Int J Semiot Law DOI 10.

1007/s11196-010-9201-x

Law and Conversational Implicatures

Springer Science+Business Media B.V. 2010

Abstract This essay investigates the applicability of Grices theory of conversational implicatures to legal interpretation, in order to highlight some of its characteristics. After introducing the notions of language and discourse, and briey explaining the most salient aspects of Grices theory, I will analyse the interpretation of two types of legal acts; authoritative legal acts and acts of private autonomy. Regarding the rst class, exemplied by statutes, I will argue against the applicability of Gricean theory due to the conictual behaviour of the addressees and, above all, to the insurmountable indeterminacy of the contextual elements. As far as acts of private autonomy are concerned, exemplied by contracts, I will argue that the cooperative principle is applicable, at least in those legal systems that include the principle of bona des among the interpretative regulations of such acts. Keywords Bona des Context Discourse Grice Legal interpretation

1 Language and Discourse This essay sets out to investigate the applicability of Grices theory of conversational implicatures to legal interpretation: I think that through a comparative analysis some characteristics of the law and some aspects of Grices theory can reciprocally come to light. Before briey explaining Grices theory, in order to clarifying its signicance, it is necessary to introduce the notions of language and discourse.1
It will not be possible to provide an exhaustive explanation of Grices theory within this paper and to mention all the problems it raises; for a more thorough explanation see, e.g., Grice [14, 15]; Gazar [12]; Levinson [22, 23]; Carston [9]; Atlas [1]. F. Poggi (&) ` dipartimento Cesare Beccaria, Researcher in Legal Philosophy, Universita degli Studi di Milano, Milan, Italy e-mail: [email protected]
1

123

F. Poggi

In accordance with a rather prevalent use of the word language, I dene it as the whole class of communication processes, from the point of view of their content and structure, independently from the concrete ways in which they occur. This understanding of language does not correspond with the meaning of the term in expressions such as English language, lost language or Braille language. Used in such phrases, the term refers to a specic system of conventional sounds (in its spoken form) and/or symbols (in its written form) considered as a concrete communication tool. The notion of language as dened before, on the other hand, is placed at a higher level of abstraction: each natural language (Italian, English, German) represents a case or instance of language, but language (in this sense) does not extend only to the totality of existing natural languages. By discourse I mean a series of occurrences of a language, produced by a given speaker at a given time, a collection of instances of parole.2 Discourse is thus a sequence of language (possibly but not necessarily a single speech act), placed in a given spatiotemporal context. The distinction between language and discourse takes on importance when it is acknowledged that a discourse is able to convey information which does not coincide with the information transmitted by the corresponding language fragment, that is, when the context is considered relevant to the communicative content. In particular, when a given language fragment, a sentence, occurs within a discourse, as an utterance, it can express more specic communications, or different communications, or even additional communications than would be the case if it were considered in a non-contextual way.3 The rst hypothesis occurs whenever the sentence in question is ambiguous or contains anaphoric,4 deictic or indexical5 expressions or when it is incomplete.6 For instance, the sentence

2 3

See, e.g., Saussure [29]; Ross [27, p. 3].

The term utterance suffers from a typical process-product ambiguity, in that it can indicate both the act of stating a linguistic expression, and the product of that act (and in this case the term tokensentence is often used). In this paper, when a distinction is needed, I will use utterance in the latter sense, whereas when referring to the act of enunciating I will use the expression act of uttering. In particular, I intend by utterance the determined spatiotemporal occurrence of a sentence; by sentence I mean a grammatically complete linguistic expression. In brief, a sentence is a fragment of language, whilst an utterance is a fragment of discourse. By anaphora I mean an expression whose meaning and reference is determined by another preceding expression. Personal pronouns (he, she, it, her, etc.) are all anaphoric expressions. Deixis concerns the ways in which languages codify and translate the context of the utterance and the communicative event into grammar: deictic reference is reference to an aspect of the context of utterance or speech event. The terms Indexicals and Deixis are often used as synonyms, although not always. For example, Crystal [10] distinguishes between different meanings of indexical: 1. (ling) Said of features of speech or writing (esp. ? voice quality) that reveal the personal characteristics of the speaker, e.g., age, sex 2. (sem) see deixis. According to many authors, sentences like I will arrive late or You will not die are incomplete, being able to mean I will arrive late at any time in the future, I will arrive late this evening (this afternoon, this week, etc.), You will not die ever, You will not die from this small injury, etc. See Recanati [26, pp. 8ff.]; Bach [7].

123

Law and Conversational Implicatures

(a)

I am reading Johns book

is ambiguous as it could mean I am reading the book that John wrote or I am reading the book that belongs to John or even I am reading the book that John gave me. An interesting case in which a fragment of discourse transmits a different message than would be conveyed if it were understood regardless of the context of utterance is represented by irony. Thus the sentence (b) Its lucky I watered the garden this morning

transmits the opposite of its literal meaning when uttered during a downpour. As can already be seen from these examples, the distinctive characteristic of a discourse does not only consist in being a fragment of language the uttering of which is spatiotemporally located, but also, and above all, in the fact that the speakers are aware of the contextual elements and they adopt them as a means of communication, as tools to make more precise the conventional meaning or to change it.7 Sentence (a) is not ambiguous if, when I state it, we both know that our colleague John has just published his new book and sentence (b) does not really communicate that I think myself lucky to have watered the garden if we both know that it is raining and that it is a waste of time and energy to water the plants in the morning if it rains in the afternoon. In particular, the relevant contextual elements include both the context of utterance, that is, the so-called triple time, place and speaker [20], to which it is opportune to add also the manner of the utterance, for example, if it was uttered laughing or with a grunt, the information relative to the listener, especially about her relationship with the speaker (if they are friends or strangers or countrymen, etc., and, especially, what they know about each other), and the so-called co-text, that is, the set of adjacent utterances, the sequence in which a given utterance occurs.8 It is important to note how, to ensure a correct understanding, the context of utterance and the co-text must be known both by the listener and by the speaker: or better still, the listener must be aware that the speaker also knows them, that the speaker knows that the listener also knows them, that the speaker knows that the listener knows that the speaker knows them and so on. In short what Schiffer termed mutual knowledge must obtain.9
7

The fact that the speakers are aware of the contextual elements and use them as a means of communication does not mean they perform any conscious or deliberate reasoning; comprehension of the discourses is, mostly, automatic. With regard to this, as Rysiew [28, p. 289] observes the immediacy of understanding, as we might call it, is just as illusory as the immediacy of vision; while there is typically no conscious interpretation (etc.) involved in utterance comprehension, some interpretative activity or processing must occur in some form. It would be appropriate to add to these contextual elements also the one which Searle terms Background. An examination of Searles theory is, however, beyond the scope of this work. See Searle [31, 32]. According to Schiffer [30, p. 34] two people may mutually know* that p even though they are not directly acquainted with one another and even though they each have entirely different grounds for thinking that p. Thus, two people who have never met but who know of each other may reasonably assume that they mutually know* that London is a city in England.

123

F. Poggi

Lastly, let us consider the following dialogues (c) A: Do you know what time it is? B: The local news has already started (d) A: Do you know what happened to the chicken? B: Well, the cat is licking its whiskers These dialogues seem perfectly comprehensible: and yet Bs responses, taken literally, would appear completely unrelated to As questions. In both cases, however, the context and, in particular, the co-text, allow an additional communicative content to be attributed to Bs reply or, more precisely, we interpret Bs reply as though it were adapted to the co-text (to As question) and therefore attribute an additional meaning to it. The theory of conversational implicatures developed by Grice provides an explanation of precisely how this is possible: how it is possible that a discourse, an utterance, expresses an additional meaning and, in some cases, one that is different and more precise than the meaning expressed by the corresponding sentence.

2 Grices Theory: A Brief Overview Grice claims that conversations are governed by a series of maxims, aimed at guaranteeing the efcient and effective use of language with a view to interaction. According to Grice, everything we say is interpreted, as far as possible, in a cooperative manner as an appropriate response to the communicative context. For this to take place, however, we often need to go beyond the meaning of what is said; we need to interpret what is said in cooperative manner or, to be more exact, as though it were in accordance (at least to a certain degree10) with the conversational maxims.11 Grice identies four fundamental maxims which express the general cooperative principle (CP): Make your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged [14, p. 26]. The maxims identied by Grice are: Quantity: 1. 2. Make your contribution as informative as is required (for the current purposes of the exchange); Do not make your contribution more informative than is required.

10 The clarication at least to a certain degree is indispensable to take account of cases of so-called exploitation or outing of the maxims: cases in which a maxim is openly violated, not with a view to not cooperating, but with a view to exploiting the evidence of the violation to convey a given message. In Grices opinion irony, metaphors, meiosis and hyperbole fall into this category: c.f. Grice [14, 15]; Levinson [22]. 11 Conversational implicatures are not to be confused with conventional implicatures: the latter are inferences (not truth-functional) that do not derive from maxims, but are connected by convention to particular expressions or lexical elements: see Grice [13, 14]; Levinson [22]; Bach [5, 6], Carston [8]. In the present paper, conventional implicatures will not be investigated and the term implicature will be used to refer exclusively to conversational implicatures.

123

Law and Conversational Implicatures

Quality: Try to make your contribution one that is true, specically: 1. 2. Do not say what you believe to be false; Do not say that for which you lack adequate evidence.

Relation: Be relevant. Manner: Be perspicuous, specically: 1. 2. 3. 4. Avoid obscurity of expression; Avoid ambiguity; Be brief (avoid unnecessary prolixity); Be orderly.

Hence, in case (c), Bs response can be considered appropriate to As question insofar as, in such a context, the utterance The local news has already started communicates I dont know what the exact time is at the momentbecause, in accordance with the rst maxim of Quantity, if I had known, I would have told youbut I can give you some informationwhich, in accordance with the maxim of Relation, I understand and you can understand as pertinent to your question and sofrom which I believe that you can approximately deduct the time, that is The local news has already started. Similarly, in the context outlined by case (d), Bs response can be understood as cooperative in that it can be interpreted as I dont know what happened to the chicken but I think the cat ate it, given that the cat is licking its whiskers. The implicatures, therefore, are based both on the meaning of what is said and on the maxims, which, in turn, are sensitive to the context. More specically, it is the assumption that the maxims (or some of them)12 are followed in the context which allows the speaker to imply, and for the listener to infer, a communicative content which does not coincide with the meaning of what is said.13 As we have seen, in dialogues (c) and (d) the maxims allow the communication of further content, while, in the case of irony, they allow a distinct content to be transmitted which is the opposite of what is said. According to Grice, irony derives precisely from an intentionally obvious violation of the rst maxim of Quality, that is to the act of uttering something that the speaker and the listeners mutually know
12 The violation of a maxim can, in fact, derive from the need to follow a different maxim, in conict with it. Consider the following example formulated by Grice: A: Where does C live? B: Somewhere in the South of France. In this case there is no reason to suppose that B is opting out, but his answer is, as he well knows, less informative than is required to meet As needs. This infringement of the rst maxim of Quantity can be explained only by the supposition that B is aware that to be more informative would be to say something that infringed the second maxim of Quality [14, pp. 3233]. For a critical discussion see Atlas [1, pp. 71ff.] . 13 According to most authors the meaning of what is said does not correspond to the meaning of the sentence, but it includes also other elements, specied by the context: it is equivalent to the proposition expressed by the use of a sentence or the truth-conditional content of the utterance, and is in turn dependent on reference resolution, indexical xing and disambiguation, Levinson [23, p. 171]. See also Grice [14, p. 31]; Atlas, [1, p. 63].

123

F. Poggi

that the speaker believes to be false (that the speaker evidently believes is false, that the listeners know the speaker believes is false, that the speaker knows the listeners know the speaker believes is false, etc.) so as to suggest the exact opposite of what is said.14 In this way, whenever (b) is uttered during a downpour, the context of utterance, mutually known by speaker and listener, makes the insincerity of it obvious, or better still, makes it clear that the speaker intends to make evident the insincerity of what she says and, therefore, suggests that she intends to communicate the opposite content. Some examples in which the maxims can transmit a more specic content are provided by Levinson. In short, according to Levinson, the second maxim of Quantity would establish the principle What is expressed simply is stereotypically exemplied: [b]rief and simple expressions thus encourage [] a tendency to select the best interpretation to the most stereotypical, most explanatory interpretation.15 This principle is behind the following implicature: (e) The spoon is in the cup (e1) The spoon has its bowl-part in the cup Lastly, referral to the maxims seems necessary also to conclude that the speaker said what she said without implying anything. For even if what the speaker means is exactly what he/she says, the hearer must recognize this, even though that (that the speaker means is exactly what he/she says) is not something that is explicitly communicated. And here too the CP plays a role: if the speaker is taken to be speaking literally, it is because that preserves the assumption that the speaker is observing CP [28, p. 293]16 At this point it is necessary to point out what the logical status of the maxims is. With regard to this, Grice seems to think that the maxims do not describe a regularity, but rather are norms (of a customary type). However, I have some doubts about the idea, for the most part undiscussed, according to which the maxims constitute (formulations of) norms in the strict sense of the term. With regard to this it could be useful to refer to a well-known test by Hart. According to Hart, there is involved in the existence of any social rules a combination of regular conduct with a distinctive attitude to that conduct as a standard [17, p. 83]. The latter behaviour nds expression in a general demand for conformity and in, equally general and serious, social pressure upon those who deviate, both accompanied and revealed by typical linguistic expressions. What happens when someone deviates from the maxims? I think that the primary effect of such a deviation does not consist in penalties, hostile reactions or generally widespread criticism, but in incomprehension, in a bad shot at communicative interaction. Certainly, those who deviate from the maxims

14 This interpretation of irony is not without its difculties: see Sperber and Wilson [33, 34]; Grice [15]; Yamanashi [38]. 15 16

Levinson [23, pp. 37ff., and pp. 114ff.]. Cf. also Strawson [35]; Bach [4].

123

Law and Conversational Implicatures

are also criticized, but they are criticized because they do not understand or do not make themselves understood.17 I would like to propose the idea that conversational maxims are primarily (formulations of) customary hermeneutic technical rules. According to von Wrights classication, technical rules (or directives) indicate a means to reach a certain goal, aiming not at directing the will of the receivers, but at indicating to them that their will is conditioned: in other words, that if they want to reach a certain goal, then they must maintain certain behaviour [37, pp. 9 ff.]. The CP and conversational maxims then could be formulated in the following way: (TR) If you want to cooperate and you want to understand and be understood, then you must follow the CP, that is m, m1, Where m, m1, stand for the content of each maxim as formulated by Grice.18 As with all the technical rules, the CP and each maxim presuppose an underlying proposition that must be true for the technical rule to be effective.19 This proposition is approximately equivalent to: (UP) If you follow the CP and, in particular if you follow m, m1,, then you cooperate, understand and make yourself understood Everyday experience seems to show how this proposition is true: so, e.g., in (c), the communication is successful if, and only if, both the speaker and the listener follow the CP; if only one of them follows the maxims, the communication fails; and, if they both violate the maxims, then it is difcult to see even a (failed) attempt to communicate. If Bs response is not (and is not interpreted as) relevant and informative in regards to As question, then it is not a response at all, it is an impromptu observation, and (c) is not a dialogue.20 At this point we could take a further step and ask ourselves why such a proposition is true. A simple response to this could be thats how the world is, or

17 A considerable exception is represented by the violation of the rst maxim of Quality: aside from the hypothesis of exploitation, someone who deviates from this maxim is telling lies and lying is, obviously, a behaviour that produces general criticism, hostile reactions and penalties. With regard to this, however, as well as the conversational maxims there can also be social norms that prescribe their content. 18 Within Gricean literature, the exact content of the CP is controversial: see Ladegaard [21]; Thomas [36]. In this paper, as claried by the above formula, cooperation is intended exclusively as communicative cooperation, as direct cooperation from the common goal of understanding and making oneself understood and not as a sharing of further non-linguistic goals or interests. 19 Not effective in the sense of effectively followed, but effective in the sense that, if the receivers carry out what is laid down by the apodosis, then they reach the goals indicated in the protasis.

For more examples, see Atlas [1], Bach [3], Gazdar [12], Grice [14], Levinson [22], Levinson [23] (as well as the next sections). Of course there can be exceptions: there can be cases in which the communication is successful although the CP is violated. I think that these cases are extremely rare for two reasons at least. Firstly, it is necessary that both the speaker and the listener dont follow the maxims. Secondly, they must share some other (private) rules. For the communication to succeed, it is not sufcient that both the parties are not cooperative: If I know that you are not truthful, informative, relevant, perspicuous, and so on, I still dont know what you are trying to tell me, if something.

20

123

F. Poggi

in other words that everyone follows the CP and the maxims, that there is a general expectation that everyone follows them and that, when this expectation is not satised, incomprehension arises.21 2.1 Generalized Implicatures, Particularized Implicatures and the Problem of Understanding Grice distinguishes between generalized implicatures, those that are always produced, except in special circumstances, and particularized implicatures, which require particular contexts. For example, the utterance (f) I went into a house

always implicates, except in particular circumstances (f1) The house was not mine and (g) (g1) (g2) Some of the guests have already left

possesses the generalized implicature Not all the guests have left

whereas it possesses the particularized implicature It must be late

only in the context of the following dialogue (h) A: Do you know what time it is? B: Some of the guests have already left (i.e., it must be late)

The generalized implicatures above all derive from the rst maxim of Quantity. In (f) if the house had been the speakers, she should have said so, given that it would be relevant information in most cases [14, p. 38]. In particular contexts, however, even similar information may not be important with regard to the communicative exchange in course. In fact, even generalized implicatures are context sensitive: in certain contexts they do not occur. For example, the implicature from An X to not the speakers X does not seem to be produced from the following utterance: (i) This evening I have backache, because I spent all day sitting at a desk

In (i) whether the desk belonged to the speaker does not seem relevant information with respect to the linguistic exchange: what is relevant is that the
21 As Levinson [23, p. 28] states, cognitive sciences have demonstrated the asymmetry between the slowness of human discourse, of the emission of acoustic signals, and the apparently greater speed of human thought. The CP and the maxims allow this asymmetry to be compensated for. However, this only proves that following the maxims is rational, but it does not explain the reason or the motivation for everyone to follow the maxims.

123

Law and Conversational Implicatures

speakers physical problems were caused by sitting all day (it doesnt matter whether at her own desk or someone elses).22 The implicature from (g) to (g1) is a scalar implicature of Quantity. By linguistic scale we mean a set of linguistic expressions of the same grammatical category (so-called scalar predicates) arranged according to a linear order of informativeness or semantic force. Given a scale of this kind, the rst maxim of Quantity makes it possible to formulate a general predictive rule according to which, if a speaker states something that refers to a weaker point on the scale, it is implicated that the same does not hold with respect to a higher point on the scale. Thus, in (g), the fact that the speaker asserted that some of the guests had already left implicates that she knows (or believes) that not all the guests have left, and this is true even if, from a logical point of view, the proposition expressed in (g) is coherent (in that it is logically entailed) with that expressed by All the guests have already left.23 Also, generalized scalar implicatures can not occur in certain contexts. (l) A: Im sorry I made Paul pay the bill B: Paul owns four big properties

In this communicative exchange, Bs response implicates that A has no reason to be sorry, as Paul is well off, while it does not seem necessarily to implicate that Paul has only four properties and no more.24 An interesting aspect of generalized implicatures is that they seem to be partially independent of the intention of the speaker. Let us consider the following example: (m) A: Yesterday a car broke down B: Whose car was it? A: Mine By saying A car broke down, A implicates The car is not mine even if that was not his intention at all. Of course the speakers intention is only partially irrelevant. In the given example, A violated the rst maxim of Quantity and, in doing so, she failed to make herself understood. She did not make herself understood because B supposed or took for granted that A was following the maxims and consequently that the car was not As. In other words, when the speakers intention does not coincide with what she should have implicated on the basis of the maxims and the CP, that is when a speaker implicates something she did not intend to implicate or does not implicate something she intended to implicate, communication is awed. The speakers intention is therefore relevant to

22 23

On the inference from An X to not the speakers X, see Hawkins [18]; Levinson [23]; Atlas [1].

It is debatable whether from the utterance of a weaker assertion one can implicate that the speaker knows that the stronger assertion is false, or that the speaker believes it is false, or even that the speaker does not know whether the stronger assertion is true. See Gazdar [12]; Atlas and Levinson [2]; Levinson [22].
24

For a critical review of this idea and a similar example from Levinson c.f. Atlas [1, pp. 210ff.].

123

F. Poggi

the success of communication. It is on the other hand irrelevant to the production of the generalized implicature: this comes into being even if the speaker did not intend it to and it is exactly this that causes misunderstandings. This also means that the communication failure is attributable to the speaker every time she did not intend to implicate what the utterance implicated (or intended to implicate something that the utterance did not implicate). In contrast, the communication failure is attributable to the listener every time the speaker followed the maxims and intended to implicate what she implicated, but the listener anyway did not understand. This conclusion derives, however, from the conguration of the maxims as hermeneutic technical rules: she who does not follow them, does not understand, does not make herself understood and causes a communication failure. As already mentioned, however, generalized implicatures are also context sensitive: in certain contexts they may not occur. With regard to this two cases can occur. It could be that the context, which impedes the production of a generalized implicature, is not mutually known: in this case the communication failure seems to be caused by the speaker who erroneously relied on this mutual knowledge. Thus if a person hears only (l1) Paul owns four properties

without hearing the whole dialogue (l), she could well believe that the generalized implicature in force is valid, according to which Paul is the owner of exactly four properties. The speaker may well have had good reason to believe that the listener was also aware of the facts relevant to the context but if that awareness did not exist, her mistaken assumption caused the communication failure.25 If, on the other hand, the context in question is mutually known by speaker and listener, it could be held that the communication failure is attributable to the party who ignored the contextual facts which impeded the production of the generalized implicature. Unfortunately, however, things are not always so simple: the context and, above all, its importance with respect to the communicative exchange, are often unclear, not univocally determinable. For example, it could be doubted that in (l) the exact number of properties owned by Paul is not relevant, or at least, it could be argued that if the speaker had known for certain that Paul owned more than four properties, according to the maxim of Quality, she should have said so. With regard to this, Grice believes that a characteristic of implicatures is in fact indeterminacy: Since, to calculate a conversational implicature is to calculate what has to be supposed in order to preserve the supposition that the Cooperative Principle is being observed, and since there may be various possible specic explanations, a list of which may be open, the conversational implicatum in such cases will
25 One might argue that the speaker should always speak clearly. That would be correct, but the clarity is always relative to the context and maxims. The speaker is clear when his utterance is clear once it is interpreted according to the CP, in that particular context as mutually known. The speaker might think to be clear, and he might also have good reasons for thinking so, but he could be wrong (because the listener actually doesnt know or doesnt remember some contextual elements, or because, although he knows them, he omits to take them into account, or, as we will see, because the context is, in some respects, opaque). Causing a communication failure is very different from being responsible for it, from being at fault.

123

Law and Conversational Implicatures

be a disjunction of such specic explanations; and if the list of these is open, the implicatum will have just the kind of indeterminacy that many actual implicata do in fact seem to possess [14, pp. 3940] In my opinion, something similar is true in the case of particularized implicatures as well: these implicatures are also partially independent of the speakers intention. The problem, however, is that, while with generalized implicatures the context is relevant only if there are certain characteristics which impede the production of implicatures which normally take place, the contextual facts are, in contrast, always decisive for the production of particularized implicatures. In this sense the indeterminacy of particularized implicatures is greater. In conclusion, the idea that implicatures are partially independent of the speakers intention does not mean that they are part of the language, or that they are a component of the sentence-meaning. Implicatures derive not just from what is said but also from the assumption according to which the speaker is cooperating and following the maxim.26 This assumption is true only as far as discourse is concerned. Language is, instead, an abstract construction, where there is no context or speakers and obviously no assumptions regarding the behaviour of these subjects.27

3 Legal Language I have argued that conversational maxims are hermeneutic technical rules that normally operate within discourses: are such rules also applicable to legal interpretation? One way to resolve this question is to investigate whether the law can be considered as discourse. I think that this inquiry is interesting for legal scholars because it may help to clarify some aspects of the law and, in particular, the role of context, the position of the speakers, and the possibility to use the conversational maxims in order to determine the meaning of legal acts. Indeed, in the previous sections it emerged that discourse has the following characteristics: (1) it is spatiotemporally determined, viz. it is characterized by a context of utterance and, often, by a co-text; (2) the speakers and the listeners are aware of the contextual elements and adopt them as a means of communication; and nally, with specic regard to conversational implicatures, (3) a CP works in discourses, i.e., each
26 A corollary of the principle of cooperation is that if the speaker does not want to cooperate, but does not want to be misunderstood either, then she must clearly indicate her intention not to involve herself in the communicative exchange. 27 According to Bach [6], denying the intentional character is one of the ten (main) mistakes linked to implicatures, as it leads to thinking of them as properties of the sentences, it does not explain cancellability and the fact that a (non-ambiguous) sentence can produce different implicatures in different contexts. We are not saying that implicatures are properties of the sentences, however: implicatures arise from certain sentences according to context sensitive maxims and this explains why the same sentence can produce different implicatures depending on the different situations in which it is uttered. In particular, generalized implicatures are implicatures that in certain contexts may not be produced and are certainly cancellable (Some of the guests have already left, actually, they have all left), but, if not cancelled, they are produced regardless of the speakers intention.

123

F. Poggi

speaker, except in the presence of signs to the contrary, assumes that the others cooperate in the conversation (i.e., that they follow Grices maxims). Does legal language have these characteristics? First of all it is necessary to explain exactly what is meant by legal language. As has been remarked by many authors, such an expression is ambiguous in that it could refer to both the language that law is, the language of the sources of law, and to the (meta) language used to speak about law. Even if we limit our analysis to the language that law is, however, this notion still risks being too wide, covering areas that are too heterogeneous for our purposes.28 The idea that I intend to put across is that among the documents which, in many legal systems, produce (and/or are authorized to produce) legal norms, both general and particular, some are depicted as discourse, while for others such categorisation is more problematic. Included in the former are private, autonomous acts, of which contracts are a typical example; included in the latter are authoritative legal acts, exemplied by legal statutes.

4 Authoritative Legal Acts: Statutory Interpretation and Conversational Implicatures Can a statute be considered as a discourse? In favour of a positive response it could be argued that the approval and subsequent issue of any statute are spatiotemporally determined acts: acts characterized by a context of utterance which includes both information relating to the speaker, that is, in this case, to members of parliament, and assumptions regarding time and place, that is, regarding the historical moment of the issue, the circumstances which caused it, the political requirements the statute aims to satisfy and the characteristics of the legal system to which it belongs. As for the latter aspect, it is indeed unquestionable that the interpretation of any statute must take into account the particular co-text represented by the legal systems set of other norms: the principle of consistency, particularly regarding the norms expressed by sources of a higher order, such as the Constitution, is in fact unanimously considered a fundamental criterion of every statutory interpretation. Moreover, every statute is, in turn, a co-text: that is every regulation contained in a statute is (and it is claimed that it must be) interpreted in the light of other provisions of the same text, not only to guarantee consistency, but also, and above all, to establish the meaning of the deictic and indexical expressions which are not at all rare in statutory texts. Certainly statutory interpretation cannot be placed within a conversation in the strict sense of the term, that is within discourse in which the parties take turns in communication, yet it could be argued that this does not at all prevent the conversational maxims from operating, not only because they also seem applicable to monologues, but also because the behaviour of the receivers takes on the value of an on-demand response, creating in this way a sort of long distance dialogue. On closer inspection, however, this solution is not convincing. Firstly, in the case of a statute, we do not have a single speaker, but many speakers who express themselves, so to speak, in chorus: nevertheless, regarding
28

For issues regarding the concept of source of law, see Guastini [16].

123

Law and Conversational Implicatures

each of these, different assumptions can apply, that is the context of utterance can be different with regard to each speaker (her opinions, the objectives she proposes etc.) and, as has been seen, the same utterance can give rise to different implicatures when some contextual assumptions change. More precisely, members of parliament often agree only on the textual formulation of legal regulations which they can interpret in different ways, just as the political aims they pursue with the same legal text can be very different. Moreover, as recently claimed by Marmor, it is not clear either which parties to the conversation are signicant, in other words it is not clear whether only the legislators must be consideredonly those who approved the statute or only those who approved it knowing its text?29or also other political agents (such as administrative agencies, various committees, lobbying groups, etc.): the main problem here is not the complexity of the situation but the fact that there are no clear criteria of relevance [25, p. 435]. Please note that the point is not whether to take into account what is generally called the intention of the legislator: the problem is that it is not clear who the subjects of this intention must be and that, even if they were determined, it would be unlikely to ascribe them univocal opinions, beliefs or aims regarding the content of the statute: the contextual assumptions relative to the speakers often change signicantly.30 All this also has an effect on the second characteristic of discourse: the awareness, of both speaker and listener, of the contextual elements and the recognition of these as a communicative means. The establishment of the context of utterance, apart from its lack of univocity, is undoubtedly problematic for the addressees: as Marmor observes, [j]udges and litigants are not parties to the legislative conversation, so to speak, and they have to rely on secondary sources to gather the relevant information [25, p. 434]. Despite Marmors apparent position, however, it is not merely a practical problem, but also a theoretical one: if the assessment of the data of the context of utterance is problematic, how can the speaker assume that the listener knows the relevant information and, thus, rely on it for a successful outcome to the communication? If I run into you on the street and I tell you that I am very agitated and I need to calm down, I cannot expect that you will offer me a cigarette if you do not know (and I know that you do not know or I do not know if you know) that I am an inveterate smoker and that I nished my own packet of cigarettes some hours ago. This is a clear case in which the

29 Frequently members of parliament vote for laws that they do not even know, just by following the indications of the party to which they belong to. This aspect becomes more serious in the case of extremely technical legislation, the text of which is generally written by experts who are not members of parliament.

Unlike Marmors claim [25], the difculty of attributing communicative intentions to a collective agency is not deemed important in this paper in that it concerns the intention to follow Grices maxims. As we saw in Sect. 2.1, implicatures are, in fact, independent of the effective intention of the speaker. Instead it is claimed in this text that the opinions, beliefs and goals of the members of parliament (that is the elements that the doctrine normally refers to under the heading intention of the legislator) are important in that they constitute elements of the context of utterance and therefore inuence the identication of the communicative content.

30

123

F. Poggi

communication failure depends on the fact that the speaker relies on contextual elements that the listener does not know (Sect. 2.1).31 Analogously in the case of law, mutual knowledge of the elements of the context of utterance, which could be relied upon to determine the content of the statute, rarely occurs. I believe that similar considerations apply to all contextual elements, even those not attributable to speakers (to their identication, aims, opinions etc.). It has already been mentioned that the particular co-text, made up of other norms of the system in some way signicant to the interpretation of a given statute, takes on importance in the legal eld, for example because regulating the same or similar material or expressing general principles they can pose problems of consistency. Once it is established that a similar co-text can be very complex and its identication extremely problematic, how is it possible to be condent of the fact that the addressees of a given statute know it? It would be like expecting someone to understand what is communicated by a sentence in a dialogue she did not hear. Of course one can imagine legal systems in which the co-text is not problematic, in which a few generally known clear rules exist, in which the majority of the members of parliament are extremely cohesive (i.e., the expression of a very close single party), in which it is clear that a given statute was issued to meet a given political requirement, viz. in which the data relative to the context of utterance (to the reasons for the statute, its intended goals, etc.) are univocal and, moreover, easily accessible. However, real legal systems are very distant from this model and even with such an ideal legal system it could be questioned whether the univocity of contextual data and their mutual knowledge are sufcient to ensure that the legislators successfully entrust them with the outcome of the communication, that is the interpretation of the statute. Here the third characteristic of discourse is in question: the effectiveness of the CP. With regard to this, it would seem that the effectiveness of the maxims, as hermeneutic technical rules, is a question of fact: in certain systems the custom of following such rules may be consolidated, in others not. Nevertheless I believe that there is a general characteristic of the law that always obstructs the effectiveness of the CP. Law is an often-conictual undertaking: it is a game in which it is convenient to cheat [19, p. 130]. The addressees of norms usually demonstrate an interest in cooperating, that is an interest often not spontaneous, but motivated by the presence of sanctions, in obeying the law. However it is a physiological characteristic of many legal systems that there are cases of violation and, before that, cases of presumed violation in which it is argued in front of a judge if a given law was
31 As an anonymous referee noted, my utterance is unintelligible also if we mutually know that you dont smoke. This case is quite different from the one considered in the text: here the communication failure depends on the fact that the speaker doesnt take into account a contextual element which, in fact, is mutually known. It seems to me that, in statutory interpretation, a similar situation takes place when the legislator enacts provisions that are clearly inconsistent with constitutional principles (or, better, with their strengthened interpretation), especially if we consider the judges, not the common citizens, as receivers. Other cases are represented by some (scholastic) examples of absurd provisions: e.g., a statute that forbids shing in a State that doesnt border on sea. Generally speaking, in statutory interpretation the assumptions relative to the receivers that do not concern the knowledge of contextual elements, but, e.g., who the receivers are, how their relation with the speakers is, are not important: normally statutory addressees are (and they are mutually known to be) indeterminate.

123

Law and Conversational Implicatures

violated. Legal trials are fought not only with regard to facts, but also with regard to words: the issue is not only if certain behaviour happened, but also if that behaviour is illegal, if the law prohibits it. The parties in litigation are interested not in understanding what the legislator really wanted to say, but in winning the case. Even in the simplest systems and with the clearest regulations, ordinary language, in which the law is formulated, presents vast traits of ambiguity, vagueness and incompleteness and it always leaves a margin of dispute. If in ordinary conversation the context can intervene and eliminate such uncertainties, in a trial even this will be the subject of controversy: as far as possible the parties will seek to twist the reconstruction of the context in their favour. It was seen in Sect. 2.1 that even in everyday discourse implicatures are (often) indeterminate: in a trial the parties will openly aim to expand this characteristic. Not only will there be discussion about which implicatures there are, which maxims are followed and the signicant contextual elements, but their effectiveness will also be the subject of controversy. In legal interpretation both the criterion according to which ubi lex voluit dixit, ubi noluit tacuit and the criterion that lex minus dixit quam voluit are normally in force, the rst seemingly excluding implicatures whereas the second, in contrast, seemingly permits them. In short, in statutory interpretation the Gricean maxims do not represent normal hermeneutic criteria reciprocally assumed by the addressees of the statute or, to be more exact, by the parties in the trial, due to their unrestrained conictual behaviour: the existence of divergent (non communicative) interests creates less interest in cooperating with a view to communicative understanding. Obviously the parties, in order to understand and be understood by the judge must follow the maxims and the CP; cooperative spirit is, however, absent from the understanding of the legal text: the understanding upon which the outcome of the trial principally depends. It follows that, if the parties effectively do not cooperate, if they do not assume that the statute is formulated according to the maxims, then the legislator cannot reasonably rely on this cooperation to specify the meaning communicated. We could be tempted to conclude that a good legislator should never implicate anything, but things are not quite so simple. As has just been said, the conictual behaviour which characterises the trial leads the parties to claim that such implicatures exist whenever it is in their interest to do so. As we have seen in Sect. 2, the conclusion that the speaker intended to communicate only what he said, without implicating anything, also requires the adoption of the CP and it is precisely that adoption which is in doubt in legal controversies. However, even assuming that the parties in the trial have no interest in cooperating, could we not claim that this interest always exists within the duty of the judges? Much depends on the characteristics of the system and on the hermeneutic criteria consolidated in the legal culture. In real legal systems cooperation is often difcult because, as previously mentioned, the context of utterance is unclear and not univocal and the identication of the co-text is often problematic. What is more, the co-text itself, i.e., the set of the legal norms which are relevant to the legal interpretation of a given statute, is often identied by the judges: more precisely it is reconstructed by the judges, with reference to the case in question, independently of any assumption relative to the speakers, i.e., the legislators, effective knowledge

123

F. Poggi

of it. It was seen in Sect. 2.1 that in ordinary discourse implicatures are independent from the effective intentions of the speaker, but not from the assumption that the speaker intends to cooperate and follow the maxims in a given mutually known context. In contrast, judges reconstruct the signicant co-text with a view to interpreting a given statute without conducting any investigation into the members of parliaments effective knowledge of it; knowledge which, particularly in difcult cases, is very hard to assume. Judges act as though the whole legal system were produced by a single (coherent) speaker, but this is evidently not the case. In terms of ordinary conversation we nd ourselves confronted here with a case of failed communication because the listener adopted a context of communication, in particular a co-text, which the speaker did not know. In the legal eld, however, similar procedures are generally accepted.32 Furthermore, since the reconstruction of the co-text and the determination of how it affects the interpretation of a given statute can be a very complex operation, different judges can arrive at different results: much will depend also on the persuasive ability of the parties lawyers. From what has been said, it doesnt follow that a statute is something like a fragment of language (an abstract, acontextual entity): if it were so, and if it were considered so, it would be much more unintelligible than it really is (Sect. 1). On the contrary, as we have seen, the contextual elements are important for interpreting statutory provisions: understanding a statute would be very difcult without taking into account the other norms of the legal system, without considering the statute as a co-text, and, at least sometimes, without making hypotheses about the legislators aims. But each of these contextual elements is, in itself, unclear and indeterminate; its reconstruction is subject to controversy and is often performed discretionarily, independently from the effective knowledge of the speakers. So a statute is a very particular discourse: a discourse in which there is no CP in force. In my opinion this explains why statutory interpretation is in many respects more complex than ordinary understanding. I believe the same conclusion applies to all texts which I propose to name authoritative legal acts: acts producing legal effects erga omnes, issued by collective agencies and containing general and abstract norms. Issued by collective agencies, they do not express particular and concrete norms and their effectiveness is not limited to a single case; all this ensures that, with respect to these acts, the context is indeterminate, it cannot act as an interpretative guide, but is open to contrasting reconstructions by the parties, who pursue contrasting interests which can even twist interest in linguistic understanding.

5 Acts of Private Autonomy: The Principle of Bona Fides We will now consider the acts of private autonomy, viz. the acts with which individuals regulate their own legal relations through singular norms that produce
32 For this reason, the role of judge doesnt gel exactly with the role of listener. Judges are peculiar listeners in the sense that they reconstruct the context of communication, irrespective of what the speaker (legislator) actually knows and believes, authoritatively xing the very meaning of legal texts.

123

Law and Conversational Implicatures

effects which are both recognized and guaranteed by the legal system. I will limit the following analysis principally to contracts, but I believe that, with appropriate adjustments, it can be extended to other acts of private autonomy, such as wills and other unilateral acts. At rst sight, contracts are clearly discourses: they are language sequences, moreover they may be in oral form and are spatiotemporally determined, that is they take place between identied speakers, within mutually known contexts of utterance and co-texts. With regard to the contracts we nalise every day, the effectiveness of the Gricean maxims and their normative reformulation seems conclusive.33 If I go into a bar in Italy and I ask for a coffee, I expect to be served a coffee (scalar implicature of Quantity), to be served a normal espresso coffee, not instant coffee, barley coffee, macchiato or decaffeinated coffee (implicature derived from the second maxim of Quantity, What is expressed simply is stereotypically exemplied), to be served now and not after an hour (normative reformulation of the maxim of Manner). Different assumptions apply in different countries: so if I wanted the same thing in the USA I would not ask for a coffee, but for a single espresso. Of course, what is stereotypical varies across cultures. In written contracts, in order to ensure the consistency of the interests at stake, the parties are more explicit and careful when specifying (often in technical terms) the exact signicance of their respective obligations: with respect to this category, however, the presence of implicatures (especially generalized) is not rare and in any case, as we saw in Sects. 2 and 4, even literal interpretation requires the adoption of the CP. It is necessary, however, to ask ourselves whether a cooperative principle is in force in the interpretation of contracts or, alternatively, whether the same considerations made with regard to statutory interpretation cannot be repeated mutatis mutandi. After all, each party pursues its own interests, which are almost always in conict with those of its counterpart and this conict can be reected in communicative cooperation: it is a fact that when a trial refers to contractual material, the parties argue not only about the interpretation of the legal provisions in question, but also about the interpretation of the contract. With regard to this it could seem that on conclusion of the contract, in order to reach agreement, the parties must have cooperated and therefore that controversy emerges only later and is caused by the bad faith of one of them. Nevertheless, as is the case with statutes, here too agreement could concern just the textual formulation of regulations the parties understood differently, while, in the case of verbal contracts, the parties could have understood their own or others declarations differently. In my opinion, the value of the cooperative principle depends principally on the interpretative norms in force: in particular, Grices theory is applicable in those legal systems that include the principle of bona des among the interpretative regulations of contracts. The principle, derived from Roman law, states that contracts must be interpreted according to bona des and it can be understood as a
The conversational maxims formulated by Grice mainly regard informative discourse, that is discourse aimed at providing information; nevertheless Grice himself [14, pp. 2829] claims that the theory can be extended, with appropriate adaptations, also to other forms of communication and provides some examples relative also to normative discourse, i.e. discourse aiming to direct others behaviour.
33

123

F. Poggi

norm which enforces the understanding of contractual declarations according to the CP, that is, it enforces the assumption that the parties demonstrated cooperative behaviour, that, in this context, they followed the maxims and therefore indirectly obliges the parties to demonstrate such behaviour. The declarations of the parties must therefore be interpreted in compliance with the Gricean maxims, taking account of the context. Let us look at some examples. If the vendor declared that a horse for sale suffers from weak hooves, according to the maxim of Quantity, this implicates that it only suffers from weak hooves and does not have any other more serious diseases or disorders [24, p. 50], unless other contextual elements override this implicature. If the estate agent declares that a new underground station will soon be built near to the house for sale, according to the maxim of Quality (unless other contextual elements override this implicature) this implicates that she has sufcient proof of this. If the landlord declared that he would whitewash the property at his own expense and hand it over to the tenants by giving them the keys according to the maxim of Manner (unless other contextual elements override this implicature) it is to be understood that the landlord will whitewash the property at his expense and then hand it over to the tenants. Moreover, the applicability of the CP also requires that a party cannot interpret a regulation according to the context that (she knew that) the other party did not know and, vice versa, neither can she interpret it ignoring the mutually known context.34 The bona des principle, furthermore, ensures that the party which caused the communication failure (Sect. 2.1) is considered legally responsible for this and is liable for compensation or compulsory redemption. Please note that this principle does not require proof that one of the parties acted in bad faith: as mentioned in Sect. 2.1, the effective intention is not signicant in the production of the implicatures, what counts is, instead, the common, shared assumption that the speaker intended to follow the maxims. In law a similar assumption is not of customary nature, it does not derive from a generalized and reciprocal expectation, but is imposed by a specic norm (which, in some systems, can also be of customary nature).35 The bona des principle represents the basic criterion of contractual interpretation but certainly cannot resolve all problems. As was seen in Sect. 2.1, one characteristic of implicatures is their indeterminacy: an indeterminacy which, in the context of a trial, the parties and their lawyers will tend to emphasize, particularly through different reconstructions of the effective circumstances which make up the context of the contractual declaration.36
34 35

Given that effective knowledge is difcult to prove, the law often accepts its knowability.

The applicability of the CP to contractual interpretation is also claimed by Frade [11] and, with regard to Roman law, by Mantovani [24]. Whereas Frade seems to claim that this applicability derives from the nature of the contract as a promise or set of promises constituting an agreement between the parties [11, p. 337], according to Mantovani, on the other hand, it derives from the principle of bona des which is equivalent to the CP. With regard to this I prefer to argue that the legal principle of bona des is not equivalent to, but requires respect of the CP: in that, unlike ordinary discourse, in contractual interpretation the maxims constitute hermeneutic rules which are not spontaneously followed, but rather required.
36 Regarding unilateral acts things are actually simpler: here it is a question of exclusively interpreting the declaration of a subject. With regard to these acts however, legal controversies can also arise as to

123

Law and Conversational Implicatures

6 Conclusions In this paper I have claimed that the theory of conversational implicatures is not applicable to statutory interpretation or interpretation of other authoritative legal acts; it is applicable, instead, to the interpretation of contracts and other acts of private autonomy. The main difference between these two categories lies in the fact that, whereas with regard to authoritative legal acts the context (including the context of utterance and the co-text) is radically indeterminate, in acts of private autonomy it is, on the other hand, delimited, even if in particular cases, its reconstruction in the trial can be problematic. Legal authoritative acts and acts of private autonomy certainly do not make up the complete set of the sources of law: further research should concern the interpretation of sentences, precedents and discretionary public administration acts.37 A comparison to ordinary discourse could prove useful also for these acts as highlighting similarities and differences contributes to the clarication of their pragmatic function.
Acknowledgments I want to thank Prof. Mario Jori and Prof. Riccardo Guastini for their helpful comments and suggestions.

References
1. Atlas, J.D. 2005. Logic, meaning and conversation. Oxford: Oxford University Press. 2. Atlas, J.D., and S.C. Levinson. 1981. It-clefts, informativeness, and logical form: Radical pragmatics. In Radical pragmatics, ed. P. Cole, 161. New York: Academic Press. 3. Bach, K. 1994. Conversational implicature. Mind and Language 9: 124162. 4. Bach, K. 1997. The semantic-pragmatic distinction: What it is and why it matters. http://online. sfsu.edu/*kbach. Accessed 15 September 2009. 5. Bach, K. 1999. The myth of conventional implicature. Linguistics and Philosophy 22: 327366. 6. Bach, K. 2006. The top 10 misconceptions about implicature. In Festschrift for larry horn, ed. B. Birner, and G.A. Ward. Amsterdam: John Benjamins. 7. Bach, K. 2006. Impliciture vs. explicature: Whats the difference? http://userwww.sfsu.edu/*kbach/. Accessed 15 September 2009. 8. Carston, R. 2002. Linguistic meaning, communicated meaning and cognitive pragmatics. Mind and Language 17: 127148. 9. Carston, R. 2004. Truth-conditional content and conversational implicature. In The semantic/pragmatics distinction, ed. C. Bianchi, 1848. Stanford, CA: Center for the Study of Language and Information. 10. Crystal, D. 1987. The Cambridge encyclopedia of the english language. Cambridge: Cambridge University Press (CUP). 11. Frade, C. 2002. The legal cooperative principle: An essay on the cooperative nature of contractual transactions. International Journal for the Semiotics of Law 15: 337343. 12. Gazdar, G. 1979. Pragmatics, implicature, presupposition and logical form. Orlando, FL and London: Academic Press.

Footnote 36 continued their interpretation secundum bonam dem that is, in our terms, as to what the implicatures of certain utterances are, and, also in this case, this will depend to a large extent on the reconstruction of the context.
37 In some legal systems even interpretation of the Constitution, which is itself an authoritative legal act, could present characteristics making a separate analysis necessary.

123

F. Poggi 13. Grice, P.H. 1961. The causal theory of perception. Proceedings of the Aristotelian Society 25(Suppl.): 121152. 14. Grice, P.H. 1967. Logic and conversation. Reprinted in Grice, P.H. 1989. Studies in the Way of Word, 2240. Cambridge, MA and London: Harvard University Press. 15. Grice, P.H. 1989. Further notes on logic and conversation. In Grice, P.H. 1989. Studies in the Way of Word, 4157. Cambridge MA and London: Harvard University Press. 16. Guastini, R. 1996. Fragments of a theory of legal sources. Ratio Juris 9: 364386. 17. Hart, H.L.A. 1961. The concept of law. Oxford: Clarendon Press. 18. Hawkins, J. 1991. On (in)denite articles: Implicatures and (un)grammaticality prediction. Journal of Linguistics 27: 405442. 19. Jori, M. 1995. Denizioni giuridiche e pragmatica. In Analisi e diritto 1995, ed. P. Comanducci, and R. Guastini, 109144. Turin: Giappichelli. 20. Kaplan, D. 1978. On the logic of demonstratives. Journal of Philosophical Logic 8: 8198. 21. Ladegaard, H.J. 2009. Pragmatic cooperation revisited: Resistance and non-cooperation as a discursive strategy in asymmetrical discourse. Journal of Pragmatics 41: 649666. 22. Levinson, S.C. 1983. Pragmatics. Cambridge: CUP. 23. Levinson, S.C. 2000. Presumptive meaning. Cambridge, MA and London: MIT Press. 24. Mantovani, D. 2008. Lingua e diritto. In Il linguaggio giuridico. Prospettive interdisciplinari, ed. ` G. Garzone, and F. Santulli, 1756. Milan: Giuffre. 25. Marmor, A. 2008. The pragmatics of legal language. Ratio Juris 21: 423452. 26. Recanati, F. 2004. Literal meaning. Cambridge: CUP. 27. Ross, A. 1968. Directives and norms. London: Routledge and Kegan Paul. 28. Rysiew, P. 2007. Beyond words: Communication, truthfulness, and understanding. Episteme 4: 285304. ne 29. de Saussure, F. 1916. Cours de linguistique ge rale. Lausanne and Paris: Payot. 30. Schiffer, S.R. 1972. Meaning. Oxford: Clarendon Press. 31. Searle, J.R. 1979. Expression and meaning. Cambridge: CUP. 32. Searle, J.R. 1980. The background of meaning. In Speech act theory and pragmatics, ed. M. Bierwisch, F. Kiefer, and J.R. Searle, 221232. Dordrecht, Boston, MA and London: D. Reidel. 33. Sperber, D., and D. Wilson. 1981. Irony and the use-mention distinction. In Radical pragmatics, ed. P. Cole, 295318. New York: Academic Press. 34. Sperber, D., and D. Wilson. 1997. Irony and relevance: a reply to Drs Seto, Hamamoto and Yamanashi. In Relevance theory: Applications and implications, ed. R. Carston, and S. Uccida, 283293. Amsterdam: John Benjamins. 35. Strawson, P.F. 1973. Austin and locutionary meaning. In Essays on J.L. Austin, ed. I. Berlin, et al., 4668. Oxford: Clarendon Press. 36. Thomas, J. 1998. Cooperative principle. In Concise encyclopedia of pragmatics, ed. J. Mey, 176179. Elsevier: Amsterdam. 37. von Wright, G.H. 1963. Norm and action. A logical enquiry. London: Routledge and Kegan Paul. 38. Yamanashi, M. 1997. Some issues in the treatment of irony and related tropes. In Relevance theory: Applications and implications, ed. R. Carston, and S. Uccida, 271282. Amsterdam: John Benjamins.

123