SciELO - Scientific Electronic Library Online

Home Pagealphabetic serial listing  

Services on Demand




Related links



Print version ISSN 0100-6045On-line version ISSN 2317-630X

Manuscrito vol.38 no.2 Campinas July/Aug. 2015  Epub Oct 06, 2015 




1Philosophy Department, National University at La Plata, Buenos Aires, Argentina,


Why would we argue about taste, norms or morality when we know that these topics are relative to taste preferences, systems of norms or values to which we are committed? Yet, disagreements over these topics are common in our evaluative discourses. I will claim that the motives to discuss rely on our attitudes towards the standard held by the speakers in each domain of discourse, relating different attitudes to different motives -mainly, conviction and correction. These notions of attitudes and motives will allow me to claim that different domains of evaluative discourse have a different distribution of disagreements driven by them.

Keyword: disagreements; evaluative expressions; Lewis; conversational dynamics

Why would we argue about taste, correctness or morality when we know, as competent speakers, that all these topics are relative to the taste preferences, systems of norms or values to which each of us is committed? Yet, disagreements over all these topics are a common part of our evaluative discourses. In this paper I inquire into the issue of what motives we have to engage in discussion about these matters. I will claim that the motives to discuss in each case rely on our attitudes towards the standards held by the speakers in each domain of discourse. I will point to a relation between these motives and two notions of disagreement: while regular disagreements over facts can be understood as first-order disagreements (where a certain semantically expressed content is rejected as false), evaluative disagreements like those mentioned above should be considered second-order disagreements: disagreements where the standard is rejected in a discursive move that I will explain using Lewisian tools. This characterization and the differences found among attitudes and motives to disagree will allow me, at the end, to propose that different domains of evaluative discourse have a different distribution of motives to disagree.

Evaluative disagreements

Humans need to engage in an amazingly wide variety of conversations and discussions in order to live a social life. Among these exchanges, many end up being disagreements. There are plenty of different kinds of disagreement: two people can be said to disagree, in some sense, if they entertain opposite non-cognitive attitudes (say, desires) towards the same object, different beliefs about the same object, or different (and mutually incompatible) plans or practical intentions towards the same object. They can disagree during a conversation, but they can also be said to disagree even if they are miles or years (or worlds!) apart from each other. 1'Disagreement' can be understood both as an activity and as a state. Even when there are as many definitions of disagreement as papers on the topic, it is not contentious to consider that a dispute counts as a disagreement when:

  • Direct intuition : we intuitively understand that one of the parties is rejecting what the other one said.

  • Rejection : one of the speakers utters a contentious answer ('No', 'I disagree', 'Nu-uh') that denies what the other party says and it is felicitous.

Disagreements can be intuitively sorted into in weak and strong: we expect weak disagreements to be resolved easily by resorting to available data, while in strong disagreements the agreement on a description of the relevant facts is not enough reason to make the parties converge. ( Blackburn, 1981) associates weak disagreement with objective topics and strong disagreements to aesthetic, moral or personal taste topics. ( Rott's, 2014) distinction between substantive disputes (where retraction happens) and merely verbal disputes (faultless) also correlates to this divide into weak and strong disagreements (respectively). This apparently natural sorting gestures towards the need to distinguish between two broadly different kinds of disagreements: first- and second-order disagreements, corresponding nicely 2to disagreements on utterances of objective sentences (1) and evaluative sentences (2): 3

    1. . John: The average diameter of planet Earth is 7,925 miles. Mary: No, it's not! It's actually 7,926 miles.

    2. . Jack: Roller coasters are fun. Jill: No, they are not! They are very unpleasant.

Were John presented with accurate astronomical information, he would retract from his utterance. However, no information about roller coasters can force Jack to stop enjoying them. And this is thus because, while objective sentences are true or false relative to the way the world is, evaluative sentences are true or false depending on the way the speaker (or the assessor, according to ( MacFarlane, 2005, 2007, 2014)) valuates the world. Without entering into technicalities, let's just say that speakers valuate the world (objects, events) according to evaluative standards, varying in nature along with the nature of the sentence itself. 4I will use the umbrella expression 'evaluative sentences' for sentences like Jack's in (2), sensitive to different valuations of the world. Although usually sentences containing taste predicates are used as a toy example for the sake of simplicity, these are not the only nor the most relevant cases of evaluative sentences. Expressions with similar context-sensitivity comprise deontic modals, deontic predicates and more generally, normative expressions. Hence, 'evaluative standard' should be also understood as an umbrella term covering orderings of taste preferences, but also dispositions, normative systems and values. In all cases, the utterer of an evaluative sentence asserts it as true according to her own commitment to a particular evaluative standard: Jack's utterance is true relative to his standard of fun, while Jill's is true according to her standard of fun. 5Disagreements over objective sentences are mostly weak, for both parties can end up agreeing when they share the same data and process it in the same way. Disagreements over evaluative sentences are in turn usually strong, for adding more data does not usually lead to a change in the way the world is valued. 6

In first-order disagreements, the rejection is directed against the (semantically expressed) content of one of the utterances. Each party assents or dissents to a common content whose truth does not rely on any special trait of the speaker and both utterances are assigned contradictory truth-values. In turn, in 'second-order' disagreements it is not the (semantically expressed) content what is being rejected, but a different (pragmatically expressed) content, presupposition or foundation of what the speaker says. Because of this, there can be disagreement and rejection even though both sentences are true or it is accepted that each utterance is true relative to each speaker. Experimental results attained by Knobe and Yalcin show that speakers can reject assertions even if they find their content true. Their results concerned sentences containing epistemic modals, but the same results are expected in disagreements involving evaluative expressions. 7Therefore, disagreements over evaluative sentences are not related to their assigned truth conditions.

If not because we find them false, why do we disagree over evaluative sentences? A nice way to understand this is via the ( Lewisian, 1979) understanding of language as a norm-governed game. In this perspective, every assertion is a move within a game that alters its score. The scoreboard keeps track of the changes each move makes in all the components of the game: the common ground (involving the propositions believed to be believed by all the participants), comparison classes, evaluative standards and every other parameter required to assign a truth-value to an utterance and to accept or reject its content. In this game, some moves update the common ground in the score, enriching it by adding a proposition to it or narrowing it by taking one proposition out of it. This is what happens with the utterance (and acceptance) of objective sentences. Another kind of move reinforces or makes changes in the values given at each time to the relevant parameters in the score: this move is previous 8to the content update. Hence, if the sentence Φ is true under the value v of the parameter p in the score at the moment of assertion, the assertion of Φ will be accepted only if (a) v is already the value of p at that time or (b) v is not the value of p, but the parties accommodate 9v as the value of p from that moment. The rejection in disagreements signal the refusal of one of the parties to accept a given value (that under which the assertion would be true) as default for a parameter in the score. Disagreeing over evaluative sentences can be understood, in most cases, as a linguistic move preventing the settlement of a given value for a standard held in the conversation.

For what reason would we want to refuse to establish a value as default? One answer in this line of thought claims that what is rejected is the establishment of a given meaning for the evaluative expression. Evaluative disagreements are hence understood as metalinguistic disputes. ( Plunkett and Sundell, 2013) claim that some disputes are non-canonical , i.e., not targeting the literally expressed content but some pragmatically communicated contents. Among these disputes, those involving context-sensitive expressions are metalinguistic: both parties mean different things by their words, and the main motivation for the dispute is to get to an agreement on a common meaning for the key terms. The value given to the parameter is considered part of the meaning: following ( Kaplan, 1978), meaning can be understood in terms of character and content . According to Plunkett and Sundell, metalinguistic disagreements can affect both: for relative gradable adjectives like 'tall', the character remains stable across all contexts of use, but each use may pick up different properties in different contexts, once a relevant parameter -a threshold along a scale of height or a comparison class, say- has been settled. Hence, a disagreement over the sentence 'Feynman is tall' in a context in which both speakers are aware of Feynman's height, is to be understood as a disagreement over opposing views of what contextually counts as 'tall'. In the same way, a dispute over the sentence 'That chili is spicy' seems more like a disagreement over what should we call 'spicy' rather than a disagreement over the content expressed by the sentence. In both cases, a metalinguistic negotiation ensues aimed to settle a unique contextual threshold for tallness and spiciness respectively. Through their assertions, both speakers advocate pragmatically for the parameters associated with their assertion. Cases of metalinguistic disagreement also can range over the character and not the content of the expression. A dispute over the classification of 'Secretariat' the race dog as an athlete or over tomato as a vegetable can hint to disputes over the stable meaning (character) of these expressions. Again, both utterances in disputes are correct for both speakers are using 'athlete' and 'vegetable' differently, and the disagreement aims to settle what should count as an athlete or a vegetable in that context.

Now, disputes over characters are metalinguistic, but there are fewer instances of metalinguistic disputes as it might seem at first sight. In the tomato case, for example, the experts on the topic have already classified it as a fruit: there is no dispute over the meaning of the words involved -at least no dispute solvable by appealing to arguments based on ways of using tomatoes. The whole dispute is best understood as a case of semantic ignorance (a fruit can be used as a vegetable and continue being a fruit). 10Most content disagreements are cases of second-order disagreements in my account, but I would not though consider them metalinguistic , for (i) the meaning (character) remains stable across contexts 11-what is discussed is the value that a parameter should have-, and (ii) the motivation leading these disputes, as I see it, it is not an interest in agreeing on a common meaning for the expression (or a common value for the parameter). Instead, it is a motivation to convince the other out of his/her perspective. The difference is subtle and both goals can be confused. But there is a difference between starting a negotiation willing to achieve a common end (and therefore being disposed to abandon our own perspective if necessary) and being reluctant to do so. In the first attitude, the value given by the parties to the parameter is merely conditional; in the second, the parameter has a fixed value that the speaker is not willing to jettison. In my understanding of second-order disagreements, they start with this latter attitude.

What motivates us then to reject a previously uttered evaluative sentence? In weak, first-order, objective disagreements, rejection is motivated by the need to correct a mistake. When it comes to statement of facts, same input of data and same inferential processing should provide the same (true) content as output. This strongly ties together the notions of falsity and mistake: considering the same evidence and arriving at a false content evinces a fault either in the gathering or in the processing of it. The main motivation to discuss is consequently to correct these mistakes whenever an utterance is considered false. However, when it comes to strong, second-order evaluative disagreements, we must look elsewhere: 12utterances of evaluative sentences only get relative truth-values: the same evaluative sentence will get different truth-values when uttered by speakers with different evaluative standards. Now, although intuitively, this relativity is known by competent speakers. Add to this the fact that a competent speaker also knows intuitively that -according to the norms of assessment- one should only assert what she thinks is true. If the speaker is a competent user of evaluative terms, she will know then that what her conversational party says is true according to his standards as much as what she says is true according to hers. If truth is thus relative, why should we feel motivated to argue what is relatively true for others? Certainly, the motivation to correct a factual mistake is not what is at stake here.

On this regard MacFarlane has claimed that the motive underlying these discussions is the need to "coordinate contexts". By this he means the search for a common context of assessment, a common evaluative standard or perspective. The bigger the interest in coordinating contexts, the more intense the disposition to discuss:

But what is the point of fostering controversy in 'subjective' domains, if there is no (nonrelative) truth on which both parties can converge? Why shouldn't we just talk about our own tastes, rather than ascribing subjective properties to the objects? Perhaps the point is to bring about agreement by leading our interlocutors into relevantly different contexts of assessment. If you say 'skiing is fun' and I contradict you, it is not because I think that the proposition you asserted is false as assessed by you in your current situation, with the affective attitudes you now have, but because I hope to change these attitudes. Perhaps, then, the point of using controversy-inducing assessment-sensitive vocabulary is to foster coordination of contexts. We have an interest in sharing standards of taste, senses of humor, and epistemic states with those around us. ( MacFarlane, 2007, p. 30)

But the notion of coordination of contexts is a bit obscure, for it does not specify the attitudes the contender should have regarding her own attitudes towards the object in dispute. The contender could be willing to arrive, after negotiation, to a common context or standard different from her initial one. If this is what "coordination of contexts" means, it does not reflect accurately the motivation to discuss in cases in which we are not willing to abandon our initial perspective. As with metalinguistic disputes, coordination-motivated discussions would be impossible without the disposition to relinquish our evaluative standard in favor of a new, commonly built one. And in most evaluative second-order disagreements this disposition is not present: when we aim to correct our contender's commitment to an unacceptable standard or to convince her out her deficient standard, we are not willing to abandon our standards in the discussion. On the other hand, MacFarlane could be referring to cases where the contender is not willing to abandon her own commitments to an evaluative standard but she aims to convince her contender to commit to it as well: in this case the notion of context-coordination needs to be explained with greater detail. In what follows I introduce a model of explanation of the different ways in which we could be motivated to discuss utterances of evaluative sentences that aims to fill this gap.

Attitudes and motives

As said, in objective disagreements we discuss mainly to correct a mistake being made by the speaker who uttered a false sentence. In evaluative discussions, though, things are a bit more complex: even though there are (rare) cases of first-order evaluative disagreements in which the discussion is motivated by the urge to correct, generally we disagree because we think that, even if true relative to the speaker, evaluative sentences express inadmissible valuations of the world or because we think that our perspective on the topic is better. Hence, there are more motivation to discuss than expected. Consider these cases:

    (a). Agreement : British John says 'Marmite is delicious' I concur. I don't feel motivated to discuss.

    (b). Acceptance : Pothead Bill says 'Smoking pot is cool' I don't find it cool myself, but I don't see any harm if others consider it thus. I don't feel motivated to discuss.

    (c). Deficiency : Pothead Dan says 'Smoking pot is cool' I don't find it cool for me, and I don't see any harm if others consider it thus. Even so, I think that drinking wine is cooler, and feel motivated to convince Pothead Dan of it.

    (d). Misapplication : British John says 'Marmite is delicious' I disagree, because I think that according to any food expert in our community, Marmite is not tasty. I feel motivated to correct British John. 13

    (e). Unacceptability : Addict George says 'Doing heroine is cool' I don't find it cool because I think that that is not a way anyone should live. I feel motivated to correct Addict George.

It is obvious for any competent user of evaluative predicates that what is valuated in a certain way may be valuated differently from any other perspective. Even so, everybody can entertain opinions or valuations over other people's valuations of an object or event, say x. The cases above involve the utterance of an evaluative sentence expressing a personal valuation of some x. In all, I take different attitudes towards these valuations. In words of ( Richard, 2008, pp. 129-130), we can find four commonsensical attitudes towards another person's valuing x: agreeing with it (as in (a)), accepting it (as in (b)), finding it deficient (as in (c)) or being intolerant towards it and finding it unacceptable (as in (e)). I add one more case: finding it mistaken through misapplication of a communal standard (as in (d)).

Agreement does not raise any kind of disagreement: it elicits agreement expressions. Misapplication of a standard leads to a disagreement, but it is a first-order disagreement: what British John says is false under his own standard and I correct the content of his utterance. This leaves us with Unacceptability, Deficiency and Acceptance : the three of them constitute the main motivations to disagree. Acceptance can be understood as a mild version of Deficiency that does not lead to an explicit discussion. I will assume then that Deficiency and Unacceptability are the main motives to discuss evaluative topics, with any other motive reducible to them or to a mix of them. Note that these are basic motives: usually there are many and more complex ulterior reasons to discuss joining them: we may want to correct someone to belittle her in front of others; we may want to set ourselves up as figures of wisdom, or maybe we need to prove ourselves more informed. But these ulterior reasons are not relevant to the following argument: at the end, all of them are based in one of the two main, basic attitudes.

The distribution of motivations to attitudes towards valuations is simple: whenever we find that the speaker has committed a mistake, we will feel the urge to correct him. When we deal with first-order evaluative mistakes - misapplications of the standard held by the speaker-, the relevant evaluative standard for the correction is that of the speaker herself: in these cases we usually correct someone when her valuing is not coherent with her past sayings or behavior (as when we say to a kid 'You don't like broccoli!' because he has systematically refused to eat it in the past) or with the values issuing from the standard we take to be the speaker's (say, a shared standard). Second-order evaluative mistakes in turn, happen when we find our contender's evaluative standard not deficient but unacceptable, and embracing it, a mistake that we will want to correct by reasoning the speaker into adopting a new standard. On the other hand, whenever we think that the speaker has not committed any kind of mistake but we still feel that our standard is preferable, we feel the urge to convince her of changing it. Note that considering the standard deficient does not mean nor imply that it is mistaken (nor that we think it is so); we acknowledge her standard but we still think there is a better way of valuing x.

Correcting and convincing can be taken, thus, as the main motivations to discuss, rooted in the different attitudes that valuations of objects or events can trigger in us.

Appropriate (and inappropriate) standards

The close tie between falsity and mistakes helps explain the motivation to discuss in objective disagreements. This puts the concept of truth in a central position when accounting for these disputes: disagreements of this sort are forcefully understood as states of truth-value contradiction between the semantically expressed contents. An input of identical evidence identically processed should result in the output of an identical belief. Any utterance whose content contradicts it must then be false, and the utterer consequently mistaken. With evaluative disagreement it is a bit more complicated, for different attitudes towards the standard used by the speaker lead to a motivation to convince or to correct (a second-order mistake). These disagreements, because they are directed to the standard and not to the content, can (and do) happen even when each party takes the other party's utterance as true. The concept of truth ceases to be central in accounting for these disputes. We need another, more suitable concept. In order to deal with these cases a notion of the appropriateness of evaluative standards in play is needed.

Discussions over evaluative, standard-dependent topics are mostly motivated by the attitudes held by the assessor towards the speaker's standard. Mistakes are not as tightly connected to falsity as to commitment to unacceptable standards. It is not the truth value of the utterance, nor the truth value that the assessor may ascribe to it from her perspective, what motivates the discussion. A different notion, other than truth, is needed: one that helps articulate these attitudes towards someone else's standards or valuations. I propose to consider a notion of appropriateness for standards and valuations not linked to truth values:

  • Given a evaluative standard (ES) and a speaker S,

  • ES is appropriate for S iff S acts consistently with the valuations of objects or events that issue from ES.

Even if we lack a precise idea of what evaluative standards might be, we do know what they do: they rank objects or events in relevant hierarchies. In our example, British John's food taste standard ranks Marmite as delicious over other spreads. The manifestations of this ranking or the valuations issuing from it are twofold: the standard determines a non-cognitive attitude towards the object or event x (rejection or attraction, tastewise) 14that could be potentially articulated into a cognitive attitude or belief. This non-cognitive attitude makes us also lean towards the object in an appropriate way: 15in the case of Marmite, it will move British John to choose it over other spreads whenever he can; holding a particular evaluative standard makes us act in a certain way and avoiding acting in another if possible. A standard is appropriate for us, then, if we are willing to act consistently with the valuations it imposes.

According to this, the standards we are already committed to are appropriate for us: we do in fact act consistently with their valuations. But what about the appropriateness of the standards we are assessing? For those we ponder what we would do in counterfactual situations in which we held them, imagining what those different versions of our actual self would do and measuring them against the self-image we have of our actual selves. Say I don't like Marmite; it is not hard for me to imagine a situation in which I value it positively over any other spread. That alternative version of me acts consistently and eats Marmite whenever possible. I can contemplate that picture from my actual self-image and embrace it as an acceptable version of me: even if I don't actually find Marmite tasty and it is not valued high according to my (appropriate) standard, I can find appropriate the standard valuing it. In the same way, say I do not do drugs. I can imagine a version of my actual self holding a standard which values doing them as cool. I can imagine what it would be like to act accordingly. And I can reject, from my actual self-image, that version of me as undesirable, unacceptable. Therefore:

  • ES is appropriate for S iff S would act consistently with the valuations of objects or events that issue from ES.

The notion of appropriateness involves then appealing to things like actual self-images and acceptable and unacceptable versions of ourselves to cover both the cases in which we act consistently with standards (our own) and cases in which we don't but we would (other standards). It might seem that I am loading the notion too much with metaphysical burden, and maybe I am. But this is just a device as good as any other to account for the fact that we can reject (or approve) a standard different from ours because we reject (or approve) the kind of actions it leads to. Therefore, I will talk about better and worse versions of our actual selves, even about unacceptable versions of ourselves.

There is a reason for not defining appropriateness in terms of cognitive attitudes or beliefs. In most cases, individuals cannot properly articulate their evaluative standards and sometimes cannot even articulate particular valuations issuing from them. Correspondingly, when it comes to know an individual's standard from a third-person perspective, direct epistemic access is impossible. All we have as such is behavior and the ability to infer preferences from it.


Trivially, we all find our evaluative standards appropriate. But cases of evaluative disagreements emerge with the assessment of the standards implicit in utterances made by others. When it comes to it, other people's evaluative standards can be assessed from two different standpoints: either from their perspective or from our own. Let's call the first one external assessment (EA) and the second one internal assessment (IA):

  • For a speaker A committed to an evaluative standard ESA and an assessor B,

  • EA: ESA is appropriate for B iff A would act consistently with the valuations of objects or events that issue from ESA.

  • IA: ESA is appropriate for B iff B would act consistently with the valuations of objects or events that issue from ESA.

EA avoids assessing a standard from a perspective other than the speaker's and therefore, no matter what B's attitudes are towards the standard held by A, it will be considered appropriate (for B) if it is appropriate for A himself. Usually A is an individual agent, but it could also stand for a community. IA allows the assessment of a standard from perspectives different than the speaker's. Mainly, the perspective will be the assessor's itself: A's evaluative standard will be considered appropriate (for B) if it is appropriate for B herself. But B could also stand for a community.

Thus armed, let's take a look at evaluative disagreements. We have singled out two different motivations to discuss: we may either want to correct or to convince our conversational party. These motivations are rooted in our attitudes towards the other party's valuation or standard: we want to say that people are inclined to discuss whenever they find that the speaker's standard is appropriate but misapplied, appropriate but deficient, or not appropriate at all.

IA considers a standard appropriate if the speaker would act consistently with it, disregarding the assessor. Consequently, an evaluative standard is inappropriate if the speaker won't act upon it:

  • Inappropriate standard for IA : ESA is inappropriate for B iff A would not act consistently with the valuations of objects or events that issue from ESA. 16

This internal perspective cannot account for disagreements based on deficiency nor unaccceptability of A's standard: remember that one can find a standard deficient whenever finds it appropriate (an alternative version of self committed to that standard is acceptable) but still considers one's actual standard better. According to IA, B could consider A's standard deficient if A himself would find it appropriate but held a second standard and considered that one better. Since only one standard can be held, comparison from the speaker's internal perspective is not possible. The same goes for unacceptability : a standard is unacceptable when we reject an alternative version of our selves committed to it. From an internal perspective, this means that, for B to find A's standard unacceptable, A would have to reject a version himself committed to it. An assessor could never find the speaker's standard unacceptable nor deficient from his perspective if the speaker acted consistently with it.

EA, in turn, allows the assessment of standards from perspectives other than the agent's. Therefore it can account for cases of deficiency and unacceptability of standards, and correspondingly, for motivations to correct and to convince. According to EA, a standard appropriate for the speaker can still be inappropriate for the assessor:

  • Inappropriate standard for EA : ESA is inappropriate for B iff B would not act consistently with the valuations of objects or events that issue from ESA.

This external perspective accounts perfectly for cases of unacceptable standards: in our case of Unacceptability above, Addict George says 'Doing heroine is cool'. In these cases, the assessor (say, Clean Mary) considers doing heroin is not a way anyone should live and consequently, feels the urge to make Addict George quit heroine by correcting him out of the standard that values it as cool, because she thinks it is a mistake to embrace it: Addict George's standard on coolness is unacceptable because it is completely inappropriate for her. This means that the heroin addict version of Clean Mary is unacceptable for her actual self-image: she would not accept to act as if she considered cool to take heroin consistently. Not rejecting this standard is, hence, a mistake: not a mere misapplication but the commitment to an inappropriate standard, a second-order mistake.

In cases of standard deficiency, the assessor acknowledges the evaluative standard in play and finds acceptable a version of himself that acts accordingly with it. Acceptable but not actual versions of ourselves can be sometimes better and sometimes worse than our actual self: a tidier version of myself is acceptable and definitely better than my actual messy self, and a version of myself that watches TV all day long is also acceptable but worse than my actual self. A standard is therefore deficient whenever it is appropriate for the assessor, but the way of acting it would impose is worse than the actual way of acting; the version of self that it sketches is worse in some respect than the actual self-image of the assessor. Recall our example, where Pothead Dan says 'Smoking pot is cool'. In cases like this the assessor (say, Clean Mary) does not find pot cool herself, but sees no harm in doing it. Even so, she may want to convince him into changing his standard into hers only because she thinks hers is better.

Different motives for different discourses

Deficiency and unacceptability of standards may be presented thus:

  • Deficient standards : ESA is deficient for B iff ESA is appropriate for B but not reflecting her best way of acting.

  • Unacceptable standards : ESA is unacceptable for B iff ESA is inappropriate for B.

Finding a standard deficient leads to try to convince the speaker into a different one (usually, our own). Finding it inappropriate leads to trying to correct the speaker into a different standard (also, usually our own). But, although the result of both goals is pretty similar, convincing and correcting a (second-order) mistake differ in the strength of the disagreement they create. Above I mentioned a natural way of classifying disagreements into strong and weak: while weak disagreements are solved easily by presenting both speakers with the same data on the topic, strong disagreements won't be solved just so. However, this distinction is too coarse for the subtleties of evaluative disagreements. A more adequate classification can be done appealing to the notions presented above.

Different discursive domains differ in the kind of disagreements more frequent in them. It is interesting to notice that the distribution of the frequency of different kinds of disagreement varies in different domains of discourse: let us distinguish among preference-expression, normative and moral discourses. Preference-expression discourse contains predicates of taste and humor, among others. Sentences like 'This is tasty' or 'Roller coasters are fun' are most likely used to inform about our preferences to others; the purpose for this expression varies contextually. These sentences rely on evaluative standards that impose a preference ordering over objects and events. Normative discourse , instead, contains deontic predicates and depends on systems of norms that determine which actions are correct and which are not. Usually, sentences in this domain range from those using deontic predicates ('It is correct to move the pawn two squares forward in its first move') or deontic modals ('You should move the horse diagonally'). Usually, these sentences are used either to educate or to correct someone's behavior inside a certain practice. Finally, moral discourse is regulated by values; it also contains sentences involving moral predicates ('It is morally wrong to lie') or moral modals ('You should not lie'). 17And as in the former case, they are also mostly used either to educate or to correct.

In the preference-explicitation domain finding standards unacceptable is much less frequent than finding them deficient. These domains present more cases of acceptance than any other. An example of this kind of conversation features speaker A and B in the ice-cream shop, disagreeing over the flavors they are going to get: 'Vanilla is delicious', says A, and B, who does not find it so, denies this assertion. This denial is motivated by the fact that B finds A's taste standard for ice-cream flavors deficient compared to his. Now, the politically correct thing to do about someone else's preferences on taste or humor is to respect it; usually, these preferences are expressed in contexts where nothing hinges on the difference between standards of taste. That is why in most cases, even though we find the other speaker's standards deficient relative to ours ('How can she like that disgusting food?' is what secretly we tell to ourselves), 18we decide to accept it and no disagreement arises. There are cases, though, in which the context forces a disagreement for practical reasons: A and B have to choose a particular ice-cream flavor for a dinner party. In that case B will find A's evaluative standard on ice-cream flavors deficient and he will try to convince A out of it because of the need to reach a solution to the problem. In turn, disagreements based on the unacceptability of the standard are rare for there is scarcely any reason to find an evaluative standard of this sort unacceptable. More precisely: there are in fact reasons to consider another person's evaluative standard unacceptable, but they are not related to taste preferences. Take for example B, a rabid vegetarian, who thinks that A's inclination to find meat delicious is unacceptable. So, whenever A voices his preference to meat by claiming it is delicious, B says it is not. There is a disagreement, but it is not based on the unacceptability of the standard of taste but for moral reasons; in fact, it is completely consistent for B to find meat delicious and to have moral reasons to avoid participating in the killing of animals.

In the normative domain, deficiency-driven disagreements are more frequent. Normative discourse issues from the human need to coordinate actions in the same way to achieve certain goals. Norms dictate how to act in order to accomplish the desired goal. Of course, for almost any goal there will be many alternative ways of achieving it, and consequently there could be as many systems of norms dictating which actions are correct and which are not provided a desire to reach that goal. This makes it reasonable to find someone else's system of norms for a certain goal inferior to ours. However, unlike preference-expressing discourse domain, the context where these sentences are uttered usually leads to disagreement and not to acceptance. Taken that the goal is shared, it is quite common to consider that ours is the best way to achieve it. Of course, there are cases where the divergent standard or system of norms is accepted: whenever the other speaker is not perceived as a member of our community (in a broad sense). 19In those cases, the cultural gap between the speakers tend to diminish the need to coordinate actions to attain a certain goal even when it is the same goal. Regarding disagreements driven by unacceptability of standards, we find here the same phenomenon we found above: even though there are reasons to find a disagreement unacceptable, these are usually related to considering the norms embraced by the other speaker morally questionable and not normatively unacceptable. Finding a norm normatively unacceptable would amount to consider it inadequate to attain the desired goal, but in this case the system of norms containing it would be clearly deficient compared to another that did attain it.

The moral domain, then, is the only one that presents a significant amount of disagreements based on unacceptability of standards. Moral discourse is based on our need to come up with a communitarian way of living, independently of particular goals we might pursue. Commitment to different values lead speakers to classify acts and events as morally right or wrong. Here, not accepting the values held by the other speaker equals to not accepting his way of living. Put in other words, we reject the alternative version of us that lives according to those values. Hence, rejecting their values can lead to finding his moral standards unacceptable. Of course, it can also lead to finding them merely deficient, and this in turn can even, in some context, issue cases of acceptance (most likely whenever the other speaker is perceived as belonging to a completely different community). But accepting alternative ways of life is, again, more a matter of political correctness than of real assessment.

Thus, it is possible to find three different distributions of motivations to discuss utterances of evaluative sentences in three different discourse domains containing them: mostly cases of acceptance in preference-expressing discourse, mostly cases of deficiency-driven disagreements in normative discourses and mostly cases of unacceptability-driven disagreement in moral discourse. This distribution is closely linked to the motives to utter the evaluative sentences in each domain. Note, for example, that we are more prone to enter into disagreements -instead of merely accepting the divergence of standards- when the discussion involves topics that affect the way we act in our society. That is why it is no that common in a discourse aimed at expressing preferences, unless those preferences affect a communal action. Among those discourse domains in which disagreement occurs, deficiency-driven disagreement seems to be more frequent in cases where the topic of discussions does not question our identity, while unacceptability-driven disagreement is more frequent where it does. Hence, C (vegan) and B (vegetarian) can disagree on what diet is more convenient once the goal of not harming animals is established: C can find B's diet deficient if she considers that consuming animal products (eggs and dairy) counts as a case of harm. But both C and B will find A's diet (based on meat) unacceptable. The means by which we do things can be more or less successful in achieving our goals and they can be discussed; the way we live -the way we think humans should live- is rarely open to discussion; divergences from it most likely are not considered deficiencies but (second-order) mistakes in deciding what life is worth of living.


BLACKBURN, S. "Following a Rule and Ethics", in Steven Holtzman and Christopher Leich (eds.) (1981), Wittgenstein: to Follow a Rule, London, Routledge & Kegan Paul. [ Links ]

EGAN, A. "Disputing about Taste". In: T. Warfield and R. Feldman (eds.) (2010), pp. 247-285. [ Links ]

EINHEUSER, I. "Three Forms of Truth Relativism". In: M. García-Carpintero and M. Kölbel (eds.) (2008), pp. 187-203. [ Links ]

FIELD, H. "Epistemology without Metaphysics". Philosophical Studies, 143, 2, pp. 249-290, 2009. [ Links ]

GARCÍA-CARPINTERO, M., KÖLBEL, (eds.) M. Relative Truth. New York: Oxford University Press, 2008. [ Links ]

HUME, D. Four Dissertations. London: A. Millar, 1757. [ Links ]

KAPLAN, D. "On the Logic of Demonstratives". Journal of Philosophical Logic, VIII, pp. 81-98 1978. [ Links ]

KNOBE, J., YALCIN, S. "Epistemic modals and context: experimental data". University of California, Berkeley, 2013: URL: ]

LASERSOHN, P. "Context Dependence, Disagreement and Predicates of Personal Taste". Linguistics and Philosophy, 28, 6, pp. 643-686, 2005. [ Links ]

LEWIS, D. "Scorekeeping in a language game" Journal of Philosophical Logic, 8, 1, pp. 339-359, 1979. [ Links ]

LÓPEZ DE SÁ, D. "The Many Relativisms and the Question of Disagreement". International Journal of Philosophical Studies, 15, 2, pp. 269-279, 2007. [ Links ]

LÓPEZ DE SÁ, D. "Presuppositions of Commonality: An Indexical Relativist Account of Disagreement". International Journal of Philosophical Studies, 15, 2, pp. 269-279, 2007. [ Links ]

MACFARLANE, J. "Making Sense of Relative Truth". Proceedings of the Aristotelian Society, 105, 1, pp. 305-323, 2005. [ Links ]

MACFARLANE, J. "Relativism and Disagreement". Philosophical Studies, 132, 1, pp. 17-31, 2007. [ Links ]

MACFARLANE, J. "Nonindexical Contextualism". Synthese, 166, 2, pp. 231-50, 2009. [ Links ]

MACFARLANE, J. Assessment Sensitivity, Oxford, Oxford University Press, 2014. [ Links ]

PARSONS, J. "Presupposition, disagreement, and predicates of taste". Forthcoming in Proceedings of the Aristotelian Society, 2013. [ Links ]

PLUNKETT, D. and SUNDELL, T. "Disagreements and the Semantics of Normative and Evaluative Terms". Philosopher's Imprint, 13, 23, 2013. [ Links ]

RICHARD, M. When Truth Gives Out. Oxford: Oxford University Press, 2008. [ Links ]

ROTT, H. (2014), (2014), "A puzzle about disagreements", Erkenntnis, pp. 1-23. [ Links ]

SARKISSIAN, H., PARK, J., TIEN, D., WRIGHT, J. C., & KNOBE, J. "Folk Moral Relativism". Mind & Language, 26, 4, pp. 482-505, 2011. [ Links ]

SCHAFER, K. "Faultless Disagreement and Realism about Matters of Taste". Philosophy and Phenomenological Research, 82, 2, pp. 265-286, 2011. [ Links ]

STALNAKER, R. Context. New York, Oxford University Press, 2014. [ Links ]

STOJANOVIC, I. "Talking about taste: disagreement, implicit arguments, and relative truth". Linguistics and Philosophy, 30, 6, pp. 691-706, 2007. [ Links ]

SUNDELL, T. "Disagreements about Taste". Philosophical Studies, 155, 2, pp. 267-288, 2011. [ Links ]

VON FINTEL, K. "What is presupposition accommodation, again?". Philosophical Perspectives, 28, Philosophy of Language, pp. 137-170, 2008. [ Links ]

20CDD: 149.946

1For a nice classification of disagreements see (MacFarlane, 2014).

2Although not perfectly: some evaluative disputes are cases of first-order disagreement.

3Evaluative sentences include those containing taste predicates and other similarly soft evaluative expressions ("funny", "interesting"). Sentences with moral, political or aesthetical predicates are also included provided a previous ontological commitment to a non-realistic position over moral, political or aesthetical facts. See (Einheuser, 2008) and (Schafer, 2011) for realist positions in the debate.

4For each kind of evaluative predicate, a different kind of evaluative standard should be assigned. For some, a dispositional account will be more adequate (Sundell (2010), Hume (1757)); for others, appealing to systems of norms seems a better choice (Field (2009)). The nature of the standard is not relevant to this paper.

5Different authors provide different explanation for this phenomenon. Paradigmatically indexical contextualism (following the classification in López de Sá (2007)) claims that evaluative expressions are associated to a hidden variable taking a contextual standard as its value (López de Sá, (2007), 2008), Sundell (2011), Stojanovic (2007), Egan (2010) and Parsons (2013)); non-indexical contextualism considers that evaluative sentences are standard-neutral but a truth-value is assigned to them according to the evaluative standard in the circumstances of evaluation (Kölbel (2003, 2004 and 2007)) and assessment relativism argues that this sentences get different truth-values in different contexts of assessments (Lasersohn (2005), MacFarlane (2005)).

6It could be objected that some evaluative disagreements will be affected by the adding of more data: as a vegetarian, knowing that some cookies that I find tasty are made with animal-based oil it can make me reconsider and find them disgusting. A response to this objection would point out that in that case I would just stop acting upon the fact that I find them tasty because I find them morally questionable, but I would not find them disgusting. Even so, I think that the divide between objective/evaluative disagreements and weak/strong disagreements are may not entirely overlap. Hence the use of 'usually' in the text.

7Epistemic modals and evaluative expressions get the same kind of semantic treatment: in both cases, truth-values of sentences are relative to a standard, which in the case of sentences containing epistemic modals consists in evidence of the speaker.

8See Von Fintel (2008) and Stalnaker (2014) for the timing of accommodation.

9A Lewisian notion of accommodation could be stated thus: If an assertion A made at time t requires any component s of the conversational score to have a value v for A to be true or otherwise acceptable, and if s does not have v as a value before t, then at t v takes r as its value some value .

10Of course, in the proper context, the dispute could be metalinguistic: imagine a context in which botanists are not the experts on classifying produce as fruit or vegetable but the classification depends on the use we give them.

11Though not for indexical contextualists, in which case these disputes could be considered metalinguistic.

12At least in most cases, for there are cases of first-order evaluative disagreements where one of the speakers misapplies his own evaluative standard and is corrected by the other.

13The example presupposes that British John is committed to the communal standards of taste. Another example could involve the misapplication of an individual standard (consider a kid claiming that he does not like French fries when she does like them).

14When we deal with other evaluative predicates (say, normative expressions), the attitudes could be different (for example, cognitive). In any case, standards always elicit positive and negative attitudes towards the valuated objet.

15If there is no other disrupting element. As happens with dispositions, preferences may not be actualized in the expected action if, for example, the agent proposes himself to act otherwise, if epistemic conditions are not given etc. Assume this exception as implicit in what follows.

16The expression "would not act consistently" may be somewhat obscure. Unwillingness to go for a walk now does not amount, per se, to the kind of unwillingness I have in mind (I may not want to go now just because I'm in the middle of something more interesting). By "would not act" I mean a rejection held consistently in time: in order to count as not willing to go for a walk in the sense above I should consistently reject going for a walk at any time, be consistently of the opinion that long walks are boring or tiresome, etc. The image (or version) of me enjoying long walks should be unacceptable for me (not agreeable for my actual self) for whatever reason.

17The difference between norms and values could be traced to Kant's distinction between categorical and hypothetical imperatives.

18There is some truth in the ideas of (López de Sá, 2007, 2008), who argues that evaluative expressions trigger a presupposition of commonality among speakers. We tend to think that others like us -as our conversational peers- are like us also in their taste/humor preferences. Therefore, acknowledgment of a divergence can be perceived as a mistake.

19(Sarkissian, Park, Tien, Wright and Knobe, 2011) conducted an experiment showing that conversational parties in the same community were most likely to consider false the utterances of their contenders, answering contentiously; conversational parties of different cultures considered most likely that they both were correct and were prone to conciliatory answers. Contentious exchanges are not that frequent: in most conversations on taste, even if the hearer does not agree with the speaker, she will not reject nor will feel the urge to discuss him. But conversations with a strong investment in the valuation do elicit contentious answers more often than not.

Received: April 16, 2015; Accepted: June 06, 2015

Creative Commons License This is an Open Access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.