang Wildgen
References Leyton, Michael 1992 Symmetry, Causality, Mind. Cambridge, MA: MIT Press. Meyer, Laure 1993 Englische Landschaftsmalerei von der Renaissance bis heute. [Translation from French.] Paris: Terrail. Moore, Henry 1968 Catalogue of an Exhibition at the Tate Gallery. [Edited by David Sylvester.] London: Humphries & Co. Moore, Henry 1969 Catalogue of an Exhibition at the Fondation Pierre Gianadda, Martigny. [Edited by David Michinson.] Milan: Electa. Pedretti, Carlo (ed.) 1995 Leonardo da Vinci: Libro di pittura. Codice Urbinate lat. 1270 nella Biblioteca Apostolica Vaticana (trascrizione critica di Carlo Vecce). Firenze: Giunti. Peirce, Charles Sanders 1986 Semiotische Schriften, vol. 1. Frankfurt/Main: Suhrkamp. Wildgen, Wolfgang 1994 Process, Image, and Meaning. Amsterdam: Benjamins. 2004a “Die Darstellung von Hand (Gestik) und Auge (Blick) in einigen Werken von Leonardo da Vinci”. In Bodies–Embodiment–Disembodiment, W. Nöth and G. Ipsen (eds.). Kassel: Kassel U.P. CD-Rom. 2004b “Éléments narratifs et argumentatifs dans l’articulation de l’espace pictural: Transformations de l’Ultime Cène du XIIe au XXe siècle”. In Espaces perçus, territoires imagés en art, S. Caliandro (ed.), 77–97. Paris: L’Harmattan. 2004c The Evolution of Human Language: Scenarios, Principles, and Cultural Dynamics. Amsterdam: Benjamins. Wilton, Andrew 1982 Turner Abroad. London: British Museum Publications.
Index
A
action 1, 3, 6, 7, 9, 10, 26, 38, 67, 75, 77, 78, 8082, 86, 97, 98, 107, 119, 144146, 179, 183, 188 acte 14, 17, 21, 22, 76, 78, 80, 86, 100102, 105, 106, 109 activité 77, 78, 80, 81 agentivity 5, 62 agreement 7, 16, 125, 153, 154, 156, 160, 167 allotemporal 30, 31 ambiguïté 22, 103106, 108111 analyse du discours 76, 78, 80 Anatolian 90, 95 animacy hierarchy 61, 62, 68, 70 animal 98100, 118, 120123, 125, 127, 140, 160, 161, 183 antiphrase 100, 106, 107, 111 Armenian 90, 91, 94 art 13, 115, 183, 185, 186, 189, 195, 196 artist, artistic 183, 190193, 195 assertive 44, 5456, 63
B
behavior 1, 2, 3, 6, 9, 10, 14, 15, 38, 67, 70, 86, 117, 118, 120122, 127, 129131, 133, 134, 136, 139141, 143, 147, 148, 155, 162, 163 Belgium 178, 180 body 66, 67, 143, 146, 147, 183, 184, 186, 191194, 196
C
Cartesianism 131, 133 category 15, 25, 31, 36, 38, 44, 5256, 63, 70, 73, 75, 77, 8791, 90, 93, 94, 118,
119, 123, 144, 145, 162 cognitive 3, 7, 10, 16, 26, 2831, 37, 38, 40, 41, 69, 70, 72, 73, 95, 132, 139, 149, 161, 168, 181, 185, 188, 193 cognitive grammar 31, 40, 73 see also grammar cognitive linguistics 72, 73, 181 cognizer 68, 69 commissive 54, 56, 57, 119, 124, 125 common language 143, 152156, 165167 community 3, 4, 6, 11, 12, 14, 15, 115, 118, 126, 153155, 157, 159, 160, 162-165, 175, 176, 178, 179 see also linguistic community, speech community computer simulations 141, 142, 144, 145, 147, 148 conceptualizer 33, 36 conditional 89, 92, 95 conjunctive 88, 94 constative 3, 4, 43, 44, 4648, 119, 124, 125 culture, cultural 25, 62, 7073, 117, 126, 127, 160, 168, 173, 176, 177, 178, 179, 181, 189, 196
D
declaration, declarational, declarative 43, 44, 5058 deictic, deixis 20, 27, 28, 65 directive 52, 53, 55, 57, 119, 124 discours 14, 18, 21, 22, 7578, 80, 8386, 107110, 112 see also analyse du discours discourse 2, 4, 6, 7, 11, 12, 14, 15, 1820, 22, 26, 27, 29, 40, 65, 69, 71, 72, 86, 96, 117, 140, 171173, 180, 181, 185
198
Index
diversity 117, 118, 160, 169, 171173, 175 177, 179, 180
E
emotion, emotional 7, 9, 185, 186, 191 English 19, 2731, 3841, 57, 6467, 70 72, 88, 89, 96, 145, 180 ethnic, ethnicity 173, 175176, 177, 180, 181 evolution 41, 69, 95, 106, 117, 118, 120, 124127, 148, 149, 195, 196 expression, expressive 1113, 16, 25, 31, 32, 34, 35, 38, 43, 49, 53, 56, 57, 69, 72, 78, 90, 97, 98, 103, 104, 106109, 111 113, 119, 127, 129137, 152, 159, 173, 183, 186
F
ªrst person 6266, 68, 70, 91 français 86, 98, 111 French 16, 20, 28, 30, 65, 125, 174, 178, 195, 196 fusion of horizons 153, 154, 159, 161, 165, 166 see also horizon
G
Genève 75, 77 see also modèle genevois Georgian 92, 95 Germanic 90, 91, 96 grammar 13, 20, 26, 31, 3840, 72, 73, 87, 89, 95, 96, 131, 132, 137, 155 see also cognitive grammar grammaticalization 33, 9395 Greek 8891, 93, 96, 97
H
Hindi 65, 92, 9496 horizon 153155, 159, 160, 161, 165, 166 see also fusion of horizons
I
identities 131, 165, 171181
ideology 10, 171 idiolect, idiolectical 158, 159, 165 illocutionary 3, 4, 7, 15, 16, 19, 4446, 49 58, 117119, 121, 123, 127, 136138 see also locutionary illocutionary act 3, 7, 16, 44, 45, 49, 50, 51 56, 58, 118, 119, 127, 137, 138 see also locutionary act imperative 14, 49, 8992, 96 implicit 6, 32, 44, 50, 51, 53, 57, 71, 171 indicative 32, 49, 63, 89, 129, 132, 135, 136 indirect speech acts 6, 18, 52, 53, 58 see also speech act Indo-European 73, 90, 91, 9496 see also Proto-Indo-European intention 58, 11, 21, 45, 46, 56, 98, 99 103, 105, 110112, 131, 137, 162, 163, 165 intercultural communication 152, 169 interpretation 5, 1012, 17, 27, 29, 32, 34, 38, 72, 109, 111, 124, 134, 142, 151159, 161, 163, 165, 167, 168, 185, 189, 191, 195 introspectionist 131, 134 ironie, ironique 97113 irony 5, 112, 113 irrealis 63, 87, 9295 see also realis
J
Japanese 6373, 168
L
language game 129131, 134, 135, 137, 139, 141, 142, 147149 language use 3, 6, 9, 148 Latin 8791, 93, 95, 96 linguistic 2, 3, 6, 9, 11, 13, 16, 17, 20, 21, 25, 26, 30, 34, 35, 40, 51, 52, 55, 58, 61, 6366, 68, 69, 7173, 89, 91, 9296, 126, 129132, 134136, 139141, 143, 144, 146149, 154160, 162165, 167, 175, 186, 189 linguistic community 157, 159, 160, 164,
Index
159, 162, 165 see also community, speech community linguistic form 135, 139141, 143, 146149 litote 103106 locutionary 3, 16, 53 see also illocutionary locutionary act 16, 53 see also illocutionary act
M
meaning 3, 713, 15, 17, 20, 2530, 33, 34, 37, 38, 40, 41, 43, 44, 53, 58, 59, 67, 87, 91, 9395, 120, 127, 129, 131, 132, 135, 137, 139141, 143, 144, 146148, 153 155, 164, 166, 171, 183, 196 media 171, 180, 181 mensonge 17, 18, 21, 22, 98100, 109 metaphor 5, 30, 72, 103 modality 2, 5, 10, 16, 31, 35, 9296 see also mood modèle genevois 76, 78 see also Genève monologic 65, 69 mood 31, 33, 34, 36, 49, 53, 58, 8892, 94, 96 see also modality multiculturalism 169, 173, 176
N
nominal 61, 62, 67
O
optative 31, 63, 8794 order 15, 39, 47, 48, 5154, 57, 70, 71, 89, 90, 93, 117, 122, 155, 158, 160, 161, 183 186, 191 organisation opérationnelle 78, 79, 83
P
parenthétique 75, 8286, 84 passing theory 156158 performative 3, 31, 4359, 118, 137 persi¶age 97, 107 perspectival 28, 34, 38
philosophe 77, 108 philosopher 3, 87, 137, 139, 152, 155, 157, 158, 167, 168 philosophie 17, 21, 100, 168 philosophy 24, 10, 12, 17, 19, 26, 40, 41, 43, 58, 59, 95, 131, 133, 137, 138, 152, 159, 161, 168, 177 political, politics 6, 169, 171, 179, 181, 185 pragmatic, pragmatics 13, 68, 10, 13, 16 20, 22, 23, 26, 28, 30, 33, 41, 43, 53, 57 59, 62, 72, 73, 112, 117, 124, 126, 129, 139, 148, 181 pragmatique 17, 18, 21, 22, 78, 101, 111 113 praxéologique 75, 78, 8085 presupposition 15, 91, 152, 171 primates 98, 120, 125 see also singes prior theory 156, 157 promise 43, 44, 47, 56, 58, 125, 137 Proto-Indo-European 13, 90, 91 see also Indo-European
R
race, racial 175, 176, 178 realis 63, 92, 93, 95 see also irrealis rule 6, 10, 11, 18, 19, 44, 64, 118, 142, 148, 155, 156, 162, 165, 178
S
second person 62, 63, 65, 66, 70, 168 semantic, semantics 3, 710, 18, 19, 2831, 33, 34, 3741, 43, 53, 55, 57, 58, 72, 93 96, 123, 148, 158, 164, 184, 186 semantic map 93, 96 semiotic, semiotics 3, 4, 6, 12, 13, 1822, 69, 72, 183186, 188, 189, 195 shared language 156, 169 sharing 25, 70, 151, 156160, 164 singes 98, 112 see also primates society 15, 20, 94, 118, 127, 149, 163, 166, 171, 174, 175, 185
199
200
Index
space 20, 34, 37, 40, 41, 69, 71, 72, 93, 166, 167, 169, 172, 176, 178, 183185, 195 speech 3-6, 1013, 15, 16, 18, 26, 27, 31, 32, 49, 52, 53, 55, 58, 59, 6365, 68, 69, 112, 117, 118, 123126, 137, 155, 157, 161 164 speech act 6, 18, 52, 53, 55, 58, 59, 112, 117, 118, 123126, 137 see also indirect speech acts speech community 4, 11, 15, 157, 163, 164 see also community, linguistic community structuralist 4, 131 subjunctive 8794 syntax 18, 53, 58, 95, 96, 148
tradition 2, 4, 37, 88, 90, 91, 97, 151155, 159, 160, 163166, 165, 170, 173, 185, 189, 190 translation 66, 89, 156, 159, 166, 167 triangulation 161165 truth 35, 711, 13, 14, 17, 29, 43, 45, 46, 55, 119, 130, 151, 153, 154, 156, 157, 168 Turkish 91, 92, 94, 95 typology 5, 72, 87, 92, 9496
T
volitive 89, 90
tagmémique 77, 78 tense 2531, 3340, 53, 94, 95, 184 third person 6166, 68, 135, 137 time 2, 9, 11, 13, 15, 20, 2531, 3341, 54, 56, 57, 92, 93, 119, 120, 124, 135, 145, 156158, 160, 165, 172, 173, 177, 179
U
understanding 4, 6, 7, 9, 11, 12, 14, 15, 17 20, 37, 40, 148, 152160, 163, 165167, 169, 176
V
W
Wittgensteinian 6, 132, 133 world 5, 810, 12, 25, 2931, 33, 37, 52, 68, 69, 95, 125, 126, 133, 137, 141, 151155, 159161, 163167, 172, 180
In the Pragmatics & Beyond New Series the following titles have been published thus far or are scheduled for publication: 79 ANDERSEN, Gisle and Thorstein FRETHEIM (eds.): Pragmatic Markers and Propositional Attitude. 2000. viii, 273 pp. 80 UNGERER, Friedrich (ed.): English Media Texts – Past and Present. Language and textual structure. 2000. xiv, 286 pp. 81 DI LUZIO, Aldo, Susanne GÜNTHNER and Franca ORLETTI (eds.): Culture in Communication. Analyses of intercultural situations. 2001. xvi, 341 pp. 82 KHALIL, Esam N.: Grounding in English and Arabic News Discourse. 2000. x, 274 pp. 83 MÁRQUEZ REITER, Rosina: Linguistic Politeness in Britain and Uruguay. A contrastive study of requests and apologies. 2000. xviii, 225 pp. 84 ANDERSEN, Gisle: Pragmatic Markers and Sociolinguistic Variation. A relevance-theoretic approach to the language of adolescents. 2001. ix, 352 pp. 85 COLLINS, Daniel E.: Reanimated Voices. Speech reporting in a historical-pragmatic perspective. 2001. xx, 384 pp. 86 IFANTIDOU, Elly: Evidentials and Relevance. 2001. xii, 225 pp. 87 MUSHIN, Ilana: Evidentiality and Epistemological Stance. Narrative Retelling. 2001. xviii, 244 pp. 88 BAYRAKTAROĞLU, Arın and Maria SIFIANOU (eds.): Linguistic Politeness Across Boundaries. The case of Greek and Turkish. 2001. xiv, 439 pp. 89 ITAKURA, Hiroko: Conversational Dominance and Gender. A study of Japanese speakers in first and second language contexts. 2001. xviii, 231 pp. 90 KENESEI, István and Robert M. HARNISH (eds.): Perspectives on Semantics, Pragmatics, and Discourse. A Festschrift for Ferenc Kiefer. 2001. xxii, 352 pp. 91 GROSS, Joan: Speaking in Other Voices. An ethnography of Walloon puppet theaters. 2001. xxviii, 341 pp. 92 GARDNER, Rod: When Listeners Talk. Response tokens and listener stance. 2001. xxii, 281 pp. 93 BARON, Bettina and Helga KOTTHOFF (eds.): Gender in Interaction. Perspectives on femininity and masculinity in ethnography and discourse. 2002. xxiv, 357 pp. 94 McILVENNY, Paul (ed.): Talking Gender and Sexuality. 2002. x, 332 pp. 95 FITZMAURICE, Susan M.: The Familiar Letter in Early Modern English. A pragmatic approach. 2002. viii, 263 pp. 96 HAVERKATE, Henk: The Syntax, Semantics and Pragmatics of Spanish Mood. 2002. vi, 241 pp. 97 MAYNARD, Senko K.: Linguistic Emotivity. Centrality of place, the topic-comment dynamic, and an ideology of pathos in Japanese discourse. 2002. xiv, 481 pp. 98 DUSZAK, Anna (ed.): Us and Others. Social identities across languages, discourses and cultures. 2002. viii, 522 pp. 99 JASZCZOLT, Katarzyna M. and Ken TURNER (eds.): Meaning Through Language Contrast. Volume 1. 2003. xii, 388 pp. 100 JASZCZOLT, Katarzyna M. and Ken TURNER (eds.): Meaning Through Language Contrast. Volume 2. 2003. viii, 496 pp. 101 LUKE, Kang Kwong and Theodossia-Soula PAVLIDOU (eds.): Telephone Calls. Unity and diversity in conversational structure across languages and cultures. 2002. x, 295 pp. 102 LEAFGREN, John: Degrees of Explicitness. Information structure and the packaging of Bulgarian subjects and objects. 2002. xii, 252 pp. 103 FETZER, Anita and Christiane MEIERKORD (eds.): Rethinking Sequentiality. Linguistics meets conversational interaction. 2002. vi, 300 pp. 104 BEECHING, Kate: Gender, Politeness and Pragmatic Particles in French. 2002. x, 251 pp. 105 BLACKWELL, Sarah E.: Implicatures in Discourse. The case of Spanish NP anaphora. 2003. xvi, 303 pp. 106 BUSSE, Ulrich: Linguistic Variation in the Shakespeare Corpus. Morpho-syntactic variability of second person pronouns. 2002. xiv, 344 pp.
107 TAAVITSAINEN, Irma and Andreas H. JUCKER (eds.): Diachronic Perspectives on Address Term Systems. 2003. viii, 446 pp. 108 BARRON, Anne: Acquisition in Interlanguage Pragmatics. Learning how to do things with words in a study abroad context. 2003. xviii, 403 pp. 109 MAYES, Patricia: Language, Social Structure, and Culture. A genre analysis of cooking classes in Japan and America. 2003. xiv, 228 pp. 110 ANDROUTSOPOULOS, Jannis K. and Alexandra GEORGAKOPOULOU (eds.): Discourse Constructions of Youth Identities. 2003. viii, 343 pp. 111 ENSINK, Titus and Christoph SAUER (eds.): Framing and Perspectivising in Discourse. 2003. viii, 227 pp. 112 LENZ, Friedrich (ed.): Deictic Conceptualisation of Space, Time and Person. 2003. xiv, 279 pp. 113 PANTHER, Klaus-Uwe and Linda L. THORNBURG (eds.): Metonymy and Pragmatic Inferencing. 2003. xii, 285 pp. 114 KÜHNLEIN, Peter, Hannes RIESER and Henk ZEEVAT (eds.): Perspectives on Dialogue in the New Millennium. 2003. xii, 400 pp. 115 KÄRKKÄINEN, Elise: Epistemic Stance in English Conversation. A description of its interactional functions, with a focus on I think. 2003. xii, 213 pp. 116 GRANT, Colin B. (ed.): Rethinking Communicative Interaction. New interdisciplinary horizons. 2003. viii, 330 pp. 117 WU, Ruey-Jiuan Regina: Stance in Talk. A conversation analysis of Mandarin final particles. 2004. xvi, 260 pp. 118 CHENG, Winnie: Intercultural Conversation. 2003. xii, 279 pp. 119 HILTUNEN, Risto and Janne SKAFFARI (eds.): Discourse Perspectives on English. Medieval to modern. 2003. viii, 243 pp. 120 AIJMER, Karin and Anna-Brita STENSTRÖM (eds.): Discourse Patterns in Spoken and Written Corpora. 2004. viii, 279 pp. 121 FETZER, Anita: Recontextualizing Context. Grammaticality meets appropriateness. 2004. x, 272 pp. 122 GONZÁLEZ, Montserrat: Pragmatic Markers in Oral Narrative. The case of English and Catalan. 2004. xvi, 410 pp. 123 MÁRQUEZ REITER, Rosina and María Elena PLACENCIA (eds.): Current Trends in the Pragmatics of Spanish. 2004. xvi, 383 pp. 124 VINE, Bernadette: Getting Things Done at Work. The discourse of power in workplace interaction. 2004. x, 278 pp. 125 LERNER, Gene H. (ed.): Conversation Analysis. Studies from the first generation. 2004. x, 302 pp. 126 WU, Yi’an: Spatial Demonstratives in English and Chinese. Text and Cognition. 2004. xviii, 236 pp. 127 BRISARD, Frank, Michael MEEUWIS and Bart VANDENABEELE (eds.): Seduction, Community, Speech. A Festschrift for Herman Parret. 2004. vi, 202 pp. 128 CORDELLA, Marisa: The Dynamic Consultation. A discourse analytical study of doctor–patient communication. 2004. xvi, 254 pp. 129 TABOADA, María Teresa: Building Coherence and Cohesion. Task-oriented dialogue in English and Spanish. 2004. xvii, 264 pp. 130 HALMARI, Helena and Tuija VIRTANEN (eds.): Persuasion Across Genres. A linguistic approach. viii, 244 pp. + index. Expected Winter 2004-2005 131 JANOSCHKA, Anja: Web Advertising. New forms of communication on the Internet. 2004. xiv, 230 pp. 132 ONODERA, Noriko O.: Japanese Discourse Markers. Synchronic and diachronic discourse analysis. 2004. xiv, 251 pp. 133 MARNETTE, Sophie: Speech and Thought Presentation in French. Concepts and strategies. xii, 357 pp. + index. Expected Spring 2005 134 SKAFFARI, Janne, Matti PEIKOLA, Ruth CARROLL, Risto HILTUNEN and Brita WÅRVIK (eds.): Opening Windows on Texts and Discourses of the Past. xi, 399 pp. + index. Expected Spring 2005
A complete list of titles in this series can be found on the publishers website, www.benjamins.com