No matter who: What makes one a relativist?

231 T H E O R I A eISSN 0495-4548 – eISSN 2171-679X No matter who: What makes one a relativist? (No importa quién: ¿qué hace a alguien relativista?) Eduardo Pérez-Navarro* Universidad de Granada ABSTRACT: As part of her argument that relativism and contextualism are nothing but notational variants of each other, Stojanovic holds that contextualism is flexible enough to achieve whatever relativism might do if the matter is what truth-value is assigned to each pair of sentence and context. In this paper, I reply to this statement by arguing that contextualism cannot be made as flexible as relativism without in fact turning it into a version of relativism. The key to my response to Stojanovic is that, while relativism relativizes utterance truth, contextualism does not, so parameters that are not fixed at the context of utterance will be accessible for the relativist, but not for the contextualist. Although the relativity of utterance truth follows as soon as propositional truth is relativized to contexts of assessment, as the relativist does, it is easy to lose sight of this fact if we identify the context of assessment with the assessor’s context. Hence, the point of this paper is that the difference between relativism and contextualism is not one as to whose parameters play a role in determining the sentence’s truth-value. If it were, contextualism could indeed be made just as flexible as relativism. KEYWORDS: relativism; contextualism; truth; context; assessment. RESUMEN: Como parte de su argumentación de que el relativismo y el contextualismo no son más que variantes notacionales el uno del otro, Stojanovic sostiene que el contextualismo es lo suficientemente flexible como para lograr cuanto esté al alcance del relativismo si la cuestión es qué valor de verdad se asigna a cada par de oración y contexto. En este artículo, respondo a esta afirmación argumentando que el contextualismo no puede hacerse tan flexible como el relativismo sin convertirlo, de hecho, en una versión del relativismo. La clave de mi respuesta a Stojanovic es que, mientras que el relativismo relativiza la verdad de la proferencia, el contextualismo no lo hace, por lo que aquellos parámetros que no estén fijados en el contexto de proferencia serán accesibles para el relativista, pero no para el contextualista. Aunque la relatividad de la verdad de la proferencia resulta tan pronto como la verdad proposicional se relativiza a un contexto de evaluación, como hace el relativista, es fácil perder de vista este hecho si identificamos el contexto de evaluación con el contexto del evaluador. Por lo tanto, la tesis de este artículo es que la diferencia entre el relativismo y el contextualismo no concierne al agente cuyos parámetros juegan un papel en la determinación del valor de verdad de la oración. Si lo hiciera, el contextualismo podría efectivamente hacerse tan flexible como el relativismo. PALABRAS CLAVE: relativismo; contextualismo; verdad; contexto; evaluación. * Correspondence to: Eduardo Pérez-Navarro. Departamento de Filosofía I, Facultad de Psicología, Universidad de Granada, Campus Universitario de Cartuja, s/n (18011, Granada, Spain)– [email protected] – https://orcid.org/0000-0002-2240-2380 How to cite: Pérez-Navarro, Eduardo (2021). «No matter who: What makes one a relativist?»; Theoria. An International Journal for Theory, History and Foundations of Science,36(2), 231-242. (https://doi.org/10.1387/theoria.22067). Received: 2020-09-28; Final version: 2021-03-08. ISSN0495-4548 - eISSN2171-679X / © 2021 UPV/EHU This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License Theoria, 2021, 36(2), 231-242 https://doi.org/10.1387/theoria.22067 Eduardo Pérez-Navarro 232 Theoria, 2021, 36/2, 231-242 1. Introduction Attempts to explain the particular behavior of sentences featuring predicates of personal taste have resulted in theories such as contextualism (Glanzberg, 2007; Schaffer, 2011) and relativism (Kölbel, 2004; Lasersohn, 2005; Recanati, 2007; MacFarlane, 2014). Stojanovic (2007, 2012), however, claims that relativism and contextualism are nothing but notational variants of each other. This claim involves two points. First, the two theories predict exactly the same truth-value for each pair of sentence and set of circumstances of evaluation, so relativist and contextualist sentences are intertranslatable (2007, pp.700703). Second, if the matter is what truth-value is assigned to each pair of sentence and context, this is not a semantic issue and, anyway, contextualism is flexible enough to achieve whatever relativism might do (2012, pp. 627-628). While the first of Stojanovic’s points has been contested by Lasersohn (2008, pp. 317-318), the second one has gone relatively unacknowledged. In this paper, I reply to it by arguing that contextualism cannot be made as flexible as relativism without in fact turning it into a version of relativism. Stojanovic is consistent in her use of the label “contextualism”, but she considers different versions of relativism. To avoid confusion, I will refer to one as “nonindexical contextualism” and the other as “assessor relativism”. What Stojanovic calls “contextualism”, for its part, will be referred to as “indexical contextualism” here. Indexical contextualism will be the view according to which a personal taste standard is supplied by the context of utterance and determines part of the proposition expressed (Glanzberg, 2007; Schaffer, 2011). Nonindexical contextualism will also make the standard determined by the context of utterance, but it will include it in the circumstances of evaluation (Kölbel, 2004; Recanati, 2007). Assessor relativism, finally, will locate the standard in the circumstances of evaluation, but make it determined by the context of assessment (Lasersohn, 2005;1 MacFarlane, 2014). Stojanovic’s first point is then that relativism, understood as either nonindexical contextualism or assessor relativism, and contextualism, understood as indexical contextualism, predict exactly the same truth-value for each pair of sentence and set of circum1 Although Lasersohn does not use the words “context of assessment”, some clues can be found in his paper that he is referring to something along these lines when he uses the word “context”. In particular, he says that “the objective facts of the situation of utterance do not uniquely determine a context”, and elaborates: (A)ny concrete situation of utterance will determine as many different contexts in our technical sense as there are individuals—one for each potential judge. We should not limit our choice of judges to individuals who are present in the situation of utterance; John’s utterance of Roller coasters are fun might be true or false relative to Mary, even if she was not present when he spoke. Nor, I think, should we even limit our choice of judges to those individuals who eventually interpret the utterance or assess it for truth; we would not want to say that The chili is tasty is objectively true simply because so few people ever consider the issue that they happen all accidentally to agree. (Lasersohn, 2005, pp. 669-670) As we will see, if propositional truth is not determined by a context that is fixed by the “concrete situation of utterance”, it will be relative to the context of assessment. In a footnote to this passage, Lasersohn claims that these considerations distinguish his proposal from MacFarlane’s, as the latter requires actual assessors (2005, p. 670, n. 18). However, this claim depends on the identification of MacFarlane’s context of assessment with the assessor’s context, against which it is the aim of this paper to argue. https://doi.org/10.1387/theoria.22067 233 No matter who: What makes one a relativist? stances of evaluation. Her second point is that, although it may seem that only assessor relativism can assign different truth-values to the same pair of sentence and context, indexical contextualism (and, by the same token, nonindexical contextualism) can do this too. While Lasersohn’s (2008, pp. 317-318) argument, which will be reviewed later, allows one to reply to Stojanovic’s first point, my aim is to reply to her second one. The key to my response to Stojanovic is that, while relativism relativizes utterance truth, contextualism does not (López de Sa, 2011, p. 108; Shirreff and Weatherson, 2017, pp. 689-690; Ferrari, 2019, p. 481). Thus, parameters that are not fixed at the context of utterance will be accessible for the relativist, but not for the contextualist. The relativity of utterance truth follows as soon as propositional truth is relativized to contexts of assessment, as the relativist does. However, it is easy to lose sight of this fact if we identify the context of assessment with the assessor’s context—if the parameters supplied by the context of assessment are always those of the assessor, contextualism will indeed be able to make them the ones relative to which propositional truth is to be assigned. Hence, the message to take home from this paper is that the difference between relativism and contextualism is not one as to whose parameters play a role in determining the sentence’s truth-value. If it were, contextualism could indeed be made just as flexible as relativism, as Stojanovic claims. The structure of the paper is as follows. In section 2, I introduce the theories under discussion; I present Stojanovic’s claim as it appears in her 2007 paper and, after reviewing Lasersohn’s reply to it, the updated version of Stojanovic’s claim that appears in her 2012 paper. In section 3, I show that relativization of propositional truth to contexts of assessment amounts to relativization of utterance truth, and use this result to reply to Stojanovic’s (2012) claim. As we will see in section 4, however, relativization of propositional truth to contexts of assessment will only amount to relativization of utterance truth if we are clear that the context of assessment is not necessarily the assessor’s context. If the parameters supplied by the context of assessment are not necessarily those of the assessor, contextualism will only be flexible enough if we turn it into a version of relativism. 2. Contextualism, relativism, and the “notational variant” claim In this section, I briefly introduce the views with which this paper will deal—indexical contextualism, nonindexical contextualism, and assessor relativism. As already advanced, when Stojanovic talks about contextualism, she is referring to indexical contextualism; when she talks about relativism, she is referring either to nonindexical contextualism or to assessor relativism. In her 2007 paper, Stojanovic defends that indexical contextualism, on the one hand, and nonindexical contextualism and assessor relativism, on the other, are nothing but notational variants of each other. In her 2012 paper, she focuses on assessor relativism. After introducing the theories, thus, I review Stojanovic’s (2007) argument, Lasersohn’s (2008, pp. 317-318) reply to it, and Stojanovic’s (2012) updated argument. Indexical contextualism, nonindexical contextualism, and relativism have been proposed as theories that allow us to account for different areas of natural language. Following Stojanovic’s own use, here I will focus on what these theories have to say about predicates of personal taste. In particular, let us consider what will happen, according to each theory, when a speaker, let us call her Alice, says “Black pudding is tasty”. According to indexical Eduardo Pérez-Navarro 234 Theoria, 2021, 36/2, 231-242 contextualism (see, for instance, Glanzberg, 2007; Schaffer, 2011), in so doing, Alice expresses a proposition that contains a personal taste standard. What standard this is will depend on the context of utterance. Thus, “Black pudding is tasty” will express one proposition or another depending on the context at which it is uttered. This feature has been said to allow indexical contextualism to explain why we can have faultless disagreements in which it seems that none of the speakers is at fault (Kölbel, 2002, 2004; Wright, 2006). According to nonindexical contextualism (Kölbel, 2004; Recanati, 2007), for its part, when Alice says “Black pudding is tasty” she expresses a proposition that is true or false only with respect to a personal taste standard determined by the context of utterance. This standard is not part of the proposition but part of the circumstances of evaluation, which are the possible sequences of parameters with respect to which we may ask for the truth-value of a proposition (see Kaplan, 1977, p. 502). By claiming this, nonindexical contextualists are able to account for the faultlessness of faultless disagreement while still rendering such exchanges as instances of disagreement. Indexical contextualism has been said not to be able to do this, as it characterizes participants in faultless disagreements as expressing compatible contents.2 Assessor relativism (Lasersohn, 2005;3 MacFarlane, 2014), finally, coincides with nonindexical contextualism in that it takes “Black pudding is tasty”, as uttered by Alice, to express a proposition that is true or false only relative to a personal taste standard. However, assessor relativism takes the standard to be determined not by the context of utterance, but by the context of assessment, i.e., the context from which we wonder whether Alice has said something true or false. MacFarlane takes this feature to make his relativism able to account not only for faultless disagreement but also for retraction—the speech act that one performs in saying “I take that back” or “I retract that” (2014, p. 108). Although indexical contextualism, nonindexical contextualism, and assessor relativism have been defended as alternatives to each other, some authors have claimed that they are not that different. This idea has been most powerfully defended by Stojanovic (2007, 2012), who argues that relativism (which encompasses both the view that I have called “nonindexical contextualism” and the one I have called “assessor relativism”) and contextualism (“indexical contextualism”) are “not much more than notational variants of one another” (2007, p. 691). She does so by defining a semantics for each theory and a bidirectional translation procedure that allows us to predict the truth-value of a relativist sentence (a sentence of the language of nonindexical contextualist or relativist semantics) given that of its (indexical) contextualist version, and vice versa (2007, pp. 700-703). This procedure makes what in a contextualist semantics is a syntactic argument a parameter of the circumstances of evaluation, and what in a relativist semantics is a parameter of the circumstances of evaluation a syntactic argument. The result is that the two alternatives will predict exactly the same truth-value for each pair of sentence and set of circumstances of evaluation. 2 Indexical contextualists have proposed a number of strategies to account for disagreement without incompatibility of contents (see, for instance, Glanzberg, 2007; Sundell, 2011; Huvenes, 2012; Plunkett and Sundell, 2013; Marques, 2014, 2015; Abreu Zavaleta, 2020; for a nice survey, see Stojanovic, 2017). My aim here, however, is only to give a picture of the reasons for indexical contextualism, nonindexical contextualism, and assessor relativism to arise. Thus, it is beyond the scope of the paper to discuss whether these reasons are sound, and, in particular, whether there was any need for contextualism to become nonindexical. 3 See n. 1. https://doi.org/10.1387/theoria.22067 235 No matter who: What makes one a relativist? Stojanovic’s claim has been contested by Lasersohn (2008, pp. 317-318), who holds that hers is not a proper translation—contextualism and relativism will assign different contents to the same sentence in context, unless contextualism fixes the values of pronouns in a way that turns it into a version of relativism. In a Kaplanian framework, a content is first assigned to an expression with respect to a context, and then a truth-value is assigned to that content with respect to an index. Either the values of pronouns are fixed in the step that goes from expressions and contexts to contents, or they are fixed in the step that goes from contents and indices to truth-values. In the former case, the contextualist sentence will be assigned different contents depending on the assignment of values to variables, while the relativist sentence will always receive the same one, so the translation will not be complete. In the latter case, the content of the contextualist sentence will be true or false only relative to an assignment of values to variables, so we will stay with an index instead of an implicit argument. Thus, the only way to make contextualism deliver the same content as relativism is to turn it into a version of the latter. Stojanovic’s claim, to be fair, is that relativism and contextualism are equivalent “from the viewpoint of semantics” (2007, p. 691). She develops this idea in her sequel to her 2007 paper (2012). There, she distinguishes between relativism as a theory that relativizes propositional truth to contexts of utterance (what I have called “nonindexical contextualism”) and relativism as a theory that relativizes it to contexts of assessment (what I have called “assessor relativism”). Relativism in the first sense, she argues just like she did in her 2007 paper, is nothing but a notational variant of contextualism, again understood as indexical contextualism. The novelty of Stojanovic’s (2012) paper lies in arguing against relativism in the second sense as a semantic alternative to contextualism as well. For it is true that indexical contextualism will assign a particular truth-value to any sentence as uttered in a context, while assessor relativism will be compatible with a variety of truth-values for the same sentence-in-context. But what truth-value we assign to a sentence-context pair is not a matter of semantics. Semantics, as Stojanovic understands it, is “the machinery that maps, in a compositional manner, the sentences of a language to truth-values (as a function of appropriate parameters)” (2012, p. 627). Thus understood, assessor relativist semantics and indexical contextualist semantics are equivalent— they will assign the same truth-value to each pair of sentence and sequence of parameters. The only difference will be that indexical contextualists will claim that it is the context of utterance that determines the sequence of parameters relative to which the sentence is to be evaluated, while assessor relativists will allow each context of assessment to determine a different sequence of parameters. This can be put in terms of what Stojanovic calls “adicity-diminishing definitions of truth” (2012, p.627). The indexical contextualist one is this: [[S]] (c) =True* iffdef for all assignments f, [[S]] (c, wc, tc, jc, f) = True where S is a sentence, c is a context, and wc, tc, and jc are respectively the world, time, and judge4 determined by c (2012, p. 627). This adicity-diminishing definition of truth allows 4 Although Stojanovic works with judges instead of personal taste standards, these are equivalent for the purposes of this paper. Eduardo Pérez-Navarro 236 Theoria, 2021, 36/2, 231-242 us to move from truth as a many-place predicate to truth as a two-place one. The assessor relativist definition of truth, by contrast, is this: [[S]] (cU, cA) =True* iffdef for all assignments f, [[S]] (c, wcU, tcU, jcA, f) = True where wcU and tcU are the world and time determined by cU, which is the context of utterance, and jcA is the judge determined by cA, which is the context of assessment (2012, p.628). Here, however, we cannot get as far as to get a two-place truth predicate; instead, we have to satisfy ourselves with a three-place one. At any rate, discussing which adicity-diminishing principle we should embrace is not doing semantics, but what Stojanovic calls, following MacFarlane (2003, 2012), “postsemantics”. So, “bridging principles” (Stojanovic, 2012, p. 629) that give us a sentence’s truth-value at a context of utterance as a function of that sentence’s truth-value relative to a certain sequence of parameters are a matter of postsemantics. Assessor relativism is distinguished from indexical contextualism only with respect to the postsemantics, not with respect to the semantics. A plausible answer to this kind of criticism would be to bite the bullet and just move the whole discussion to the postsemantic level; the question would then be which of these two postsemantic principles accounts better for the linguistic evidence that we have. Stojanovic, however, thinks that such a discussion would make no sense, for it would depend on the assumption that we need some bridging principle or another. This is the kind of assumption that she rejects—which context determines the relevant parameter is not set once and for all, but is a matter of pragmatics. Different conversational settings can make different contexts the ones to look at; sometimes, the relevant personal taste standard, for instance, will be that of the speaker, while in other cases it may be that of the assessor (2012, pp. 631-632). In fact, it may be neither of them, since the absence of a bridging principle allows parameters to take virtually any value, if needed. Stojanovic thinks that this flexibility vindicates her approach. The aim of this paper, though, is to show that contextualism cannot be made as flexible as Stojanovic needs it to be without turning it into a version of relativism. Contextualism is flexible enough to make the assessor’s standard the relevant one, but this will make it as powerful as relativism only if what relativism tells us to take into account is the assessor’s standard. In the next two sections, I show that this is not how relativism should be characterized. Although I find Lasersohn’s argument convincing, in what follows I will proceed as if it were not enough to reply to Stojanovic. I will argue that, even if Lasersohn’s argument did not succeed, contextualism and relativism would still be interestingly different—even if their semantics were equivalent, their postsemantics would not be. In a way, thus, my argument can be seen as adding up to Lasersohn’s. 3. Against the “notational variant” claim The way I have characterized it in this paper, assessor relativism is the view that relativizes propositional truth to a context of assessment. In this section, I show that this move amounts to the relativization of utterance truth as well. This is the kind of thing that will next be shown to stand out of the contextualist’s reach. https://doi.org/10.1387/theoria.22067 237 No matter who: What makes one a relativist? Assessor relativism relativizes propositional truth to contexts of assessment. By relativizing propositional truth to contexts of assessment, assessor relativism also relativizes utterance truth.5 This is so because relativizing propositional truth to contexts of assessment means to claim that propositional truth is a function of sentences, contexts of utterance, and contexts of assessment. In claiming this, the relativist denies that propositional truth is a function of sentences and contexts of utterance alone. Sentences and contexts of utterance together are enough to characterize utterances—there is nothing left to say about an utterance once we have said what sentence has been uttered and at what context. Hence, the assessor relativist denies that propositional truth is a function of utterances, and thus takes more than one truth-value to be compatible with the same utterance. For instance, if Alice’s utterance of the sentence “Black pudding is tasty” expresses a proposition whose truth-value is relative to a context of assessment, the truth of Alice’s utterance itself will be relative—the truth of the utterance will not get settled by looking at the utterance alone. This is the feature that makes assessor relativism interestingly different from both indexical and nonindexical contextualism. Of course, the characterization of assessor relativism as the theory that relativizes utterance truth is not novel. Several authors (see, for instance, López de Sa, 2011, p. 108; Shirreff and Weatherson, 2017, pp. 689-690; Ferrari, 2019, p. 481) have taken this to be assessor relativism’s defining feature, and MacFarlane himself is quite conscious that his characterization of assessor relativism as the view that relativizes propositional truth to contexts of assessment can be shown to be equivalent to this. Next, though, I show that the idea that assessor relativism relativizes utterance truth can be effectively put at the service of replying to Stojanovic’s claim that contextualism can be made to be just as flexible as relativism. Remember where we left Stojanovic’s argumentation. Assessor relativism and indexical contextualism, she said, can be distinguished only by their postsemantics, but not by their semantics. Even if this were so, I think that the fact that two theories predict different truth-values for a single utterance is enough to make them interestingly different, and to make us engage in the discussion about which of them is worth embracing. If this discussion does not belong to semantics, then, as suggested in the previous section, we should 5 A natural worry might arise when assessing the adequacy of this characterization of relativism that there is no such thing as utterance truth—utterances are acts, and as such, they can be said to be correct or incorrect, but not true or false (see Strawson, 1950, p. 130; Bar-Hillel, 1973, p. 304). Although we can talk about utterances’ having truth-values in virtue of their being the acts through which we express propositions that have truth-values in a primary sense (see Davidson, 1990, p. 310), deeper objections can be raised about talk of utterance truth. MacFarlane, who at some point characterizes relativism in these terms (2003), later rejects this move on two separate grounds (2014, pp. 48-49; see also 2008). First, the notion of an utterance is not a semantic notion (see Kaplan, 1977, p. 563), while it would be reasonable to expect the difference between relativism and other proposals to be a semantic one. However, as we saw in the previous section, relativism is not distinguished by its semantics, but by its postsemantics, as MacFarlane himself emphasizes. Second, understanding an utterance as true if and only if the proposition expressed by the uttered sentence at the context of the utterance is true requires for utterances to take place at a unique context, but some would claim that utterances take place in all the contexts that feature possible worlds compatible with such an event (see López de Sa, 2009). However, it feels quite ad hoc to adapt our characterization of relativism to make room for these possibilities. Eduardo Pérez-Navarro 238 Theoria, 2021, 36/2, 231-242 move to the postsemantic level and try to establish which theory explains the postsemantics of our language better. At any rate, Stojanovic has an answer for those who are disposed to move to the postsemantic level, as I said in the previous section—which context is the one in charge of supplying the required parameter is a matter of pragmatics, and some conversational settings may make the assessor’s context the one to look at (2012, pp. 631-632). But I think that Stojanovic’s defense of her approach as more flexible than the assessor relativist one stems from assigning indexical contextualism more flexibility than it is in fact capable of, or from assigning assessor relativism less flexibility than it is in fact capable of. It may well be that some contexts of utterance select the standard supplied by a particular context of assessment as the relevant one. However, they will select it once and forever; thus, it will be enough to look at the sentence and the context of utterance to have a truth-value determined. In other words, the standard supplied by the context of assessment will now be part of the context of utterance. Assessor relativism, by contrast, does not make truth depend on the standard supplied by a particular context of assessment, but on the standard that is relevant at the context of assessment, whichever that might be. Since there is only one context of utterance and infinitely many potential contexts of assessment, assessor relativism will still be much more flexible than contextualism. Indexical contextualism is committed to the claim that utterances are true or false once and forever, that is, to absolutism about utterance truth. I have proved that assessor relativism relativizes utterance truth given that it relativizes propositional truth to contexts of assessment. However, relativization of propositional truth to contexts of assessment will only amount to relativization of utterance truth if we understand contexts of assessment properly. As we will see in the next section, there is an intuitive understanding of the notion that threatens to break this link. 4. The crux of the difference In this section, I distinguish between the context of the assessment and the assessor’s context. Relativism will relativize utterance truth if it relativizes propositional truth to the former, but not if it relativizes it to the latter. Since, as we saw in the previous section, it is the relativization of utterance truth that grants relativism the flexibility it needs, we can only save relativism from Stojanovic’s criticism if we refuse to understand the context of assessment as the assessor’s context. When we say that nonindexical contextualism relativizes propositional truth to contexts of utterance and assessor relativism relativizes it to contexts of assessment, it is tempting to understand the difference between these two theories as a difference as to whose standard plays a role in determining the proposition’s truth-value. If it is the speaker’s standard, we are talking about contextualism; if it is the assessor’s, we are talking about relativism. I think that Stojanovic has this in mind when she claims that relativism and contextualism are nothing but notational variants of each other. However, I think this is misguided, for neither contextualism requires the parameters to be those of the speaker nor relativism requires them to be those of the assessor. For instance, as Recanati points out, the location that the context of utterance supplies as part of the circumstances of evaluation need not be the location at which the speaker is, but “some other place that is being talked about” (2007, p. 6). In other words, it is possible for the context of utterance to https://doi.org/10.1387/theoria.22067 239 No matter who: What makes one a relativist? provide a value for a parameter in the circumstances of evaluation that is different from the one that figures in the index of the context, understood as the index that includes the speaker who made the utterance, the time and location at which it was made, the speaker’s standards, etc.6,7 Likewise, the context of assessment can supply a parameter that is not the assessor’s. To see this, consider the following case. Alice and Beth are at a restaurant and they are looking at the menu, trying to choose a starter that they can share. However, while Alice came to the restaurant thinking about the delicious food they serve, Beth does not really care about food—she eats just to feel her stomach filled, and does not really have culinary preferences. There is nothing she does not like, but there is nothing she especially likes either. All she wanted to do was sharing a meal with her friend Alice, no matter what meal it was. It is against this background that Alice says “Black pudding is tasty” and Alice wonders whether Alice has said something true or false. The context from which Beth does this is the context of assessment. It makes no sense, however, for Beth to assess the taste of black pudding against her personal taste standard, since it will not tell her anything. Thus, the only option left for her is to rely on Alice’s taste standard, and say that Alice has spoken truly only if black pudding is tasty relative to that standard.8 The difference between context of utterance and context of assessment will amount to a difference in the status that utterance truth has in each theory only if the context of assessment is not reducible to the assessor’s context. If these two notions are different enough, contextualism will take utterance truth to be absolute, while relativism will relativize it. In particular, contextualism will take utterance truth to be absolute even if it 6 There is a choice as to how to understand the index of a context involved here. In a broad enough sense, the index of a context would be that composed of the values that the context determines for each parameter, whether they are those associated with the speaker or not. However, the same context could determine, for instance, different personal taste standards as the relevant ones to assess different assertions even by the same speaker. Thus, it is not possible to talk about the index of the context in this sense, as there will be more than one index compatible with the same context. This is why I prefer to talk about the index of the context independently of the values that the context actually determines. 7 It could be argued that, if we allow the context of utterance to supply values other than those related to the speaker, we deprive ourselves of the resources that allow the contextualist to deem sentences such as “I am here now” logically true. If this is so, though, it just means that contextualism is less flexible than Recanati takes it to be, for it requires every parameter in the circumstances of evaluation to be fixed by the speaker for “I am here now” to be logically true. Relativism does not need to be this flexible with respect to indexicals—it can allow for personal taste standards, for instance, other than the speaker’s to get into the circumstances of evaluation while restricting the values that can be supplied by the context of utterance so that “I am here now” remains logically true. 8 This case could be argued to be an example of an exocentric use of “tasty” (Lasersohn, 2005, pp. 670674). However, note that Beth’s evaluation has practical consequences for her, as it will help her decide whether she eats one thing or another. (Compare the truly exocentric use of “tasty” we make when we say that the cat’s food is tasty.) Beth is genuinely trying to decide whether black pudding is tasty or not, not wondering whether it is so according to Alice’s taste. In doing so, though, it is Alice’s standard that she takes into account. Thus, just like the context of utterance, the context of assessment can supply values for the parameters different from the ones in the index of that context. This means that, no matter who the assessor is, the context of assessment can select virtually almost any value for a given parameter.