Discursive pluralism: Inferentialist expressivism and the integration challenge

Discursive pluralism, recently fostered by anti-representationalist views, by stating that not all assertions conform to a descriptive model of language, poses an interesting challenge to representationalism. Although in recent years alethic pluralism has become more and more popular as an interesting way out for this issue, the discussion also hosts other interesting minority approaches in the anti-representationalist camp.

with moral statements we express our attitudes. 2This insight often entails a liberal approach, for we do different things with language, and many of them can hardly be reduced to mere representing.
This paper tries to approach discursive pluralism as it emerges from the problems of representationalism and proceeds along three directions: a connection with "placement problems" in metaphysics; semantic alternatives to representationalism; and the implications of alethic deflationism for representationalism.
Discursive pluralism is indeed relevant for the puzzles generated by attempting to understand nonfactual vocabularies in representationalist terms.Should our use of moral vocabulary commit us to moral facts?Moral expressivism provides the paradigmatic retraction of commitment to moral facts by acknowledging that ethical statements do not describe.Hence, we can achieve a more sober ontology through splitting the expressive and descriptive functions of the language by assigning them to different vocabularies.
Discursive pluralism thus appears as an alternative to representationalism.For example, Ludwig Wittgenstein in his Philosophical Investigations criticized representationalism (which he had formerly defended in the Tractatus Logico-Philosophicus) by acknowledging a stunning plurality of "language games.""Representing," he concluded, is just one among many functions; language is a praxis, and if we seek an understanding of the meanings of our words, we should look at how we use them in our language games.
Discursive pluralism also addresses alternatives to substantive-that is, explanatoryconceptions of truth.Not only is alethic deflationism, the idea that "true" plays an expressive role, an option that avoids many long-standing issues about truth, it also entails that truth is not an explanatory concept. 3This is important, since representationalism likewise depends on an explanatory conception of truth (see below).Deflationary truth opens the way for different explanatory resources-resources that are legitimizing alternatives to representationalismand permits understanding statements like "It is true that war is unjust" without a correspondence between statements and facts.
Among expressivist views there are options that fruitfully intersect, constituting the spine of neopragmatist conceptions of language.In the context of the views defended by Huw Price and Robert Brandom-and independently of some issues that may divide them-the central commitments of their versions of expressivism neatly align with the overall issues of discursive pluralism, both in unifying the semantic explanation of it and in deflating its metaphysical complications while granting a default acknowledgment of it.
In what follows, section 2 sketches the antecedents of anti-representationalism by shaping discursive pluralism in the perspective promoted by Price.Then, section 3 addresses the placement problems for representationalism that motivate its alternatives.The fourth section explores the connections between expressivism and semantic anti-representationalism.Here, Brandom's inferentialism is presented as the main anti-representationalist account of our assertory practices.At this point, section 5 concerns Brandom's expressivism, thanks to the theoretical apparatus of "analytic pragmatism."From this Brandomian conception, it is easy to see how pluralism naturally follows, as language shows up as an ensemble of pragmatic functions and expressively powerful resources.Finally, section 6 explores how this expressivist pluralism offers advantages over alternative accounts from three points of view: metaphysics, semantics, and theory of truth.Thus, in addition to arguments supporting discursive pluralism within each thematic expressivist/anti-representationalist perspective, there seem to be external considerations that emerge from the exploration of the main theoretical commitments' entailments.I argue in particular that in the context of discursive pluralism, (a) inferentialist expressivism is metaphysically convenient because it avoids placement problems, (b) it is semantically unified in terms of inferential norms, and (c) it is unified by endorsing alethic deflationism.These three aspects compose a unified framework, concerning discursive pluralism that is advantageous and easily demonstrates how alternative conceptions must often rely on ad hoc solutions to achieve comparable results.

| A M I NOR I T Y T R A DI T ION: CON T E M POR A RY EX PR E S SI V I SM A N D I TS DEV ELOPM EN TS
According to representationalism, concepts like "truth," "reference," and "representation" play an explanatory role in semantics.In representationalism, therefore, knowing the meaning of a sentence S means knowing how things would be if S were true.Representationalism shapes many research programs in the philosophy of language, in linguistics, in cognitive science, and in epistemology.This does not mean, however, that we lack alternative conceptions.
Among these alternatives, we can cite the family of ideas called "expressivism."Even though expressivism has a peculiar history in the meta-ethical debates between cognitivism and noncognitivism, its beginnings were already concerned with language. 4In this context, Simon Blackburn's "quasi-realism" has been an outstanding option for its defence of the expressivist legacy without denying the fact that moral statements share the surface grammar of factual assertions.Inspired by the later Wittgenstein, Blackburn focused on the linguistic functions and ontological commitments of expressive statements: according to him, we express our attitudes and do not undertake ontological commitments in doing so (Blackburn 1984 and 1993).Blackburn distinguishes two linguistic functions: descriptive functions, which concern representing facts, and expressive functions, which concern non-factual domains (morality/modality/mathematics).
We have, however, seen a fast and ramified evolution of expressivism in recent decades.Many authors have contributed to this ongoing evolution, for example Allan Gibbard, Robert Brandom, Paul Horwich, Michael Williams, Mark Schroeder, and Huw Price.Their work, moreover, is not just a development of Blackburn's ideas but rather derives from philosophers like Wittgenstein, who is cited above; Wilfrid Sellars, who developed expressivist insights in criticizing logical empiricism; and Richard Rorty, who promoted neopragmatism in opposition to representationalism.These authors are in fact responsible as much as Blackburn for the basic alternatives to representationalism and their underlying motivations.
It all began with Wittgenstein's reconsideration of tractiarian representationalism in the Investigations: to a pictorial/representational account of language, Wittgenstein contraposed the idea of a social practice comprising many "language games" like "counting," "naming," and so on.Furthermore, some language games cannot be understood in representationalist terms (Wittgenstein 1953, § § 1-2); for instance, when workers cry "Slab!" they are not representing slabs but giving an order; when one reads "five red apples" in a shopping list, one does not understand it as representing a number, a colour, and a collection of fruit-such words do not represent anything without the practice of eating fruit already in place and capacities like 4 According to cognitivism, moral judgments can be true/false.Non-cognitivism denies that.counting, recognizing colours, and so forth.Thus, Wittgenstein withdrew a step from representationalism, and this became the root for similar views.
Another towering figure in the critique of representationalism is Sellars, who contrasted the dominant empiricism of the first half of the twentieth century and its urge to reduce meaningful discourse to describing, on the one hand, with, on the other hand, the cases of modal and normative statements, which were particularly pressing in the quest for an alternative view.Sellars complained against this view, which we can call "descriptivism": "[O]nce the tautology 'The world is described by descriptive concepts' is freed from the idea that the business of all non-logical concepts is to describe, the way is clear to an ungrudging recognition that many expressions which empiricists have relegated to second-class citizenship in discourse are not inferior, just different" (Sellars 1957,  § 79).By rejecting "descriptivism," we abandon the idea that the primary point of language is describing, and thus that non-descriptive forms of language are secondary.This idea also helps supply alternatives to representationalism by explaining Sellars's approach to meaning based on the idea of "role in reasoning" (Sellars 1953,  1974, and 1997 [1956]).Furthermore, this is close to Wittgenstein's idea of language as a multifarious collection of uses/practices (though Sellars, unlike Wittgenstein, did not give up on semantic theorizing).
These lead to Blackburn's expressivism, in which moral discourse, unlike factual discourse, does not work in terms of representations or truth conditions.Thus, understanding moral vocabulary as expressive entails a "bifurcation" between factual and non-factual vocabularies: there are vocabularies which represent facts, and non-factual ones where we mimic representing facts, but without ontological commitments; these latter are merely expressions of attitudes.In fact, moral statements play a function that differs from describing/stating facts, thus explaining the difference in their ontological commitments.Blackburn's bifurcation clarifies, despite its admission of representationalism for factual discourse, how representationalism cannot cover all domains.There are clear cases-such as moral statements like "Stealing is wrong"-where representationalism is not helpful, and where alternatives are more promising.Even though this is a local retreat from representationalism-as it concerns morality/modality/mathematics-it is a first step towards sound alternatives.It would be hard to deny that this proposal has been the paradigm for further accounts.
Then there is Rorty, who contested the representationalism of the post-Cartesian epistemological tradition (Rorty 1979).He promoted a radical anti-representationalist alternative that opposed the role of representation in the understanding of knowledge, mind, and language.The problem here is that the concept of representation is both the main theoretical player in the tradition and mostly unclear.Hence, Rorty diagnosed how representationalism is the source of metaphysical disputes (for example, realism versus anti-realism) or the chain of philosophers' replies to scepticism.He claimed that by abandoning representationalism we would dissolve, in a therapeutic Wittgensteinian spirit, such issues as foundational mistakes.The work of Sellars, Quine, and Davidson, in his opinion, left no space for a representationalist programme, and so he urged a pragmatist overcoming of the issue.Thus, together with the abandonment of representationalism, Rorty promoted a pragmatist revival.The upshot of this reconsideration was a shift of focus towards our social practices and language games.
These insights help us to reconsider representationalism both traditionally and in ongoing debates.A problem in combining them, however, is that anti-representationalism comes in degrees.For example, a tension appears between Blackburn's views and Rorty's: while Rorty simply rejects representationalism, Blackburn embraces it as the soundest option for the empirical domain.Therefore, we need to orient ourselves clearly in the anti-representationalist context before approaching discursive pluralism.At least we should distinguish between global (GA) and local (LA) versions of anti-representationalism: According to this distinction, the contribution of anti-representationalist views to an understanding of expressivist conceptions may vary in important ways.For example, Price defends GA as the basis for a global expressivist (GE) view; to him, vocabularies do not represent, and their role is entirely expressive. 6Brandom's view is different: by admitting an expressive role for the representational vocabulary, he endorses a kind of local expressivism (LE) and LA (Brandom 2013).Before introducing anti-representationalism, however, let us explore the main issues concerning representationalism.

| R E PR E SE N TAT ION A L I SM' S PUZZ L E S
In his later works, Huw Price (2011 and 2013) assembles insights from different authors and traditions, but his major point presents a problem for representationalism.At the beginning, representationalism is seen as an implicit commitment of a reductive attitude in contemporary philosophy that Price calls "object naturalism" and is defined as follows: As an ontological doctrine, it is the view that in some important sense, all there is is the world studied by science.As an epistemological doctrine, it is the view that all genuine knowledge is scientific knowledge.(Price 2011, 185)   When we think about the world as studied by science, we may become aware that "in many interesting cases it is hard to see what natural facts we could be talking about" (Price 2011, 186).Such difficulties concern an implicit connection between object naturalism and representationalism. 7 But before going into this, we should introduce Price's alternative to object naturalism, which he calls "subject naturalism," defined as follows: According to this second view, philosophy needs to begin with what science tells us about ourselves.Science tells us that we humans are natural creatures, and if the claims and ambitions of philosophy conflict with this view, then philosophy needs to give way.(Price 2011, 186)   What science suggests about us as human beings indicates how we can be naturalist without the reductionism of object naturalism.Price's project here goes far from the focus of the present discussion in its concern with discursive pluralism, so I will not discuss object naturalism and subject naturalism here.Nonetheless, some of this contraposition's presuppositions are relevant: while subject naturalism is a pragmatist option as its focus on humans is de facto on their practices and abilities, object naturalism embeds representationalist presuppositions.Hence, we face a renewed 5 Price's view is more complicated than this and requires further qualifications, since it evolved over time.See Price 2011 and 2013. 6Price exploits, for this globalization, both Blackburn's expressivism and Brandom's inferentialism.Characterizing global expressivism as a mere globalization of Blackburn's (local) view would entail, as nicely highlighted in Simpson 2020, that global expressivism is not at all different from local expressivism. 7Even though semantic representationalism is not a scientific truth about meaning: naturalism should commit us only to entities and facts grasped by the a posteriori descriptive and explanatory methods of empirical inquiry: that is, we are not (naturalistically) entitled to representationalist word-world relations, which are not grasped by means of empirical inquiry (Price 2013, 13-14).
version of the contraposition between pragmatism and representationalism, but with a semantic focus.
These representationalist presuppositions of object naturalism are often implicit and thus not quite evident; according to Price, they become visible by the exploration of certain controversial facts we talk about within an object-naturalist framework.In fact, such cases highlight some metaphysical difficulties that depend on these presuppositions.These hard cases show up in the following questions: "If all reality is ultimately natural reality, how are we to 'place' moral facts, mathematical facts, meaning facts, and so on?How are we to locate topics of these kinds within a naturalistic framework, thus conceived?"(Price 2011, 187).These questions undermine the framework that generates them.In fact, in view of these puzzles one could feel entitled either to drop naturalism or to endorse a form of non-naturalism about the controversial cases, such as meanings, values, and so on.Yet Price endorses a third option and chooses to maintain naturalism, but without its representationalist presuppositions.What about these puzzles?
According to representationalism, vocabularies represent; hence, the meanings of our statements depend on how things are.Language is taken to be primarily representational in nature, and every vocabulary is understood as devoted to representing something-this is uncontroversial for empirical statements, at least those like "The cat is on the mat."Furthermore, special vocabularies-like normative or semantic ones-can also be taken to represent a portion of reality.This projection of representationalism across the board faces complications, however, leading to puzzling metaphysical commitments that generate what Price calls "placement problems."Let us look more closely at these commitments.
As we can see in table 1, vocabularies are associated with the facts these are supposed to represent.These "naturalistically controversial facts" (Price 2011, 220), in fact, appear every time we attempt to develop a representational understanding of vocabularies that hardly appear to work this way.Why are such facts troublesome for their representationalist interpretation?This is evident for the moral case: Do moral statements represent facts?As this is a notorious issue, the moral terrain is the source of alternatives to a representationalist understanding-to avoid commitments to moral realism.In this context, Blackburn's expressivism has for decades been a valuable alternative.Non-descriptive statements are not liable to be true or false (in a representationalist sense), but they are true of our rules, meanings, actions, and practices. 8Expressive vocabularies mimic factual discourse-that is, expressive statements share a grammatical appearance with factual statements-but without a metaphysical substance under the surface.If moral discourse commits us to mysterious facts, then we had better drop moral representationalism, while keeping the descriptive part of language in line with representationalism and standard ontological commitments.This reasoning holds if one wants to avoid such commitments, but this is not mandatory.A moral realist, for instance, would be happy to commit herself to values as independent entities.This option, however, as Price remarks, is in tension with our culture, where science is authoritative about what populates reality.Therefore, according to Price, we have naturalistic reasons for suspicion of realism concerning certain recalcitrant entities-not only values but also numbers, meanings, and so forth. 9Therefore, a naturalistic answer suggests endorsing expressivism and quarantining the controversial commitments.The expressivist alternative is Price's starting point, but his goal goes beyond endorsing moral expressivism; moral vocabulary, in fact, is not alone in leading us to posit controversial facts.All the vocabularies in table 1, for example, are liable to undergo a parallel treatment: modal and probabilistic vocabularies would represent, accordingly, possibilities and probabilities.The representationalist projection can be widely extended, generating puzzles across the board.
These clues suffice, if one rejects such commitments, to explore alternative options.These vocabularies are not devoted to representing facts, and their representationalist interpretation leads to facts that hardly populate the natural world: there are regions of discourse where representationalism is untenable because it generates puzzles.Therefore, the conclusion we feel entitled to draw is a need to limit the scope of representationalism to empirical locutions and to embrace, for such recalcitrant vocabularies as those in table 1, some alternative understanding.In the presence of questionable commitments, we should drop representationalism for that local context, as Blackburn has done for the moral, modal, and mathematical idioms.Nevertheless, as Price notices, other vocabularies can undergo a parallel treatment.This situation contraposes representationalism about empirical discourse with a constellation of discursive provinces emptied of ontological commitments.This is a first step towards liberalizing expressivism about the vocabularies bearing problems with their representationalist interpretation.It avoids placement problems, but the theorist here can opt for a more ambitious programme: if we endorse deflationary views on the use of "true" in debates between cognitivism and non-cognitivism, then we can also read factual/empirical claims as deprived of their representationalist significance.Hence, Price endorses anti-representationalism as a global exceptionless option concerning the extension of Blackburn's treatment to all vocabularies, whether non-factual or factual: they are both a route towards GE.Just as Blackburn's pragmatic account for moral discourse does, we could characterize other vocabularies in terms of various "pragmatic functions." This move, according to Price, finds its terrain thanks to a conception of the "game of assertion" that escapes representationalism.Brandom's inferentialism is this kind of account liable to pave the way to this expressivist liberalization, explaining the meaning of our assertions in terms of their role in reasoning rather than exploiting the represented facts.

| T H E GA M E OF A S SERT ION
Assertive speech acts affirm that things are thus and so.Such an innocent statement often presupposes a representationalist reading supported by certain examples like "The keys are in your pocket!"With an assertion like this I am usually understood to be representing the keys 9 It may seem that here Price disqualifies many central problems in the history of philosophy, like the nature of numbers or values, and so forth.Price is, however, mostly denouncing the untenable connection between naturalism and representationalism concerning such problems.
as located in your pocket.The idea is that there is a correspondence between statements and states of affairs; for example, the assertion "The cat is on the mat" says that there is really a cat on the mat.So far, this approach shows an undeniable intuitive appeal, but it is just half of the story.
If we consider statements like "Stealing is wrong" and "La Gioconda is beautiful," then things change, and the notion of correspondence loses much of its appeal. 10We certainly do assert such things, but are we then committed to the corresponding facts?This question seems to depend on the semantic presuppositions that we accept.If we stick to representationalism, then we must explain how "blatantly non-factual" statements-like "Honesty is good"-are supposed to represent.In a representationalist context this can be a problem.Representationalism, notwithstanding its popularity, is not the only game in town, though.One can, in fact, following Price's advice, throw out the representationalist baby together with this metaphysical bathwater.Yet for serious consideration this option needs an alternative view on assertion.
According to Brandom, assertions are explained not in a representationalist way but as expressions of inferential "commitments" and as "moves" affecting the deontic statuses of speakers, which are understood as participants in normative social practices.Advancing claims means making moves in the game of "giving and asking for reasons," where assertions can play the role of premises for further claims or of entailments of other statements.Therefore, assertions work according to inferential rules: different assertions, which belong to different vocabularies, are good or bad according to their inferential norms of use and not according to corresponding facts.In this way, statements like "The frog is on the log" and "The death penalty is wrong" work according to a unified model that is not representationalist and without undertaking controversial ontological commitments. 11From a semantic point of view, both statements have their meaning explained in terms of their premises and consequences.From a pragmatic point of view, both statements work as speech acts used to publicly undertake discursive commitments.This view provides an alternative to explanations in terms of truth conditions while maintaining, even though in inferentialist terms, that assertions say that things are thus and so-so we need not think of this exclusively in terms of representation. 12o say that assertion is not explained in terms of some substantive notion also means endorsing a deflationary attitude.According to deflationism, truth plays an expressive role and the meaning of "true" depends on its linguistic use, permitting us to make generalizations like "All Roger said is true."Hence, the inferentialist account of assertion is not only a unified view about the content of statements belonging to different vocabularies but is also coherent with the explanatory anti-representationalism endorsed by Brandom and deflationist about semantic vocabulary. 10Usually, speech act theorists avoid inserting value judgments in the realm of assertions.This is a way to say, more freely, that assertions are in general truth-apt.When, however, one utters "The death penalty is wrong," one behaves just as one does in asserting: in fact, one is making a claim, just as uttering "The cat is on the mat" would be.Ruling out value judgments from the realm of assertion seems to be an ad hoc move to avoid complications that are likely to follow.11 I am not saying with this that anti-representationalists and expressivists all agree on the same ideas about the expressivist understanding of ethical vocabulary-they just agree on the fact that it does not work in terms of representations.In fact, while Price explicitly endorses metaethical non-cognitivism, Sellars's expressivism is much more nuanced and complicated (see, e.g., Sellars 1980 and Koons 2018).Brandom is mostly silent on his account of ethical vocabulary, but in the few passages in which he addresses the issue he seems to distance himself from non-cognitivism (see Brandom 2009, 210).Thanks to an anonymous referee for pointing this out. 12This account is explanatorily anti-representationalist, as it avoids the explanatory notion of representation (or sibling notions like "reference," "truth," and "truth conditions").Nevertheless, inferentialism purports to explain the representational dimension of our descriptive statements in non-representationalist terms.See Brandom 2013 and 2022 and Salis 2020.

| BR A N DOM'S "V ERT ICA L" EX PR E S SI V I SM: PR AGM AT IC F U NCT ION A N D E X PR E S SI V E POW ER
At this point, we can introduce Brandom's view of expressive vocabularies.Its preliminary feature is that such expressivism does not share the traditional idea that non-descriptive moves/statements express one's mental states/attitudes.Hence, this view cannot be understood as a stretch of metaethical expressivism towards other vocabularies. 13Many considerations follow.
First, "genuine" descriptive moves/statements, and the elements of descriptive language used, are explained in terms of their inferential roles, and hence the traditional expressivist bifurcation loses any definitional function.This means that a segment of language is genuinely descriptive (ordinary empirical descriptive vocabulary-OEDV), that this descriptive function need not endorse (explanatory) representationalism, and that this descriptive layer is compatible with deflationism (as explained in terms of inferential role).
Second, this account concerns normative, intentional, semantic, modal, and logical vocabularies.This is an aspect that lines up with Price's programme, as he perceives expressivism as a kind of discursive pluralism and understands expressive vocabularies in terms of their pragmatic functions.This pluralism, however, comes with noteworthy differences from Brandom's programme that I think can put some pressure on certain aspects of Price's.The most important difference concerning this plurality of vocabularies lies in the insufficiency, for Brandom, of the notion of pragmatic function (for example, saying how things are, expressing our attitudes/feelings, and the like) to account for such resources.To this, Brandom adds the notion of the "expressive power" that a vocabulary may possess (for example, the expressive power of semantic vocabulary is poorer than OEDV's, since it lets us say fewer things than the latter does-even though semantic vocabulary can be used to specify ordinary claims in terms of truth conditions).This aspect marks differences in expressive power among vocabularies, complicating the relations they entertain with other practices or vocabularies.Furthermore, such relations are even more complicated, since differences in expressive power and pragmatic function combine in relevant ways; the upshot is a "hierarchical" understanding of expressivism, which contrasts with an "egalitarian" understanding of discursive pluralism understood in terms of pragmatic function.
Third, expressive vocabularies are, according to Brandom, metalinguistic and specify at a higher level of explicitness what we say with OEDV (so we can specify a claim as "correct," "necessary," and so forth).Just as Alfred Tarski emphasized how semantic vocabulary works as a metalanguage for ordinary talk, so we can understand, in a similar vein, other vocabularies as performing analogous functions.These vocabularies specify, with a higher level of explicitness and with certain targets (theoretical or practical) in view, assertive moves.This feature can prima facie be troublesome for Price's programme, as it suggests a hierarchy of expressive power among vocabularies going beyond functional differentiations.Hence, there is the concern over the compatibility of Brandom's hierarchical and metalinguistic expressivism with Price's conception; while it is safe to say that Brandom's inferentialism can be enrolled in Price's globalization, the case for Brandom's expressivism is challenging, since Price acknowledges as metalinguistic only the metavocabulary used to specify the "functions" played by many vocabularies (Price 2019).
Therefore, not only does this metalinguistic expressivism differ from meta-ethical expressivism in scope (viz., ethics versus many vocabularies), it also establishes a hierarchy of expressive power that codifies the divide between OEDV and "pragmatic metalanguages" (Brandom's name for expressive vocabularies).While meta-ethical expressivism was local and horizontal (as moral and descriptive discourses were distinguished by pragmatic inferentialism, in fact, explicitly addresses the problem of sentential embeddings and declares itself on the Fregean side (see Brandom 2009, 208-14).Furthermore, his metalinguistic understanding of expressive vocabularies makes the application of expressive predicates assimilable to how "true" fits into sentential embeddings (except transparency).
function and semantic explanation), Brandomian expressivism is plurally local as it concerns a constellation of local resources and "vertical" as these resources enrich the expressive power of the language and do not belong to OEDV's level of expressive power.Brandom provided a formal characterization of logical, modal, normative, and intentional locutions by means of meaning-use diagrams (MUDs) (Brandom 2008).A preliminary understanding of expressive vocabularies can, however, be provided by his account of the role of logical vocabulary in Making It Explicit (Brandom 1994).
According to logical expressivism, logical vocabulary plays an expressive function. 14For example, by mastering conditionals-such as "If X is an iguana, then X is a reptile"-I can make it explicit that "Chuck is a reptile" follows from "Chuck is an iguana."This view about conditionals also provides a plausible alternative to positing "conditional facts" as corresponding to the use of conditionals.Negation performs the task of codifying incompatibility relations that structure giving reasons against certain claims: "[A] premise-set implies not-A just in case A is incompatible with that premise-set" (Brandom 2018, 148). 15y mastering the locutions of negation, one can indicate that certain statements are mutually incompatible, for example that one cannot coherently endorse both.Alice's commitment to "The cat is on the mat" is incompatible with that undertaken with statements like "The thing you refer to as 'the cat' is a CIA sleeper agent."Therefore, the commitment undertaken with "The cat is on the mat" prevents Alice committing herself to the other statement.Also, this reading of negation avoids an understanding of it as corresponding to negative facts.Losing the expressive function of conditionals and negation would, according to logical expressivism, impoverish our resources, as our discursive practices would have a lower level of achievable explicitness.Such a loss would hinder the possibility of making explicit what we implicitly do in discursive practice: saying what follows from our assertions and what we are committed to by means of them.This would not affect our capacities of asserting and inferring; this practice would survive the loss of logical locutions, but such loss would affect the capacity of elucidating what we do with our assertions and inferences.This is crucial in the context of Brandom's inferentialism where this level of explicitness is needed to talk about the norms that we implicitly follow in our discursive interactions.With the logical vocabulary in place, participants in a discursive practice can make explicit the contents of the concepts that they apply with their linguistic moves. 16This also means that the expressive power of logical vocabulary is different from that of OEDV: it is rather a further layer enabling us to do more and specify our linguistic moves.
We can now approach other expressive vocabularies by exploiting this take on logical expressivism.Every expressive vocabulary, according to Brandom, may empower us in connection to specific goals.This amounts to a functionalist characterization of expressive vocabularies through which we specify with different resources and different purposes what we say and do in our practices.Since expressive vocabularies play an elucidative function, such a view requires that we understand these vocabularies as metalinguistic. 17Besides these metalinguistic aspects there are functional ones, but functional aspects must be understood thanks to their metalinguistic features.As the logical vocabulary plays an elucidating role in making conceptual contents explicit, these vocabularies permit us to explicitly characterize certain aspects of what we say and do concerning possibility or necessity (alethic modal vocabulary), correctness/incorrectness (normative vocabulary), aboutness 14 This view is relevant for the "demarcation question" in philosophy of logic: What characteristics should be taken as identifying some bit of vocabulary as distinctively logical?(See Brandom 2000, chap. 1, 2008, chap.2, and 2018.)   15 "¬A is the minimal incompatible of A, in the sense of being implied by everything that is incompatible with A" (Brandom 2018,  148). 16For a historical reconstruction of the background of these ideas, see Brandom 2000, chap. 1, andTurbanti 2017, 86-89. 17 The view about this metalinguistic function is developed also in Brandom 2015, chap. 1, by reconstructing Kant's legacy in Sellars's thought.(representational vocabulary), truth (semantic vocabulary), and so forth.Hence, depending on what we need to do, these vocabularies help specify what we say in different terms for different purposes.They provide substantial improvements for our practices, particularly the practice of giving and asking for reasons, which coordinates the use of such (meta)vocabularies, as it is the practice comprising the basic assertive and inferential moves.The practice of giving and asking for reasons is in this sense an autonomous discursive practice; practitioners, who make assertions and draw inferences, need no further resources to be able to engage in this practice.Other resources like expressive metavocabularies need other resources in place (and hence are not autonomous); for example, such resources presuppose a practice, already in place, consisting of moves like assertions and inferences.This means, for example, that the moves that we make in our discursive practices belong to what we can loosely understand as an "object language."This is OEDV, which comprises words like "red," "heavy," "loud," "dog," and all the locutions we use to say how things are.OEDV plays a noncontroversial descriptive function; like Blackburn, Brandom here distinguishes between descriptive and non-descriptive functions in the language-but, unlike him, without representationalism. 18t is a noteworthy feature of this view that it can count as a "pragmatist pluralism" about expressive vocabularies.The improvement those vocabularies bring to our practices can in principle be as multidirectional as the plurality of our goals and activities.Accomplishing certain tasks demands adopting certain resources and not others, and this type of instrumental maxim applies also to our vocabularies.Brandom's expressive vocabularies allow us to specify, each in its own terms, what we say and do with our assertions.For example, the normative vocabulary is useful in making the normative dimension (correctness), implicit in what we say and do, explicit.The modal vocabulary, on the other hand, can specify certain assertions in terms of necessity, possibility, dispositions, and so on.These resources are indeed helpful to complement, elucidate, evaluate, and improve our moves made with OEDV. 19his plurality does not, however, entail mutual independence among vocabularies, since they may be connected to others by what Brandom calls pragmatically mediated semantic relations.For example, normative and modal vocabularies can be related; we could use normative vocabulary as a pragmatic metalanguage to specify the practice of making modal claims, so a pragmatic metalanguage can be the "base" vocabulary for another pragmatic metalanguage.A modal statement like "It cannot be Sunday today and not be Monday tomorrow" can be read as containing conditionals that work as inferential licences, and hence as norms.The pragmatically mediated semantic relations between specific vocabularies, which establish "special relations" between them, imply that the plurality of vocabularies does not entail an egalitarian view, since not all vocabularies display the same level of expressive power.Hence, we face two basic ways of characterizing these resources: we can distinguish them by the functions they perform, on the one hand, and by their expressive power, on the other.We can represent the functional differentiation among vocabulariesthe same one that Price exploits for his GE-as a horizontal line in which we can move leftward or rightward between vocabularies by changing the pragmatic function (see figure 1).Therefore, I read the hierarchical differentiation of metavocabularies in terms of expressive power as contrasting a simply horizontal differentiation of vocabularies in terms of function; we must complement that with a vertical characterization in terms of expressive power (Brandom 2013).This vertical differentiation among vocabularies should 18 Anti-representationalism sets the divide between Brandom and Blackburn: whereas Blackburn accepts representationalism for descriptive vocabulary-while denying it for non-descriptive vocabularies-Brandom gives up explanatory representationalism for all vocabularies. 19So, these pragmatic metavocabularies make a difference in the reckoning of the deontic score of speakers.See Brandom 2013, 103-6, for the case of the representational vocabulary (de re ascriptions of propositional attitudes) as helpful from this point of view.
not, however, be conceived as an orthogonal line in respect to the horizontal line used in figure 1 to represent pragmatic functions; rather, we must add relations of "pragmatic dependence" between vocabularies and practices. 20e can now sketch certain differences in expressive power among vocabularies.For example, OEDV is fundamental for every autonomous discursive practice; without it, we could hardly have discursive practice or other vocabularies at all.We could not perform the basic assertions and inferences that can be specified by means of expressive vocabularies.Roughly, we could describe the hierarchy of expressive power among vocabularies as having OEDV (containing also observational vocabulary) as its base, a practice of making assertions and inferences-and this practice permits us to deploy an OEDV-and above the various pragmatic metavocabularies that help us specify and improve what we say in the base vocabulary and do in the practice.These expressive metavocabularies depend in many ways on the abilities involved in the practices of making claims and of drawing inferences, and on the vocabulary deployed by participation in that practice (that is, OEDV). 21This hierarchical conception confers a central position on some vocabularies, and it not only impedes an egalitarian assimilation among them but also complicates the relations between practices (such as making claims) and expressive vocabularies (such as normative vocabulary).

| EXPRESSIVISM, PLURALISM, AND THE INTEGRATION CHALLENGE: A TRIPLE PARSIMONIOUS UNIFICATION
This anti-egalitarianism about vocabularies contrasts with pluralism prima facie.Usually, inegalitarian views have been considered "anti-pluralist" in spirit.For example, Rorty complained about understanding certain vocabularies as more fundamental because of their supposedly "corresponding" with reality, and so he promoted a liberation from bad "correspondence" metaphors by claiming that pluralism was the upshot of this approach.Every vocabulary's legitimacy depends on our practices and not on its "correspondence" degree.From this point of view, one could suppose that pluralism requires some egalitarianism.This is not mandatory, however, since one can be pluralist even while assigning different weights to different vocabularies; the pre-eminence that Brandom assigns to some vocabularies is immune from Rorty's criticism by being based on pragmatically mediated semantic relations and not on correspondence.The centrality of the game of assertion that is not explained in terms of representations is what permits the proliferation of expressive vocabularies.The hierarchy of 20 Two examples are the following: (P)ractice(V)ocabulary-(suff)iciency: "engaging in a practice (or set of practices) is sufficient for someone to count as deploying a specific vocabulary" (Brandom 2008, 9); (P)ractice(V)ocabulary-(nec)essity: "a practice P is necessary for deploying a vocabulary V if and only if the ability to deploy the vocabulary V requires being able to engage in the practice P" (Brandom 2008, 28).

21
For example, the reliable capacity to respond differentially to perceptual stimuli.expressive resources can therefore be understood as a pluralist expansion of discursive practice, not as an obstacle.
Moreover, inferentialism about assertory practices provides an understanding of pluralism that is unitary and metaphysically sober; we no longer need to explain the meanings of certain language elements in one way and of other elements in other ways; we likewise no longer need to multiply facts as corresponding to the use of our vocabularies.Understanding the meanings of all claims in inferentialist terms provides a "uniform clothing" for our language-to employ Price's adaptation of Wittgenstein's phrase-that does not threaten a functional differentiation of vocabularies.As Price put it: "[W]hile Brandom's account may impose a degree of uniformity on language that some Wittgensteinian pluralists might wish to reject-offering us a uniform account of the way in which Wittgenstein's common linguistic 'clothing' is held together …-it not only allows but actually requires that this uniformity coexist with an underlying functional diversity of the kind that expressivists such as Blackburn and Gibbard require.It not only allows but insists that different pieces of linguistic clothing do different things, even though … they are all put together in the same way and all belong to the same assertoric game" (Price 2013, 34).
Thus, inferentialism about the assertion game provides an understanding of language as semantically uniform even though it may perform many pragmatic functions.Without such an account, we face a dilemma: either unify the understanding of language by endorsing representationalism, with the entailment of placement problems, or differentiate the semantics of our vocabularies by distinguishing those that are legitimately representational from those that are expressive and so save the metaphysics with the consequence of losing the explanatory unity of the semantics.Inferentialism here provides a third option by deflating placement problems without losing the semantic uniformity.In fact, together with the explanatory unity in semantics, inferentialism provides an ontological perspective that is conservative, since only descriptive elements of language bear nonconservative commitments (Williams 2013) like those involved in the representationalist interpretation of nondescriptive vocabularies. 22rice's motivations for pluralism concern naturalism, since the metaphysics generated by the adoption of representationalism and the representationalist semantics presupposed by naturalistic attempts tend to violate their own constraints.Placement problems signal that something has gone wrong both for metaphysics and for naturalism-as we end up with naturalistically controversial facts.Price uses this consequence to criticize contemporary metaphysics.The use I am making of Brandom's insights in the context of Price's programme does not necessarily underwrite these motivations; I see these options more as providing alternatives in order to understand discursive pluralism than as developing a critique of metaphysics.
This pluralism concerning the expressive power and the pragmatic functions of our vocabularies is not just the upshot of these conceptions but a general valuable entailment.One could say that non-reductionist views on language consider pluralism as an evident and pervasive phenomenon (after the later Wittgenstein), so one could say that my emphasis on the pluralist entailments of these expressivist views is trivial, or at least less significant than I claimed.Indeed, it could be understood as nothing more than a slightly Brandomian version of the story that Price has promoted over the years.If we widen the perspective, though, further implications become visible.In fact, I am going to highlight how such a perspective prima facie solves at once many problems in three interrelated areas: metaphysics, semantics, and truth.
The first result concerns metaphysics.Inferentialist expressivism by rejecting explanatory representationalism avoids any placement problems.Our vocabularies do not work by representing, 22 Nonconservative commitments concern facts that hardly populate the natural world-that is, those that involve new facts.so we find ourselves happy without commitments to questionable facts.A relevant consequence is this parsimony about the metaphysics underwritten by our use of these vocabularies.According to this expressivist metaphysical parsimony, nonfactual statements do not bear ontological commitments, so they are ontologically conservative.Certainly, there are alternative ways to achieve such a parsimony, but the provision of a viable alternative to representationalism is enough.Both factual and nonfactual claims are explained in terms of inferential norms and not as representing facts, and this avoids the metaphysical puzzles of representationalism.Finally, this attitude towards metaphysics need not be the somewhat "Carnapian" expulsion that Price sometimes suggests (at other times he calls it, in a more irenic way, a kind of quietism); rather, it merely entails the deflation of placement problems for the controversial cases.
Next, there is semantics.If we approach such an understanding of language from a semantic point of view, we realize that this form of expressivism explains the content of all the assertions in our vocabularies in terms of inferential role, so it provides a unified framework.Every assertible judgment from every vocabulary, like "The cat is on the mat," "The death penalty is wrong," "La Gioconda is beautiful," and so forth, depends on its premises and conclusions.This is a huge result because we have not only achieved a unified view for all vocabularies though granting to each functional autonomy but also avoided any bifurcation in the explanation of meanings.This means that we can distinguish the many pragmatic functions played by various vocabularies without breaking their semantic explanatory unity.Defenders of bifurcation at least prima facie in the presence of unified alternatives should provide an explanation and a defence for such a divide. 23ruth is the third aspect.Approaching this conception with truth in view, we acknowledge that inferentialist expressivism embraces deflationism, which is the idea that "true" is expressive and does not play any explanatory role.Also, deflationism undermines representationalism about semantic vocabulary, since locutions such as "true" do not correspond to alethic facts or properties, so semantic vocabulary presents its own placement problem.In this respect, as Price says, deflationism counts as an elegant solution for the placement problem generated by the representationalist understanding of semantic vocabulary (Price 2011, 15,  25, 32, 182, 194, 254-59; Tebben 2015, 5).Not only is this a unified framework, it also avoids multiplying our concepts and functions of truth, as sometimes pluralist views seem compelled to do-I have in mind the range of options going from Crispin Wright's to Michael Lynch's, for example (Wright 1992; Lynch 2009; see also Pedersen and Wright 2013).Since multiplying our concepts of truth in order to match every vocabulary or domain does not seem intuitively promising from a metaphysical point of view, and also because such an alethic pluralism tends to break a unitary perspective, I think we should rejoice that expressivism entails pluralism in a unified deflationist perspective.
If we consider discursive pluralism through a grid of options concerning metaphysics, semantics, and truth, we can see how inferentialist expressivism fares by comparison with non-expressivist functionalism and representationalism (figure 2).It highlights how expressivism ticks all the right boxes, especially when compared to the alternative conceptions.As we can see, inferentialist expressivism's meta-theoretical advantages present themselves convincingly and are evident in this integration challenge: a "triple"-as extended to the three domains-"parsimonious"-as deflating placement problems-"unification"-as truth and meaning receive uniform treatments.Surely, non-expressivist functionalism can improve its score in metaphysics by identifying certain accounts and not others, but it must deal with the potential problems entailed by breaking the unity both in semantics and in theory of truth. 2423 A nice attempt at drawing a more tenable bifurcation is provided in Tebben 2021.

24
I am referring to the problem about mixed inferences/mixed discourse, where truth appears to be one despite the use of many vocabularies (see Tappolet 1997).
This triple parsimonious unification is provided on the basis of few basic metaphilosophical choices at the level of explanatory resources.The first just lies in renouncing substantial representational notions as playing explanatory roles in semantics and in theory of truth: notions like "truth," "truth conditions," "representation," and so forth, are removed from the explanatory toolkit.The second move comes by replacing them with notions that are much less compromised with representationalism, such as attempting to explain meaning in terms of inferences, along the lines traced by Brandom's inferentialism.The third basic move is Price's idea of explaining discursive pluralism in terms of the pragmatic functions performed by our vocabularies (to be reconciled with Brandom's idea of the specific expressive power of our vocabularies) as providing a more flexible way to manage the metaphysical commitments underlying our understanding of language.

| CONC LUSION
An immediate upshot of the replacement of representationalism with expressivism is the endorsement of a pluralist attitude towards language.The non-representationalist accounts of the game of assertion, such as Brandom's, lie at the very heart of this alternative: correctness of assertions depends on inferential norms and not on substantive semantic notions.This point explains the underlying connection between expressivism and deflationism in which truth does not play any explanatory role.This account successfully connects the functional plurality of our vocabularies with a metaphysical parsimony that avoids placement problems.
Emerging from what I called the triple parsimonious unification, inferentialist expressivism fares well in metaphysics, in semantics, and in the theory of truth, not just for each account taken in isolation but for the general vision that it shapes.In metaphysics, by rejecting explanatory representationalism we deflate placement problems altogether; in semantics, with the inferentialist unification in the explanation of meaning we avoid any bifurcation; and in the theory of truth, this expressivism embraces deflationism, avoiding the implications of truth pluralism (and of representationalism) in a unified way.

AC K NOW L E DGM E N T S
I wish to thank an anonymous referee for this journal, whose comments greatly helped to improve the paper.Early versions of this work were presented at the ECAP.10 (Tenth Congress of the European Society for Analytic Philosophy) in 2020 and at the Sifa2021 (Italian Society for Analytic Philosophy) conferences: my thanks to all attendees for valuable comments and suggestions.Open Access provided by the (CRUI-CARE) agreement between Università degli Studi di Cagliari and Wiley.
LA: Some vocabularies play a legitimate representational function.(This view, in different versions, is implicit in the work of Wittgenstein, Sellars, Blackburn, and Brandom.)GA: No vocabularies play a representational function.(This view has been endorsed by Rorty and Price.) 5