Minimal contents, lying, and conventions of language

One recurrent objection against minimalism is that minimal contents have no theoretical role. It has recently been argued that minimal contents serve to draw the distinction between lying and misleading. In Sect. 1 and Sect. 2 I summarise the main argument in support of that claim and contend that it is inconclusive. In Sect. 3 I discuss some cases of lying and some of misleading that raise difficulties for minimalism. In Sect. 4 I make a diagnosis of the failure of minimalism with those cases. In Sect. 5 I strengthen the case against minimal contents by addressing two received says-based definitions of lying. My analysis of the failure of minimalism suggests that the distinction between lying and misleading, at least in some important cases, calls for a kind of utterance content that is grounded on conventions of language.


Introduction
Borg (2019) argues that minimal contents 1 serve to explain intuitive judgements on strict linguistic liability. In this paper I focus on Borg's discussion of linguistic liability only in relation to the distinction between lying and misleading (Borg 2019: §3). 2 1 According to Borg (2010: 24-25), semantics delivers minimal contents that are maximally free from contextual effects and provide the literal meanings of sentences. According to semantic minimalism, context is semantically relevant only when it is required by a standardly context-sensitive syntactic element, e.g. indexicals, demonstratives, tense markers (e.g. the Basic Set in Cappelen & Lepore 2005). Borg (2004) argues that minimal contents are not speech act contents. 2 In her paper Borg addresses many important issues other than the distinction between lying and misleading. In particular Borg argues that the notion of what is said is multifarious and plays so different theoreti-In the next section I argue that Borg's argument to the conclusion that minimal contents play a role in the explanation of the distinction between lying and misleading is inconclusive. In the following two sections I discuss some cases that are problematic for Borg's minimalism and I make a diagnosis of the failure of minimalism with those cases. In the last section I strengthen the case against minimal contents by addressing two received says-based definitions of lying. My analysis of the failure of minimalism leads to the conclusion that the distinction between lying and misleading, at least in some important cases, calls for a kind of utterance content that is grounded on conventions of language and is independent from communicative intentions.
Borg argues that the notion of what is said is multifarious and arises from judgments on speakers' responsibility for contents of their utterances. She claims that there is a kind of strict linguistic liability according to which speakers are held responsible for minimal contents. People make judgments that track minimal contents when they distinguish lying from misleading. Elaborating on Saul (2012: 37), Borg (2019: 522) discusses the following case. A rich Catholic decides to leave his fortune to Jack, as long as Jack has lived his life in full compliance with the precepts of Catholicism. The rich Catholic asks John for information about Jack. John intends to favour his friend and, knowing that Jack had two children before getting married, he says: (1) Jack got married and had two children.
John expects the rich Catholic to understand that Jack got married and then had two children.
Borg says that people have the intuition that John is misleading, not lying. The content for which John is responsible in the strict linguistic sense is the truth functional content, not the temporally ordered content. Borg argues that minimalism is necessary in order to capture intuitions on this case of misleading. 3 Before moving to the next section some preliminary clarifications are in order. First, throughout the paper I will assume that people have the intuition that John did not lie, as Borg and many other scholars claim. That laypeople have such an intuition has been put into question by Wiegmann et al., (2021) and Reins & Wiegmann (2021) with two recent surveys. Wiegmann et al. and Reins & Wiegmann collected data that show that in many cases people have intuitions that speakers lied even if true minimal contents for the uttered sentences are available. More importantly, in many cases people have intuitions that speakers lied with generalised or particularised conversational implicatures, though they are cancellable. Wiegmann et al.'s and Reins & Wiegmann's main theoretical purpose is to argue against says-based definitions of lying showing that they return too narrow a concept of lying which cal roles that they can hardly be covered by a unique theoretical entity. I agree with Borg on this claim. In this paper I focus only on that part of Borg's paper in which she discusses the role of minimal contents in relation to the distinction between lying and misleading. This issue has gained centrality for philosophers of language who investigate what notion of what is said is best suited to philosophical accounts of lying. They test philosophical theories on the interface between semantics and pragmatics against accounts of the distinction between lying and misleading (e.g. Saul 2012;Michaelson 2016;Stokke 2018). 3 Borg (2019: 521) gives the following definitions (p is the minimal content of sentence S and p* is another content):Lying: the speaker believes that p is false and utters S to convey that p;Misleading: the speaker believes that p is true and p* is false and utters S to convey that p*;Deceiving: the speaker believes that p and p* are false and utters S to convey that p*.
does not correspond to the folk concept of lying. They favour a commitment-based definition of lying close to Viebahn's one (2020Viebahn's one ( , 2021. (See Marsili 2020 for another commitment-based definition).
My purpose in this paper is not to take a direct stand in the debate on philosophical definitions of lying. 4 Rather, my purpose is to argue against minimalism in semantics by showing that minimal contents have no theoretical role in the explanation of the distinction between lying and misleading. Although my interest is more in the field of philosophy of language than in the filed of ethics, my purpose is of interest for scholars who work at definitions of lying, independently of the kind of definition they favour. In Sect. 5 I argue that two received says-based definitions of lying suffers from difficulties that are related to minimal contents. Excluding that minimal contents have a theoretical role in the analysis of lying is of interest also for scholars who favour commitment-based definitions of lying. One way of being committed to a content p is to assert that p by saying that p, and one way of not being committed to a content q is to retain deniability for q and to retreat to a content p by claiming that p and not q is the content of the utterance. If Wiegmann et al.'s and Reins and Wiegmann's surveys are reliable, they show that in many cases this form of deniability and defence from the charge with lying is not possible. Yet, for sure there are cases in which that form of deniability is possible. Thus, even if Wiegmann et al. and Reins and Wiegmann are right, Borg's attempt retains centrality at least with respect to some cases in which lying is distinct from misleading. And if successful, Borg's attempt would establish a theoretical role for minimal contents. 5 4 Nor is my purpose to take a stand in the philosophical debate about the definition of assertion, though many scholars, Borg included, draw interesting conceptual connections between the project of providing a definition of lying and the project of giving a definition of assertion. Supporters of says-based definitions of lying tend to work with a notion of assertion grounded on a notion of what is said (Stokke 2018). Supporters of commitment-based definitions tend to spell out the notion of assertion in terms of commitment to contents (Marsili 2020; Viebahn 2021). 5 Some considerations are in order with respect to Wiegmann et al.'s (2021) and Reins & Wiegmann's (2021) surveys, and with respect to empirical surveys on the concept of lying in general (For an overview of experimental studies on the folk concept of lying see Wiegmann & Meibauer 2019). First, these surveys try to show that philosophical definitions do not capture the folk concept of lying. Even if they are reliable, this does not mean that there is no room for a philosophical analysis of lying that addresses a narrower phenomenon than the folk concept. For example forms of lying like perjury in courtrooms, and in general forms of lying that can be targeted by laws and penal codes (see Skoczeń 2021 for a survey whose results motivate the project of defining a concept of lying in the field of jurisprudence that diverges from the folk concept). And even if limited to such a narrower phenomenon, the theoretical question as to what notion of what is said is best suited for a correct philosophical analysis still remains. And it remains even from the perspective of commitment-based definitions, since at least one form of commitment to a content p comes from asserting that p by saying that p. Second, empirical surveys like Wiegmann et al.'s (2021) and Reins & Wiegmann's (2021) raise a lot of conceptual and methodological issues that should be addressed with great care before taking their results to be reliable. For example, laypeople might use two concepts of lying, a broad one and a narrow one. One objection, then, is that the tests in those surveys trigger intuitions related to the broad concept only. The authors of those surveys address this objection, and I am not claiming that their responses are wrong. I think, however, that they are not conclusive. Another aspect is that surveys should gather empirical data also on how laypeople lie and mislead avoiding lying, and not only on when laypeople judge others speakers to be lying or misleading. Third, it is a highly debated methodological issue whether, and to what extent, intuitions have an evidential role in philosophy and in the method of cases. Wiegmann et al.'s (2021) and Reins & Wiegmann's My claim is that Borg's attempt is not successful. In order to capture the form of deniability in cases in which the speaker is able to discharge the accusation of lying, Borg's minimalist notions of linguistic liability and semantic content are not suitable. 6 Second, one might wonder whether the need to find a theoretical role for minimal contents is really pressing for minimalists. After all, don't we need minimal contents as the output of a semantic module or a semantic competence that works independently of contextual information? I reject the view that minimal contents are returned by a semantic module or a semantic competence working independently of contextual information. I follow the view (see Korta & Perry 2007b, Perry 2007) that a Kaplanesque two dimensional semantics, with quantifications over the parameters demanded by the meaning (character) of expressions, returns token-reflexive contents that provide a kind of truth conditions for utterances when contextual information is not available or ignored. 7 However, the kind of theoretical role that Borg has in mind with respect to the distinction between lying and misleading cannot be covered by token-reflexive contents. Thus, Borg's attempt deserves close attention in philosophy of language, because if it were successful, it would establish a role for minimal contents beyond the issue on what kind of contents are returned by a semantic module or a semantic competence.
Third, I have to clarify some distinctions I will use in the following sections. I contrast minimalism with contextualism. Minimalism (Borg 2004(Borg , 2012Cappelen & Lepore 2005) covers semantic theories according to which the conventional meanings of the expressions occurring in a declarative sentence determine, in accord with the syntactic structure of the sentence, a full propositional content with determinate truth conditions. Exceptions are the conventional meanings of overt context-sensitive expressions of the Basic Set (Cappelen & Lepore 2005), which do not determine context invariant contents but provide guidance for speakers to express different contents in different contexts of utterance.
Contextualism (some scholars prefer to call it Linguistic Pragmatism, see Neale 2004, Devitt 2021) comprises a family of theories (Carston 2002;Recanati 2004Recanati , (2021 surveys are groundbreaking because they open the field to a lot of future interesting works, both conceptual and empirical. 6 If some readers have the intuition that John lied, they can enrich the setting of the scenario up to a point where John did not lie because he retains deniability for the temporally enriched content and is allowed to retreat to the truth functional one (a point where John is not committed to the temporally enriched content). 7 Take for example the sentence 'I am tired' uttered by speaker A. Minimalists (Borg 2004Cappelen & Lepore 2005) say that semantics returns the content that A is tired. Korta & Perry (2007b: 100-101) hold that semantics returns the token-reflexive content that the speaker of the utterance is tired. That A is tired is the locutionary content, which is returned by exploiting the contextual information that A is the speaker of the utterance. But the contextual information that A is the speaker of the utterance is not semantic information. The semantic machinery works with compositional clauses that quantify over the parameters that figure in the index that represents the contextual information. Indeed, this is how Kaplan's (1989) semantics of indexicals and demonstratives works. Conventionalism attempts to extend Kaplan's approach to a large amount of expressions. Consider the sentence 'I'll fix the car soon' in one of Korta & Perry's (2007b: 101) examples. Semantics returns the truth-conditions that the speaker of the utterance will fix the car she is referring to with 'the car' within the length of time that counts as upper bound of what counts as 'soon' according to her intention. 2010; Sperber & Wilson 1986) 8 that converge on semantic underdetermination, although they diverge with respect to many other important theoretical aspects and purposes. Semantic underdetermination says that the conventional meanings of many expressions (perhaps all, according to radical versions of contextualism, see Travis 2008, Unnsteinsson 2014) underdetermine their semantic contents. According to contextualists, for many sentences the composition of the conventional meanings of the expressions occurring in them returns at most semantic schemata and only with the help of contextual supplementations full propositional contents with determinate truth conditions are obtained. Moreover, even when the composition of conventional meanings returns a full propositional content, in most cases that content is not what the speaker intends to say with the utterance. Central for contextualists is the difference between the context-dependence of overt context sensitive expressions in the Basic Set and the context-dependence of semantically underdetermined expressions. The context-dependence of overt context sensitive expressions is linguistically governed by conventions of language, and it is codified in their conventional meanings as in Kaplanesque characters. The context-dependence of semantically underdetermined expressions is not codified in conventions of language and is the result of pragmatic supplementations that are not linguistically governed.
Contextualists need not deny that for some expressions the conventional meaning determines a literal content. As I explain in the next section with respect to the conjunction 'and', this is in part the reason why Borg's argument is inconclusive. Yet, according to contextualists, there is no theoretical point in positing a level of composition of literal contents, if any, independent from pragmatic modulations. For example, even granting the honorific title of literal content to the truth functional meaning for the conjunction 'and', the fact that such content enters the propositional content of an utterance is the result of a level of composition of truth conditions that involves pragmatic modulations. The pragmatically modulated composition can operate with the literal content of an expression, if any (e.g. the truth functional content of 'and') or with a pragmatic supplementation of it (e.g. the temporally or the causally enriched content of 'and').
Finally, I contrast contextualism with conventionalism 9 . Conventionalism (Devitt 2021; Korta & Perry 2007a, 2007bLepore & Stone 2015;Perry 2007) covers theories that explain away semantic underdetermination in terms of a linguistic phenomenon similar to the context dependence of overt context sensitive expressions. The idea is that for most expressions the contextual supplementations that enter the truth conditional contents of utterances are linguistically governed by conventions of language that are codified in conventional meanings like Kaplanesque characters for indexicals and demonstratives. 10 8 Bach (1994Bach ( , 2001 can be counted as a contextualist with respect to implicitures, and as a minimalist with respect to literal semantic contents. Notice, however, that Bach's semantic contents mostly fall short of full propositionality. 9 I am grateful to an anonymous reviewer for suggesting this label to me. 10 Semantics captures conventions of language by Kaplanesque characters. For some expressions, Kaplanesque characters do not return denotations, but rules of use. For instance, the character of 'I' does not return a denotation but a rule that says that 'I' is used to refer to the speaker. Of course, once the identity of the speaker is given, the character fixes the denotation of the utterance of 'I'. But fixing the identity Let me explain the point with an example with 'ready'. When speakers use the expression 'ready', they take on the semantic burden of making it explicit what someone or something is said to be ready for. The commitment to provide a completion is independent of communicative intentions. It stems from the speakers' locutionary intention 11 to use 'ready' in accord with its conventional meaning, and it belongs to an expressive level, not to a communicative level. If speaker A utters 'Helga is not ready', A cannot dismiss a request for being more explicit. If hearer B asks 'What for?', A cannot answer 'I have no idea'. 12 The following dialogue is odd: A: Helga is not ready. B: What for? A: I have no idea. The oddity is not simply a matter of lack of cooperativeness. The response 'I have no idea' reveals incompetence on the use of 'ready'. B takes that response as linguistically unacceptable and remains in the dark about what A said. If it were a matter of lack of cooperativeness and not a misuse of language, B would understand what A said but would remain in the dark about why A said that. Yet, if A responds 'I have no idea', there is no proposition said by A that B can understand. Conventionalism introduces a level of content (I will borrow Korta and Perry's terminology and call it locutionary content) that is distinct from (i) what the speaker intends to say-in the Gricean sense of saying that requires audience-oriented intentions-and (ii) what the hearer thinks the speaker intends to say. Some scholars, notably Neale (2005Neale ( , 2016, claim that there is no theoretical need for a level of utterance content distinct from (i) what the speaker intends to say and (ii) what the hearer thinks the speaker intends to say. When (i) and (ii) coincide, there is no theoretical role for a notion of what is said upon which (i) and (ii) are supposed to converge. And when (i) and (ii) are distinct, there is no theoretical point at all in bestowing the honorific title of 'what is said' on a different kind of utterance content (Neale 2005: 182). 13 I will explain that the level of utterance contents introduced by conventionalism covers the theoretical role that Borg thinks is covered by minimal contents. After all, Borg is right on one important point: in order to capture the distinction between lying and misleading, at least in those cases in which speakers can defend themselves from the charge with lying by denying responsibility for the disbelieved information, we need a kind of linguistic liability and a kind of contents that are independent from communicative intentions. I argue that conventionalism provides that kind of linguistic commitment and that kind of utterance contents. 14 of the speaker is not a task for semantics. Kaplan (1999: 3) brings to light the potential of his analysis of indexicals and demonstratives in line with the formalist tradition for developing a semantics of use that, for some expressions, does not ask after worldly meanings but after rules of use. 11 Readers who dislike talking of intentions can replace 'intentions' with 'intentional states'.

Borg's argument is inconclusive
Borg argues that minimalism is necessary to explain the intuition that John is not lying because contextualism fails to explain it. The content for which John is responsible is the truth functional content that Jack got married and had two children in one order or the other. Borg says that this truth functional content cannot be an explicature, because it is the temporally ordered content that first reaches the relevance threshold of the hearer. Contextualists cannot maintain that the intuition that John is not lying tracks an explicature with the truth functional content without running counter to their own account of explicatures. Borg's argument can be reconstructed as follows: 1. The intuition that John is not lying tracks the truth functional content. 2. Contextualism cannot explain such intuition because contextualism (i) predicts that the explicature in John's example is the temporally ordered content and (ii) affirms that explicatures are what is said. 3. If contextualism cannot explain the intuition that John is not lying, then only minimalism can. Therefore: 4. Minimalism plays a substantive theoretical role.
I grant premise 1. Yet, premise 2 is controversial and premise 3 is false.
Premise 2. Borg is too rash to affirm that the explicature in John's example is the temporally ordered content that Jack got married before having two children. Borg's reason is that the temporally ordered content first crosses the relevance threshold of the hearer. In doing so, Borg plays with too naïve a reconstruction of the contextualist picture.
The notion of explicature is often defined in terms of the content the hearer figures out as what the speaker says. Representative of this approach is Recanati's Availability Principle, which says that what is said must be intuitively accessible to the participants to the conversation. However, several authors (Bach 2005;Devitt 2021;Neale 2016) have argued that the metaphysical question of the determination of what is said must be kept distinct from the epistemological question of how the hearer figures out what the speaker says. What the hearer does or is expected to do is not determinative of what is said. What determines that an utterance has a given content are (i) the fact that the speaker participates in a set of linguistic conventions, (ii) the fact that the speaker has in mind certain contextual values that the meanings of the uttered words demand to be filled in the content, and (iii) the fact that the speaker adds some free pragmatic adjustments.
Contextualists need not deny that some expressions possess a literal content. Contextualists might identify the literal content of 'and' with the truth functional conjunction. For example, Recanati (2010: 45) defines literalness as a special case of modulation. Modulation is a pragmatic process that maps the content of an expression onto another content and the content so modulated enters the composition of truth conditions. Literalness is defined as zero modulation that takes the literal content of an expression, if any, and returns the same content. The conjunction 'and' might have a literal content, the truth functional one, which depending on the context is modulated onto the temporally or the causally enriched content, but it might also be modulated onto itself (zero modulation).
This picture provides contextualists 15 with the resources to cope with John's example. Contextualists can maintain that John uses the conjunction 'and' with its literal content. What John says is that Jack got married and had two children in one order or the other. The fact that John expects the rich Catholic to understand that Jack got married before having two children and the fact that the rich Catholic does interpret John's utterance that way have nothing to do with the determination of what John said. John's communicative strategy is tricky because he expresses the truth functional content, although he expects the rich Catholic to understand the temporally ordered content. To say, as Borg does, that for contextualists John is committed to the temporally enriched content because that is the content that the rich Catholic understands is to conflate the metaphysical determination of what is said with the epistemology of understanding.
There is one important lesson, though. The notion of saying that is involved in the claim that John says that Jack got married and had two children in one order or the other cannot be the Gricean notion of saying. According to Grice, saying something entails meaning it, and speakers cannot mean something and expect that the audience will not understand it. It is a requirement for speakers to mean a certain content that they expect the audience to recognize their intention to communicate that content. Yet, as Saul (2012: 53) points out, in many cases misleaders do not mean what they say. They do not intend their audience to focus on what they say. Rather, they intend the audience to be misled into believing something different from what they say. John expects the rich Catholic to understand the temporally enriched content that Jack got married before having two children, but he is not linguistically committed to that content.
There is another important lesson. The distinction between the metaphysical determination of what is said and the epistemology of understanding is not sufficient for responding to Borg's argument. It is necessary to keep the speaker's intentions that are determinative of what is said apart from communicative intentions. Here I have in mind a notion of utterance content inspired by Korta andPerry's (2007a, 2007b) locutionary content and Devitt's (2021) what is said. Korta & Perry (2007a: 172) develop the notion of locutionary content as a kind of utterance content that is determined by the following intentions of the speaker (assuming English is the language): Second, there are important connections between Devitt's properties (i) to (iii) and Korta and Perry's (a) to (g). Devitt's properties arising from linguistic conventions (i) correspond to Korta and Perry's (a) and (b). Devitt's disambiguation (ii) covers to a large extent Korta and Perry's (b), (c), and (d). Devitt's saturation (iii) covers to a large extent Korta and Perry's (e), (f), and (g).
Devitt holds that the meaning of, for example, quantifiers, genitives, 'it is raining', and 'ready' demands saturation in context by an implicit reference the speaker has in mind. The linguistic conventions that govern the use of those expressions demand 16 Korta and Perry do not call 'semantic' their locutionary contents because on their view the output of semantics are token-reflexive contents, whereas Devitt calls 'semantic' his what is said because it is governed by rules of language, mainly established by conventions. 17 Devitt speaks of intentional states instead of intentions.
that the speaker has in mind a particular restriction of the domain of quantification for a quantifier, a particular relation for a genitive, a particular location where it is said to be raining, a particular course of action for which someone or something is said to be ready. Devitt says that the conventionally established meaning of those expressions has an implicit slot to be filled. Although Perry dislikes the term 'saturation', his idea (subscribed by Korta) of unarticulated constituents is quite close (and precedent) to Devitt's idea of slots to be filled. Perry (2007) is a very helpful text where to look for evidence for conceptual connections between Perry's unarticulated constituents and Devitt's slots to be filled. Just to give a couple of examples, Perry (2007: 553) says that the presence of an unarticulated location in the content of 'it is raining' is demanded by the conventions that govern the use of that expression. Devitt has the same view about weather reports. In the same vein, Perry (2007: 558) says that the rule that governs the meaning of 'late' demands that the speaker has in mind a particular event for which someone or something is said to be late. This is the same view that Devitt holds with respect to 'ready' and 'enough'. The central point is that conventions of language commit speakers to providing unarticulated constituents (Perry) or to filling slots (Devitt), and commit speakers to contents for their utterances independently of communicative intentions.
One important idea that Korta, Perry, and Devitt share is the rejection of the view that Devitt calls 'the tyranny of syntax' (Devitt 2021: 216). Neither Perry's unarticulated constituents nor Devitt's slots to be filled are explained by positing elements in the syntactic structure or logical form that match them. Korta, Perry, and Devitt reject commitment to the Isomorphic Principle according to which there must be a correspondence between the structure of a sentence and the structure of the content it expresses. 18 Perry's unarticulated constituents and Devitt's slots to be filled are to be found in linguistic rules, which are posited to provide the best explanation of regularities of use. Perry (2007: 550) says that linguistic rules are registered in the 'lexicon'. The lexicon Perry has in mind is a dictionary that supplies the information an artificial system would need in order to parse and use natural language sentences. That is the kind of information that is (implicitly) used by humans and needs to be made explicit for artificial systems. The lexicon registers thematic rules, like the one that demands a location for 'it is raining', and permissive rules, like the one that says that 'it is raining' can be used to say of a location that it has many of a number of properties: from sprinkling to monsoon-like rain. Perry (2007: 555) says that most of the conventions that govern words are permissive and permit to use them to stand for a number of things, often related to each other. It is one's intention that determines which thing one uses the word to stand for. Perry's thematic rules correspond to Devitt's rules that demand slots filling, and Perry's permissive rules correspond to Devitt's conventions that specify a plurality of related meanings governing expressions 19 , which Devitt (2021: 203) pushes to the extent of putting in doubt the existence of most generalized conversational implicatures as a pragmatic phenomenon (e.g. the temporal meaning of 'and').
There are important philosophical differences between Korta and Perry's view and Devitt's one, and I do not want to oversimplify them. The point to be stressed is that they recognize a level of utterance content that is determined by the fact that speakers exploit linguistic conventions and undertake the burden of filling the slots that those conventions demand to be filled. Linguistic conventions and what they demand generate a form of linguistic liability of speakers for the contents of their utterances that is independent from communicative intentions. This form of linguistic liability for utterance contents is independent from what hearers do or are expected to do, from their actual or hypothetical mental states, and from actual or hypothetical effects on them. Henceforth I will adopt Korta and Perry's expression locutionary content for such a kind of utterance content. As said in the introduction, I call Conventionalism the view that recognises the fundamental role of conventions of language in the constitution of utterance contents that are independent of communicative intentions.
Borg is right on one very important point after all. In order to capture intuitions on the lying/misleading distinction (at least in those cases where speakers retain deniability for disbelieved information) we need a notion of linguistic liability according to which speakers are responsible for expressing contents that need not be meant in the Gricean sense. Minimal contents satisfy this constraint, but they are too minimal. I will argue that we need conventions of language that commit speakers to locutionary contents for their utterances.
Premise 3. Even granting the failure of contextualism, conventionalism provides an alternative explanation of why John is not lying. Some philosophers treat certain phenomena as semantic rather than pragmatic. In particular, some generalised implicatures are treated as cases of ambiguity (Strawson 1952; Lepore & Stone 2015; Devitt 2021). On this approach, the conjunction 'and' is governed by different conventions specifying different meanings. 20 One is the truth-functional meaning, another is the temporally ordered meaning, and still another might be the causally ordered meaning. People might judge that John is not lying because, if charged with lying, he can retreat to the truth functional meaning of 'and'. John chooses the truth functional meaning and can think in good conscience that he is saying that Jack got married and had two children in one order or the other. Nonetheless John expects that the rich Catholic understands that Jack got married before having two children. Suppose John's moral precepts prohibit him from lying. John can truly think that he is keeping faith with his moral precepts.
John's communicative strategy is tricky just because he can play with the semantic polysemy 21 of 'and'. John might intentionally use the conjunction 'and' with the truth functional meaning and at the same time have the expectation that the rich Catholic 20 See Posner (1980) for some early objections to conventionalism. 21 Here I borrow Devitt's terminology. Devitt uses the terms 'semantic polysemy' and 'polysemous ambiguity' to express his idea that some expressions, for example 'and', have different conventional meanings, and the pragmatic process that intervenes in the assignment of one of those meanings is a process of selection and not a process of modulation. will misunderstand his utterance with the temporal meaning. John plans his misleading strategy relying precisely on the semantic polysemy of 'and'. 22

Controversial cases
Borg's argument rests on the analysis of John's example with the conjunction 'and'. With respect to the lying/misleading distinction Borg discusses no other example with different expressions. The debate between minimalists and their opponents focuses on expressions like 'ready', 'enough', quantified noun phrases, etc. Borg's line of reasoning would be much stronger if she could offer some uncontroversial cases with those expressions.
I argue that it is not easy for Borg to provide uncontroversial cases with expressions like 'ready', 'enough', and quantified noun phrases. There is a tension between Borg's account of the minimal contents of those expressions and her view that minimal contents serve to capture intuitions on lying and misleading. To see why, it is helpful to recall Borg's account of the minimal contents of expressions like 'ready', 'enough', and quantified noun phrases.
Borg (2012: 92-102) argues that 'ready' is lexically marked with two argumentplaces and denotes the relation of readiness. When only one argument-place is filled at the surface level, the other is marked by an existentially bound variable in the logical form. For example, in a context where the property of being ready to join the fire service is salient, the sentence 'Al is ready' literally expresses the proposition that Al is ready for something not that Al is ready to join the fire service. The same account holds for 'enough'. In a context where Al has had enough beer, the sentence 'Al has had enough' expresses the minimal content that Al has had enough of something, although the speaker might convey that Al has had enough beer. As to quantified noun phrases, minimal contents return unrestricted interpretations. The sentence 'there are no beers' might be uttered to convey that there are no beers left in the house. Yet, its minimal content is that there are no beers simpliciter.
Consider now the following examples, which I borrow from Saul (2012). 23 22 One might object that there is no real conflict between minimalism and the view that 'and' is ambiguous, given that minimalists accept the existence of ambiguous expressions. But there is a conflict between the motivations that underlie the two frameworks. According to minimalism, semantics concerns primarily the contents of sentences, not of speech acts, and differences of contents of utterances are to be explained in pragmatic terms. Ambiguity theories inspired by conventionalism, instead, are developed inside a conception of semantics that is interested in contents of speech acts and tries to explain differences in contents in terms of conventional meanings by strengthening the role of ambiguity and saturation and weakening the role of pragmatic modulation. In the same vein, one might object that there is no real conflict between minimalism and contextualism, given that contextualists might accept that the truth functional content of 'and' deserves the honorific title of literal content. Again, the difference regards the motivations that underlie the two frameworks, since contextualism, like conventionalism, focuses on speech acts contents, although it explains differences in contents in terms of semantic underdetermination instead of conventions of language. That the point is substantive with respect to the distinction between lying and misleading becomes clear as soon as we focus on expressions like 'ready', 'enough', and quantifiers, as I explain in the following section. 23 I assume that the intuitions in the following examples are uncontroversial. However, if some readers have different intuitions, they can enrich the scenario up to a point where the speaker retains deniability Case A. Mary wants to sabotage Helga's friendship with Iggy. Iggy shows up at the house and asks for Helga. Mary utters (2) with the intent to deceive Iggy that Helga is not ready to go out with him, which Mary knows to be false.
(2) Helga is not ready. Saul says that Mary has lied. Case B.
Dave is lying in bed, and two nurses are discussing the treatment he needs. Ed holds up a bottle of heart medicine, points at it just in front of Fred, and utters (3): (3) Has Dave had enough? Fred replies with (4): (4) Dave has had enough. Fred plans Dave's death by denying him his heart medicine. When he uttered (4), Fred meant something like (4*), which he knew to be false.
(4*) Dave's had enough heart medicine. Saul says that Fred has lied. Case C. Saul discusses Bill Clinton's utterance: (5) There is no improper relationship. 24 Saul proposes this case as an example of misleading. She says that (5) is a carefully worded denial, designed to say something true (that there was no improper relationship between Monica Lewinsky and Bill Clinton at the time of the utterance) and to convey something false (that there had never been an improper relationship between Monica Lewinsky and Bill Clinton before the time of the utterance).
These cases are problematic for Borg's minimalism. Saul says that in cases A and B the intuition is that speakers lied. Borg's minimalism predicts that the minimal contents expressed by (2) and (4) are trivially true. (2) expresses the minimal content that Helga is not ready for something, and (4) expresses the minimal content that Dave has had enough of something. If Borg's analysis is taken to provide necessary and sufficient conditions for lying, her analysis and her theory of the minimal content of 'ready' and 'enough' predict that Mary and Fred did not lie. It is at least controversial that these examples are not cases of lying. One might reply that this is a clash of intuitions. Yet, the point remains that we are given no uncontroversial case involving expressions like 'ready' and 'enough', and we are not told how their minimal contents are tracked in intuitions on lying and misleading.
Borg (2019: n 18) hedges her view and acknowledges that her analysis might provide only sufficient conditions for lying, and there might be cases of lying that involve pragmatically enriched contents. Borg might then admit that cases A and B are genuine cases of lying, in which intuitions track pragmatically enriched contents. Yet, this manoeuvre jeopardizes Borg's dialectical position. Not only we are still for the disbelieved information and is allowed to retreat to the locutionary content in order to avoid the charge with lying. 24 We are invited to imagine Clinton making that statement in the course of a speech that is not an answer to a question like 'Was there ever an improper relationship between Monica Lewinsky and you?'. Otherwise one might treat Clinton's utterance as a case of ellipsis. waiting for clear cases involving expressions like 'ready' and 'enough'. But now we are given clear cases in which intuitions on lying and misleading do not track minimal contents.
Case C is subtler. Saul argues that Clinton was not lying because he said something he knew to be true, although he tried to convey something he knew to be false. Borg agrees with Saul that Clinton was not lying. Yet, Borg defends the view that what Clinton literally said is that there is no improper relationship simpliciter. Given that it is trivially false that there is no improper relationship simpliciter, Borg explains the intuition the Clinton was not lying by saying that Clinton semantically expressed a trivial falsehood and tried to convey another falsehood. According to Borg, then, Clinton was engaged in an act of deceiving, not of misleading. The minimal content that there is no improper relationship simpliciter serves to capture the intuition that Clinton was not lying. Since Clinton did not intend to convey that content and deceive his audience with that content, Clinton did not lie.
However, Borg cannot avoid the difficulty. One might ask whether Clinton would have lied, had he said (6): (6) There has never been any improper relationship. Saul's intuition is that Clinton would have lied. If Clinton had uttered (6), he would have said that there had never been any improper relationship between Monica Lewinsky and himself, which he knew to be false. Borg ought to say that Clinton would not have lied, if she maintains that it is the minimal content of (6) that matters. (6) expresses the minimal content that there has never been any improper relationship simpliciter, which is trivially false as much as the minimal content of (5). But Clinton would have said that trivial falsehood with the intent to convey that there had never been any improper relationship between Monica Lewinsky and himself, another falsehood. As in (5), on Borg's view Clinton would have performed an act of deceiving, not of lying.
It is very controversial that Clinton would not have lied, had he uttered (6) instead of (5). Borg might respond that with (6) matters are different and it is not the minimal content that counts for intuitions of lying but the pragmatically enriched content. But in this response, Borg would owe us a principled explanation for treating (6) differently than (5).

Lying and misleading
I argue that if what drives intuitions on lying and misleading in cases like John's is that which seemingly drives Borg's intuition that John is not lying, then intuitions on lying and misleading in that kind of cases do not track minimal contents.
Borg incidentally gives a hint of what drives her intuition that John is not lying. Here is the relevant passage from Borg's paper: Imagine that I'm talking to a rich fundamentalist who is considering leaving her fortune to Jack, so long as he has lived his life fully in line with Christian teaching. Wishing him to get the money but knowing that he had his children out of wedlock, it seems that I can utter: Jack got married and had two children.
I fully expect that my utterance will lead the rich fundamentalist to believe that Jack got married and then had children, but nevertheless if challenged it seems I can protest that this is not what I said. My utterance is clearly misleading, and intended to be misleading, but intuitively it seems it is not a lie. (Borg 2019: 522).
Borg suggests that the intuition that the speaker is not lying prevails over the intuition that the speaker is lying because the speaker can protest that the false proposition that the hearer understood is not what was said.
The key point is to understand the sense of 'can protest' and the sense of 'say' in Borg's passage. Notice that Borg is here hinting at what drives the intuitions of laypeople. It is as if Borg were reporting the answer that she expected laypeople would give to the question on why they have the intuition that John is not lying. It should be obvious, then, that the sense of 'can protest' and the sense of 'say' cannot be the theoretical notions of the experts in philosophy of language. For laypeople do not master those notions. Arguably, from the point of view of laypeople, that the speaker can protest that the content the hearer understood is not what was said means that the speaker is not forced to retract her utterance and can claim that the content understood by the audience does not correspond to the content asserted. This means that, in the above example, the speaker can retreat to the truth functional content and commit her utterance to that content. The speaker can leave her utterance in place as a move in the discourse that retains its force of assertion with the truth functional content.
My conjecture is that in typical 25 cases of misleading it is the possibility to retreat to the locutionary content and to commit one's utterance to it that people track in their intuitions. People judge that the speaker is misleading and not lying when they allow for the speaker to retreat to the locutionary content, to commit her previous assertion to that content, and to discharge responsibility for the disbelieved information understood by the addressee.
In John's example, John cleverly expresses the truth functional content, which he knows to be true, with the expectation that the audience understands the temporally enriched content, which he knows to be false. Faced with accusation of lying, John can retreat to the truth functional content and proclaim that that content is what he asserted. To the extent that people judge that nothing in the envisaged discourse prevents John from retreating to that content as the content of his assertion, they have the intuition that John is not lying. 25 There might be complicated cases of misleading. Stokke (2018: 121) presents the following example. Marion intends to make Mick falsely believe that Tim's baby is female.Mick. Have you seen Tim's baby yet?Marion. Yes, she is lovely.Marion's use of the feminine pronoun has the effect of making it common ground that the baby is female, which she knows to be false. Marion does not retain deniability for the disbelieved information that Tim's baby is female. This case is intricate because not only it involves misleading with a presupposition but it also involves a misuse of language. Marion knows that Tim's baby is male, so she is intentionally breaking the linguistic convention governing the use of 'she'. Strictly speaking Marion's utterance has no locutionary content. If one accepts Stokke's view that this is a case of misleading, one might limit my claim to plain cases of misleading in which there are no linguistic errors or misuses of language. Notice that according to some commitment-based definitions of lying (Viebahn 2017(Viebahn , 2020, it is possible to lie with presuppositions and deniability of disbelieved information is a necessary condition for misleading. On that approach, this is a case of lying. I do not take side in the debate between says-based and commitment-based definitions of lying. My point is that minimal contents are not the contents which misleaders are taken to be responsible for when they deny disbelieved information for discharging accusations of lying. It is important that the misleader is not forced to retract her utterance in the face of an accusation of lying. The intent of the clever misleader is to avoid the charge with lying without being irrational, or linguistically incompetent, or silly, and in some important contexts (e.g. courtroom testimonies, official speeches, public interviews, etc.) avoiding irony, figurative use of language, and non-literalness. To this end, the utterance of the misleader needs to be carefully worded for expressing a content that can be claimed to be the content of the utterance without violating rationality and intelligibility. 26 It is also important that retreating to the locutionary content is distinct not only from retraction of assertion but also from reassignment of content. By retreating to the locutionary content, the misleader makes it clear that what was asserted is the locutionary content.
My conjecture is that laypeople track contents that satisfy this constraint in intuitions that speakers are not lying when speakers retain deniability for the disbelieved information. After all, laypeople master strategies for misleading without lying. It is a plausible psychological hypothesis that in their intuitions laypeople track contents that they themselves would have expressed for the purpose of misleading without lying.
If my conjecture is correct, then Borg's minimal contents are not suited for being so tracked. Consider the Clinton example. Borg says that her account predicts that Clinton is not lying. The prediction is in line with intuitions, but the account is wrong. For Clinton's misleading strategy to be careful and clever, it is necessary that Clinton's utterance be rational, intelligible, and appropriate to an official speech of the President of USA. Clinton's utterance need express a content that it is rational, intelligible, and appropriate for Clinton to assert. It is not rational, intelligible, and appropriate for Clinton to say that he asserted that there is no improper relationship simpliciter. The following dialogue in which Clinton tries to avoid the charge with lying by retreating to the minimal content is implausible.
Case D.
After the press conference, the question time starts. Journalist: Mr President, you said that there was no improper relationship between you and Miss Lewinsky. We have got proofs that there was an improper relationship between you and Miss Lewinsky. Mr President, you lied.
Clinton: #I did not lie. I said that there is no improper relationship simpliciter. As Borg (2012) acknowledges, one characteristic of minimal contents with existentially bound variables or unrestricted interpretation of quantifiers like, say, that Helga is not ready for something, that there is no improper relationship simpliciter, is that they are either trivially true or trivially false. This characteristic makes it irrational, unintelligible, and inappropriate to assert minimal contents in almost all discourses. 27 This is not to deny that there are contexts in which the speaker might 26 The retreat to locutionary contents reveals that the misleader was uncooperative, but not unintelligible or irrational. See Stokke (2018: 84-87, 112). See Kissine (2013: 89-92) for some considerations according to which uncooperative misleaders keep faith to the evolutionary function of assertion because what they directly and primarily assert is true. 27 In light of some accounts of assertion (and constative speech acts in general), speakers cannot make assertions when the content or its negation is already part of the common set. Contents that are trivially true or trivially false are already part of the common set in most conversational contexts (See Kissine 2013: intelligibly assert, say, that Helga is not ready for something. Think of a philosophical talk on the metaphysics of readiness. Notice, however, as Bach (1994) pointed out long ago, that when speakers intend to assert that Helga is not ready for something they do not utter the sentence 'Helga is not ready', but the sentence 'Helga is not ready for something'.
Borg's account predicts that Clinton is not lying, but at the price of making Clinton's communicative strategy absurd. Moreover, as said in the previous section, there would be no difference had Clinton uttered the sentence 'There was no improper relationship' with the past tense. It seems an immediate explanation of why Clinton chose to utter the sentence 'there is no improper relationship' with the present tense that people do not track minimal contents in their judgments on lying and misleading, and speakers do not rely on minimal contents for planning their misleading strategies. Compare case D with case E.
Case E. Journalist: Mr President, you said that there was no improper relationship between you and Miss Lewinsky. We have got proofs that there was an improper relationship between you and Miss Lewinsky. Mr President, you lied.
Clinton: I did not lie. I said that at present there is no improper relationship between Miss Lewinsky and me.
In E Clinton's defence achieves the result of discharging the accusation of lying. Borg (2019: n 30) mentions the case of a child who says 'I am ready' and then, in order to avoid censure when his mother finds him not to be ready to go to school, he says 'I did not say I was ready to go to school, I said I was ready for something'. Borg proposes this case as one in which the speaker exploits a minimal content.
I do not find Borg's example persuasive. I do not deny the existence of linguistic exchanges like this one. Yet, the child's reply is a joke or an evasion with the purpose of avoiding a quarrel with the mother and it works provided that the mother is in the mood of avoiding the quarrel. The child's reply is not a genuine defence of his assertion. The child is not speaking literally. It is not accurate to describe the child as proposing to uptake the common ground with the information that he is ready for something. Indeed, it is difficult to envisage a similar exchange in contexts where speakers are not allowed to speak non-literally. Consider the following case.
Case F. Fred is on trial for the death of Dave. The witness, Mary, has described the dialogue between Ed and Fred. Mary has said that Ed asked 'has Dave had enough?' pointing at the bottle of the heart medicine, and Fred answered 'Yes, Dave has had enough'.
Prosecutor: You lied. You said that Dave had had enough hearth medicine, with the murderous intent to cause Dave's death.
Fred: #I did not lie. I said that Dave had had enough of something. No court would accept Fred's defence. And if Fred persisted obstinately in this line of defence, he would get sentenced for contempt of court as well as for Dave's murder.
67-76 for a discussion of this point). The view that misleaders can retreat to minimal contents is incompatible with such accounts of assertion.
Consider one last example. Case G. The rich Catholic asks John the following direct question and John gives the following answer: Rich Catholic: Did Jack get married and have two children or did he have two children and get married?
John: Jack got married and had two children.
Suppose Mary is attending this dialogue, knows that Jack had his children before getting married, and intends to sabotage John's plan.
Mary: John is lying, what he said is false. Jack had his children before getting married.
John: #I did not lie. I said that Jack got married and had two children in one order or the other.
In this case retreating to the truth functional content in order to avoid the charge with lying is implausible. Borg might respond that, given the context of the conversation, the speaker is lying with the temporally enriched content. I think this is the right thing to say, but this response weakens Borg's view. Here we have a case in which the same sentence that Borg employs in her argument is used to trigger intuitions that do not track the minimal content.
Case G points out that there are contextual constraints on locutionary intentions. The linguistic formulation of the question constrains the way John can uses 'and'. John cannot retreat to the truth functional content and leave his assertion in place. John might defend himself from the charge with lying claiming that he did not understand the question. I do not dispute that. The point remains that in that case John must retract his assertion. Likewise, the intuitions that both Mary and Fred lied in cases A and B depend on contextual constraints on their locutionary intentions about the use of 'ready' and 'enough'. Consider Fred's case. Ed holds up a bottle of heart medicine, points at it just in front of Fred, and asks 'Has Dave had enough?'. Ed's pointing at the bottle of heart medicine manifests his intention to refer to the heart medicine. If Fred's understands Ed's question, he undertakes the locutionary burden of saturating 'enough' with the hearth medicine. Fred cannot retreat to the minimal content that Dave has had enough of something, but he cannot retreat to the content, say, that Dave has had enough tea either. Faced with accusation of lying Fred might try to say that he misunderstood Ed's question. But then he must retract his assertion.
My point is that locutionary contents are apt to account for the difference between lying and misleading in these cases. Clever misleaders plan their deceiving strategies in such a way that they do not have to retract their assertions. This is not to say that locutionary intentions are not contextually constrained. To the contrary, very often they are contextually constrained, and that explains why certain misleading strategies can be achieved in some contexts but not in others. For example, John can plan his misleading strategy with the ambiguity of 'and' in Borg's example, but he cannot in case G.
In this section I strengthen the case against minimalism by discussing two received says-based definitions of lying by Saul (2012) and Stokke (2018). My purpose is not to take a stand in the philosophical debate on the definition of lying. I discuss Saul's and Stokke's definitions because they directly address the question of what notion of what is said is apt to figure in a says-based definition of lying. My purpose here is to show that the notions of what is said that Saul and Stokke propose are not satisfactory just because in many cases they return (or cannot avoid in a principled way) contents that are identical to Borg's minimal ones. 28 I close this section with some programmatic reflections on conventionalism, locutionary contents, and philosophical definitions of lying.
Here are the definitions of lying by Saul (2012) and Stokke (2018).

Saul's definition.
If the speaker is not the victim of linguistic error/malapropism or using metaphor, hyperbole, or irony, then they lie iff (1) they say that p, (2) they believe p to be false, (3) they take themself to be in a warranting context (Saul 2012: 3).
A lies to B iff there is a proposition p such that (1) A says that p to B, (2) A proposes to make it common ground that p, (3) A believes that p is false (Stokke 2018: 31).
The notion of what is said is central in both definitions. The notions of what is said that Saul and Stokke propose suffer from difficulties related to minimal contents.
Saul's what is said. Saul proposes a linguistically constrained notion of what is said that admits only pragmatic enrichments that are required for full propositionality. 29 Stokke (2013Stokke ( , 2018 argues that Saul's notion of what is said suffers from undergeneration and overgeneration. Case H (undergeneration). The week after the Community Week, during which people were invited to help neighbours out with various tasks, Jasper is having dinner with Doris and is keen to give her a good impression of himself, although Jasper fixed the roof of his own house during the Community Week.
Doris. How did you help out during the Community Week? Jasper. I fixed a roof. Stokke says that Jasper lied. This intuition is captured if Jasper said that he fixed a roof of a house of his neighbours. This interpretation requires a restriction of the quantified noun phrase 'a roof' to the roofs of the houses of neighbours. The difficulty for Saul is that the restriction of the quantified noun phrase is not required for full propositionality. Notice that the content that Jasper fixed a roof-with no restriction-is the minimal content of Jasper's utterance in Borg's account. In general, Saul's account of what is said does not rule out Borg's minimal contents in a principled way, since minimal contents do not require completions for full propositionality. When Saul discusses Borg's minimal contents, her argument for rejecting them rests entirely on intuitions and on what she (2012: 58) takes to be consensus among scholars that expressions like 'ready', 'enough', and quantified noun phrases do not work as Borg says. My conjecture that laypeople track locutionary contents in their judgements on lying and misleading as the contents to which speakers commit their utterances provides a principled reason for rejecting the view that minimal contents play that theoretical role.
Case I (overgeneration). Larry has been asked to design a cover for a book about logic, a subject he does not know anything about. He knows that Norma knows that he has just finished writing a book, although she does not know that it is a book about cats. Larry wants to look attractive to Norma and knows that Norma is interested in logic.
Norma. Do you know a lot about logic? Larry. My book is about logic. Stokke says that Larry's reply is not a lie. If charged with lying, Larry can reply that he said that the book he has been assigned to design a cover for is about logic. Stokke argues that this case raises a problem for Saul. Truth evaluability demands that the possessive 'my' be assigned a contextually specified relation between Larry and the book, and the relation that most plausibly is selected is the relation of being the author of the book. Stokke says that on Saul's view the content said is that the book Larry wrote is about logic. Given that Larry knows it to be false and takes himself to be in a warranting context, Saul is forced to the conclusion that Larry lied.
Stokke's objection, like Borg's argument, overlooks that the contextual value for the possessive 'my' that Larry might have in mind could have nothing to do with what Norma is expected to understand. Larry might express the locutionary content that the book he has been assigned to design a cover for is about logic, although he expects that Norma understands the content that the book he wrote is about logic.
However, the difficulty for Saul is that she cannot avail herself of this response. Saul too overlooks the locutionary notion of what is said, whose constitutive intentions are expressive and not communicative. Indeed, Saul (2012: 37-38) anticipates Borg's argument that, for constrained notions of what is said, in a context in which the addressee understands a conjunction with the temporally enriched content, that is the content that is said. This suggests that Saul (i) assumes that for constrained notions of what is said the saturations and completions that contribute to what is said are guided by communicative intentions and (ii) overlooks the possibility of a plurality of conventions governing the same expression like, for example, the plurality of meanings for the conjunction 'and'. Actually, Saul (2012: 35-36) discusses a case that she considers problematic for constrained notions of what is said.
Case L.
A watches Billy go to the top of the Empire State Building, walk to the edge, and jump into the air. Appalled, A turns and runs away, thinking that she has just watched Billy throw himself off the top of the Empire State Building. A tells B her story, by uttering (7).
(7) Billy went to the top of the Empire State Building and jumped.
By this, A means that Billy threw himself off the edge of the building. B, however, knows that Billy is a performance artist, rehearsing a piece that involves much bouncing up and down on the top of the Empire State Building.
Saul claims that when A learns that Billy was rehearsing a piece she will not take herself to have accidentally said something false. Instead, A will think that what she said was true.
I find Saul's claim implausible. Of course, Saul might reply that this is a clash of intuitions. But the problem is another. English dictionaries distinguish at least two meanings for 'to jump'. One is the meaning of pushing against the ground with your feet and move quickly upwards into the air, and the other is the meaning of jumping from something above the ground for deliberately pushing yourself into the air for falling vertically towards the ground. Thus, when A utters (7) to mean that Billy threw himself off the edge of the building, A is saying that content because she is exploiting a conventional meaning of 'to jump'. The best explanation according to English dictionaries of how A expresses the content that Billy threw himself off the edge of the building is that A expresses that content by using conventional meanings of English words. Moreover, this explains A's commitment to that content.
Stokke's what is said. Stokke (2018: 99) advances an ingenious notion of what is said, developing ideas in Stalnaker (2002) and Roberts (2012). On Stokke's view, what is said by an utterance is a function of the literal meaning of the uttered sentence and the question under discussion (QUD). 30 What is said is the weakest answer to the local QUD that either entails or is entailed by a minimal content that is obtained by the composition, in context, of the lexical meanings of the expressions occurring in the uttered sentence. Consider the following example.
Case M. Same situation as case I, but different dialogue: Norma: What's the topic of the book you wrote? Larry: My book is about logic. Larry lies because he says that the book he wrote is about logic, he knows it to be false, and he proposes that it becomes common ground. The QUD raised by Norma admits of several answers: 'the book Larry wrote is about logic', 'the book Larry wrote is about cats', 'the book Larry wrote is about architecture', etc. The minimal content of the sentence uttered by Larry is either determinate, if the value of the possessive 'my' is determined in the context, or indeterminate across a range of propositions: that the book Larry wrote is about logic, that the book Larry is assigned to design a cover for is about logic, that the book Larry is assigned to read is about logic, etc. The weakest answer to the local QUD that entails a minimal content is the proposition that the book Larry wrote is about logic. On Stokke's account, this content is what is said.
Yet, Stokke's notion of what is said faces a difficulty. Consider case I. Suppose Larry's utterance is indeterminate across a range of minimal contents. None of them is entailed by or entails either the positive or the negative answer to the QUD raised by Norma. The difficulty for Stokke is to specify a content that is said by Larry's utterance. Stokke's (2018: 109) says that when an utterance is indeterminate across a range of minimal contents and cannot be construed as saying anything relative to the QUD, the interpretation takes a generalized proposition as what is said. In Larry's case, the generalized proposition is the following: (8) Larry bears some relation to a book about logic.
In fact, Stokke's generalized propositions are identical to Borg's minimal contents, as it is evident in another example (Stokke 2018: 111-112).
Case N.
Russell is an engineer involved in building a bank. Pamela is a consultant who specialized in materials for high-level security buildings. Russell does not know that Pamela conspires with some bank robbers who plan to break into the bank when it's finished. Pamela thinks steel is strong enough for the roof of the bank, but not for the security booth at the front.
Russell. We're thinking about using steel. What do you think? Pamela. Steel is strong enough. Stokke says that Pamela is not lying, as the continuation of the dialogue shows.
Russell. Right, so you think steel is strong enough for the security booth? Pamela. No, I just meant that steel is strong enough for the roof. On Stokke's account, Pamela avoids lying because what she says is: (9) Steel is strong enough for something. Stokke's generalized proposition in (9) is identical to Borg's minimal content. I argued that minimal contents are not well suited to play the role of what is said that is tracked in intuitions on lying and misleading. They cannot serve the clever misleader's strategy for deception and ordinary speakers lack an intuitive grip on them. It is not rational for Larry to reply to a charge with lying that he said that he bears some relation to a book about logic. And it is not rational for Pamela to reply that she said that steel is strong enough for something. Notice that Stokke himself envisages the continuation of the above dialogue in which Pamela says that she meant that steel is strong enough for the roof. She does not say that she meant that steel is strong enough for something. That would make her assertion unintelligible and her attempt to avoid commitment to lying irrational.
There is another difficulty for Stokke's generalized propositions. Consider case I again. Stokke says that Larry conveys the disbelieved information that he knows a lot about logic through an implicature. Stokke does not give a full account of the rational reconstruction of the implicature, but he claims that the generalized proposition, as what is said, is the input for the derivation of the implicature. I think that his sketch of the rational reconstruction of the implicature is implausible, because generalized propositions are an idle wheel. Stokke (2018: 110) suggests that Larry exploits the fact that Norma believes that he wrote a book about logic and the Maxim of Relation. I agree on this. Yet, the explanation of how Norma comes to believe that Larry wrote a book about logic is that she understands Larry's use of the possessive 'my' as if Larry intended to saturate it with the relation of being the author of the book. She is led to that understanding by the information that Larry has just finished writing a book and by the linguistic convention that the possessive 'my' is in need of saturation. Larry expects that Norma understands his utterance with the content that the book he wrote is about logic. And this content is the input for the derivation of the implicature that Larry knows a lot about logic. Larry avoids lying to the extent that he can retreat to the locutionary content that the book he has been assigned to design a cover for is about logic. In this case, too, we need to keep the locutionary content apart from the content that Norma is expected to understand and exploit as input for the implicature. Stokke's generalized proposition that Larry bears some relation to a book about logic plays no role at all.
Locutionary contents cover the role of contents that misleaders are taken to be responsible for when they retain deniability for disbelieved information and are allowed to discharge accusations of lying. Borg's minimal contents are not apt to this theoretical role. Saul's and Stokke's notions of what is said are not apt to it either because in many cases they yield contents that are identical to Borg's minimal ones. The attempt to find a theoretical role for minimal contents in the explanation of the difference between lying and misleading fails.
This result is relevant to says-based definitions of lying. But it is important also for commitment-based definitions of lying (Marsili 2020; Viebahn 2020, 2021). On commitment-based definitions of lying some kinds of lies require a notion of what is said, because one way of committing oneself to a proposition p is to assert that p by saying that p (see Viebahn 2020: 743). Thus, even if says-based definitions will be superseded by commitment-based definitions, a notion of what is said will remain central for the explanation of the difference between lying and misleading at least in certain important cases.
There is another important point that deserves attention. I argued that conventions of language generate a form of linguistic commitment of speakers to locutionary contents for their utterances. Yet, utterances have a plurality of contents in virtue of the linguistic conventions that speakers exploit. Several scholars accept the distinction between contents that are at issue (what is said), like the locutionary contents in my examples in this paper, and contents that are not at issue, like presuppositions and conventional implicatures. Several scholars argue that illocutionary entailments constitute another kind of utterance contents. The point is that all of them are contents to which speakers are committed in virtue of the linguistic conventions they exploit in their utterances and it is for that reason that they do not retain sincere deniability for those contents. Viebahn (2020) discusses some examples of lying with presuppositions, Stokke (2017) discusses some examples of lying with conventional implicatures, and Marsili (2020) discusses some examples of lying with illocutionary entailments of explicit performatives. Presuppositions, conventional implicatures, and illocutionary entailments of explicit performatives are triggered by conventional meanings of words. It is not implausible, then, to hypothesise that conventions of language are part of the source of the commitment 31 to contents that is demanded 31 Marsili (2020) provides a characterization of commitment in terms of accountability and discursive responsibility. Viebahn (2021) provides a characterization of commitment in terms of justificatory responsibility. Conventionalism might contribute to explain accountability and discursive/justificatory responsibility, at least in cases of literal uses of language, in terms of a linguistic commitment stemming from the exploitation of conventions of language. Some important issues remain open though. For example, whether it is possible to be committed to contents by non-literal uses of language (Viebahn, 2017), or by uses of pictures (Viebahn et al. 2019), or by non verbal actions (Reins & Weigmann 2021).