KEYNOTE: Expression and Meaning: Acts, Products, and ‘Normative Language’

Dorit Bar-On (University of Connecticut)

[Jump to comments]

 

Expression is a notion that’s used ubiquitously, yet it has received surprisingly little direct theoretical attention.  I have been interested in understanding the use and usefulness of expression in connection with three seemingly unrelated puzzles – about first-person authority, about motivational internalism, and about the origins of meaning.  I begin with a brief review of the puzzles and well-known, expressivist attempts to solve them, which have been dismissed (Section 1).  In each case, opponents have appealed to an apparent incongruence between what is ‘merely expressive’ and what is linguistically meaningful.  In Sections 2 and 3, I sketch the view of expression and expressive behavior I favor, which (I have argued elsewhere) allows us to articulate viable, neo-expressivist solutions to the puzzles.  One of my main aims in this paper will be to bring out the myriad ways in which the expressive and the linguistic interact, both in the origins of meaningful language and in its current, everyday use.  In the final section, I offer some tentative reflections on consequences of the view I outline for the alleged distinctiveness of so-called normative language.  

1. Expression and Three Puzzles

1.1 Origins of Meaning

Although the communicative repertoires of nonhuman animals can exhibit considerable complexity and subtlety, even the most sophisticated known systems of animal communication possess no discrete combinatorial structure, and their use appears to be both affect-driven and keyed to environmental stimuli.[1]  Animal communication systems, it is often thought, are wholly reflexive-reactive, while human language use is reflective-creative.  Thus, there seems to be an unbridgeable gap between the character of meaningful human language and that of all known animal communication systems.  Hence the puzzle concerning how language could have emerged in nature.[2]  Several thinkers have considered expressive behaviors of the sort we share with nonhuman animals (yelps, growls, calls, facial expressions, bodily postures, and such) as potentially providing insight into the ancient source of human speech.  However, opponents of these expressivist proposals have argued that, given their expressive nature, animals’ calls and gestures form a poor breeding ground for human language, with its distinctively rule-governed, intentional, symbolic, and compositional character.[3]

1.2 First-Person Authority

Our second puzzle concerns articulate avowals – spontaneous utterances involving self-ascriptions of occurrent mental states, such as “I feel awful”, “I’m scared of that snake”, or “I’m wondering whether he got my email”.[4]  Despite their surface resemblance to ordinary reports of contingent states of affairs, avowals appear to enjoy unique epistemic security.  Unlike nonmental ascriptions, third-person mental ascriptions, and even first-person evidential mental reports, avowals are typically taken at face value, and are rarely questioned.  It seems out of place to challenge an avowal’s truth, or to ask an avower to provide reasons or justification for her avowal.  Yet, though they seem to be made on no epistemic basis, avowals are taken to represent a paradigmatic case of privileged knowledge.

According to simple avowal expressivism, the epistemic asymmetries between avowals and other claims are to be explained by the fact that, appearances to the contrary, in terms of their ‘logical grammar’, avowals are more like natural expressions of sensa­tions, such as moans, grimaces, and giggles: they are direct expressions – as opposed to reports or descriptions of the avowed states of mind.  However, simple avowal expressivism purchases the explanation of the epistemic asymmetries at the cost of compro­mis­ing notable semantic continuities between avowals and other ascriptions, and by assigning avowals no epistemic status; it portrays them as excluded from the realm of linguistic meaning, inference, judgment, and knowledge.

1.3 Motivational Internalism

Finally, again in the realm of meaningful, structured discourse, consider an ethical claim such as “Hunting for sport is wrong”.[5]  Proponents of internalism in metaethics maintain that one who makes such a claim competently can’t be completely motivationally indifferent to hunting; whereas one who makes the grammatically similar claim “Hunting for sport is expensive” can be.  The view is that, in general, a person cannot competently and sincerely make an ethical claim without being some­what motivated to act (or refrain from acting) in accordance with it.

According to ethical expressivism, the internal connection to motivation can be captured by taking ethical claims to betray the relevant motivational attitudes.  On one radical version of the view, due to Ayer, saying “Hunting for sport is bad” is in effect semantically equivalent to saying something like “Boo to hunting for sport!”, or to “Hunting for sport!” uttered in a disapproving tone of voice.  So ethical terms are taken to make no semantic contribution to sentences containing them.[6]  This, however, seems to fly in the face of the fact that, in ordinary discourse, ethical claims exhibit logical, semantic, and epistemological continuities with ordinary descriptive claims (just as do avowals).  The noncognitivist expressivist, it seems, is purchasing the explanation of motivational asymmetry at the cost of denying these continuities; and many have taken this to be the major stumbling block for ethical expressivism.[7]

2. What Is Expression?

In each of the three cases just described, the appeal to expression is designed to fulfill a certain explanatory role: to bridge a seeming evolutionary gulf between the non-linguistic and the linguistic, in the first case, and – within the linguistic domain – to capture apparent epistemic asymmetries and accommodate a surprising motivational contrast, in the second and third cases, respectively.  Abstracting from some details, the key reason for rejecting the expressivist proposal in each case is the observation that there is too big a gap between expressive utterances, on the one hand, and semantically articulate, truth-evaluable, descriptive, compositional sentences, on the other hand, resulting in a fundamental incongruity between the expressive and the semantic domains.  The underlying thought seems to be that where rational, conceptual-propositional language begins is where undifferentiated expressive-emotional behavior ends; and that bits of language whose function is to express mental states or attitudes can’t have proper semantics or epistemology.  I believe this thought relies on a misguided conception of expression and expressive behavior.  Below I sketch an alternative conception, which (I’ve argued elsewhere) allows for viable, neo-expressivist solutions to the puzzles.

2.1 A-Expression vs. S-Expression; Acts, Vehicles, and Products

The class of behaviors we ordinarily describe as ‘expressive’ spans a wide range.  At one end of the spectrum, we have so-called natural expressions, such as cries and laughs, grimaces, and various gestures, where both the behavior and its connection to the expressed states are supposed to be due to nature.  There are also mimicked or acquired facial expressions, sounds, or gestures that become ‘second nature’, such as shrugging shoulders or tut-tutting.  Then we have conven­tional nonverbal expressions, such as tipping one’s hat, giving the thumbs up, or sticking out one’s tongue.  (And there are, of course, many cases in between.)  Still in the conventional realm, we have expressive verbal utterances such as “Darn it!” or “Ouch!”, “Sorry!”, “This is great!” “I hate you!” and so on.  We also find in the verbal domain utterances such as “It’s getting late”, which on an occasion may simply serve to express a thought that has occurred to one (as opposed to constituting an assertion, or request to leave, etc.), or “Oh for some sun,” which would typically be taken to express the speaker’s wish for sun.  Finally, at the far end of the conventional side of the spectrum, we have speech acts, such as assertion, promising, or commanding, which are said to have the expression of certain mental states as part of their so-called felicity conditions.[8]

Still, we can discern the following commonality among the expressive behaviors mentioned so far: they all express states of minds.  Following Sellars (1969), we can characterize expression as a relation that holds between performers of acts and the mental states these acts directly express: expression in the action sense, or “a-expression”, for short.  (So a-expression is a three-place relation: an agent J a-expresses mental state M by using expressive means E.)  Your dog, when he gets up and walks over to give you a lick, is a-expressing his affectionate feeling, nonverbally, of course.  And when you give a friend a hug, or say “You’re here!” in an excited tone of voice, or, alternatively, say: “It’s so great to see you”, or “I’m so glad to see you”, you a-express your joy at seeing her through an intentionally produced act, using diverse expressive vehicles.

A-expression contrasts with expression in the semantic sense (“s-expression”, for short).  So, for example, an utterance of “It is raining” or “Es regnet”, as well as the that-clauses of “Sheila thinks that it’s raining” and “George hopes that it’s raining”, or a particular thought-token It’s raining, can all be said to s-express one and the same proposition.  And of the word “charity” and its translations into other languages, we can say that they s-express the concept CHARITY.  S-expression, then, is a relation that holds between linguistic strings, such as sentences and other contentful tokens, and their semantic contents.[9]  A-expression, by contrast, is something a minded creature does, be it through bodily demeanor, facial expres­sion or gesture, or else through speech, using natural, culturally acquired, or even idiosyncratic expressive vehicles.

When one a-expresses a state of mind using a sentence token (or some other symbolic vehicle), the vehicle retains its symbolic meaning.  “It’s great to see you” and “I’m so glad to see you” each have different linguistic meanings.  Following tradition, we can take each to s-express a distinct proposition, in virtue of the linguistic rules governing the lexical items and the relevant mode of composition.  What proposition?  Well, setting aside some nuances about the context-sensitivity of indexicals, it’s most natural to say that the former sentence simply expresses the proposition that it’s great to see the add­res­see, whereas the latter sentence expresses the proposition that the speaker is happy to see her add­ressee.[10]  For all that, when producing tokens of these sentence types, in speech or in thought, one can be a-expressing one’s joy.  Intuitively, a-expression is a more basic phenomenon than s-expression.  It’s certainly more ubiquitous; nonhuman animals and prelinguistic children express states of mind through a variety of nonlinguistic means.  They don’t have at their disposal a discrete, combinatorial, recursive system of learned, arbitrary, symbolic vehicles capable of s-expression; but they are nevertheless capable of a-expressing their states of mind through nonlinguistic, natural, nonsymbolic expressive means (which is not to say that they are able to express nonverbally all the same states of mind we express using language!).

Now, our first puzzle, about the origins of meaning, is, in effect, the puzzle of how linguistic vehicles of expression could emerge out of nonlinguistic natural expressions.  However, it’s important to note (and relevant to our other two puzzles) that the emergence of such linguistic vehicles in no way supersedes the use of expressive vehicles (of whatever stripe) to a-express one’s states of mind.  As we’ll see below, among linguistic creatures, familiar learning processes lead to the increasing use of more or less conventional means to give voice to – articulate, vent, air, share – present states of mind.  Distinguishing between expressive acts, or performances, on the one hand, and the expressive vehicles used in them, or (as I sometimes put it) the products of those acts, on the other, allows us to capture underlying action-theoretic similarities between expressing one’s annoyance through a gesture, a facial contortion, a tone of voice, an inarticulate sound, or a full sentence, while still acknowledging significant differences.

2.2 Expressing and Showing

Consider early expressive behavior in our own species.  Upon seeing a fluffy new teddy bear, little Jenny’s face may light up; or she may let out an excited gasp, pointing at the teddy bear; or she may emit a distinctive sound (“Uh!”), or call out: “Teddy!” as she reaches to grab the toy; or she may exclaim: “Gimme Teddy!” perhaps with no reaching.  Jenny’s facial expression and her gasp are what we ordinarily think of as ‘purely natural’, unlearned expressions; her eager reaching and subsequent utterances are expressive behaviors she voluntarily – perhaps even intentionally – engages in, and which serve to give vent to her desire to get the toy.  The same, we saw, holds for adult language users, who may express their amusement at a joke by laughing (which, as product, does not s-express amusement), as well as by uttering a sentence such as “This is so funny!” or “I find this hilarious”.  These sentences, I say, s-express propositions, but the acts performed using them are similar to those performed using nonverbal expressive means, whether acquired or natural, inasmuch as they serve to give vent to speakers’ occurrent states of mind.  Similar expressive performances or acts, different products or vehicles of expression, with different etiologies.

What holds the different performances together?  An idea that takes its inspiration from earlier philosophical work on expression, as well as contemporary empirical work, is that when an agent engages in an act of expressing her state of mind, what she does is show that state.  When confronted with an animal baring its teeth in anger, a child smiling in pleasure, a person raising an eyebrow, or even blurting out a snide remark, we regard the behavior as allowing us to witness how things are with the expresser.  We often speak of seeing someone’s anger, hearing the nervousness in someone’s uneven voice, feeling the tension in someone’s body, and so on.  Pioneering work done by the psychologist Paul Ekman on human facial expressions, as well as more recent applications of it to nonhuman primates,[11] supports the idea that a suitably endowed individual can perceive an emotion such as anger, in the face of another, possibly by perceiving a naturally characteristic component of it, say, the baring of teeth, which foretells the animal’s impending action of striking at the target of its anger, and perceiving, e.g. the state’s target, through following the expresser’s gaze or bodily orientation.  The capacity to show and perceive states of mind in behavior is a capacity with deep phylogenetic roots.[12]

But the idea, I think, has application that goes beyond so-called natural expressions, which, as expressive vehicles, are unlearned or innate.  For example, through a process known as ‘ontogenetic ritualization’, a baby chimp will learn to tap lightly on his mother’s back, rather than pulling her down, as a way of showing his desire to climb on her back.  And the mother learns to recognize immediately what the tap shows.  More or less similar processes in humans lead to the acquisition of various culturally and even individually diverse facial expressions, vocalizations, gestures, and intonation contours – all of which can still be regarded as characteristic components (albeit acquired, and in that sense not ‘natural’) of (being in) the relevant states. The same is true for expressive behavior involving linguistic vehicles.[13]

2.3 Showing in Language?

Thus go back to little Jenny, eagerly stretching out her hand toward the toy.  It’s worth noting that we don’t expect to be able to locate the object of Jenny’s desire in the eagerly reaching arm.  We recognize Jenny’s desire for Teddy in part by spontaneously following the direction of her outstretched arm.  Having recognized what it is she wants, a parent witnessing her behavior may say: “Teddy – you want Teddy!” so that Jenny can next utter “Teddy!” as she reaches.  Jenny’s acquisition of the word “Teddy” allows her to give more articulate voice to her specific desire for Teddy.  The verbal articulation shoulders some of the burden initially placed on the nonverbal behavior to reveal aspects of the state of mind the child is expressing.  We have here a familiar acquisition paradigm: the child produces non-linguistic voluntary expressive behavior and the adult passes on to her a new expressive means – a linguistic vehicle – for articulating aspects of the psychological state that are shown through the behavior, thereby facilitating the child’s transition to at least minimally linguistic behavior.[14]

Linguistic exposure, habituation, enculturation, and other social experiences enable expressers to integrate into their expressive repertoires a wide variety of acquired (‘nonnatural’) expressive vehicles with which to give voice to their states of mind.  And, as noted earlier, the emergence of specifically linguistic vehicles (in both ontogeny and phylogeny) in no way crowds out the use of expressive vehicles of all kinds to a-express one’s states of mind.  For, through processes of acquisition and enculturation, linguistic creatures make increasing use of more or less conventional means – intonation, gestures, words and phrases, grammatical moods, and many other devices – to show their present states of mind (making a disapproving face, thinking out loud, airing opinions, and so on).  Once a repertoire of acquired expressive vehicles is in place, both the use of expressive devices – linguistic or not – to express the presence (and various aspects) of expressers’ states of mind and their uptake by observers become second nature.  Their use can become as spontaneous, uncalculated, and non-deliberate – and their uptake by observers as immediate and direct – as the use and uptake of natural expressions.  And, especially when it comes to specifically linguistic devices, expressive purposes can be accomplished without any nonverbal expressive accompani­ments (they can even be used in print!).[15]^{, [16]}  Thus, in creatures like us, some of the communicative roles played by the more ‘visceral’ showing and perceiving afforded by animals’ growls, bared teeth, grimaces, and so on, are taken over by spontaneous, competent use and immediate uptake of linguistic vehicles.[17]

As we turn to consider ways in which language is used to give expression to states of mind, we should keep in mind the enormously wide variety of expressive vehicles afforded in language.  To begin with, we have – as illustrated by our earlier example of Jenny – words that can serve to articulate intentional aspects of expressed states of mind “Teddy”, “ball” “hug”.  While these mark an important point of entry into language, they represent only one variety of expressive devices.   More overtly expressive devices include, inter alia, inarticulate interjections of the “Oh/Ooh” variety, inexplicit slurs, derogatory and laudatory terms, more articulate exclamatives (“What a great performance that was!”), and expressive sentential constructions (such as “It’s exciting that p/to Ҩ” or “Ҩ’ing is annoying”).[18]  Linguistic expressives are subject – albeit to varying degrees – to phono­logi­cal, syntactic, semantic, and pragmatic constraints that govern other linguistic constructions.  Moreover, even devices whose expressive function appears to exhaust their use – words such as “Yuck”, “Yum” – can take on adjectival form (“Yucky”, “Yummy”), allowing them to be seamlessly integrated into propositional-compositional language.  At the same time, so-called purely descriptive terms admit of use in the exclamative form (“What a tall boy he is!” or “How late she arrived!”).  Thus, we must not lose sight of the fact that, in a mature natural language, linguistic devices – whether ‘purely descriptive’ or ‘purely expressive’ – belong in a structured, compositional linguistic system that has a life of its own.  (This is a point to which we will return later on.)

3. Expressive Communication: A Neo-Expressivist Approach

Let me now turn very briefly to the application of my proposed understanding of expression and expressive behavior to the puzzles with which we began.

3.1 Neo-Expressivism: Origins of Meaning[19]

In his seminal work – The Expression of the Emotions in Man and Animals (1872) – Darwin identifies expressive behavior as representing an evolutionarily significant common ground between ‘man and animals’.  By expressive behavior, Darwin had in mind these sorts of displays

to which we could add yelps, growls, wagging tails, fear barks and grimaces, lip smacks, ground slaps, food-begging gestures, ‘play faces’, alarm, distress and food calls, grooming grunts, open-mouth and ear-flap threats, eyebrow flashes, and so on.[20]  In the literature on animal communication and language evolution, it is common to find such behaviors described as purely reflexive-responsive ‘natural signs’, assimilated to mere physio­logi­cal symptoms, such as red spots on the skin and sneezes, or to unintentional displays that merely convey information about biologically significant features of the displayer (like a peacock’s tail). And their character is often contrasted sharply with the reflective-intentional, rule-governed character of conventional-symbolic linguistic signs.[21]

With Darwin, I think that such characterizations fail to do justice to the richness, complexity, and evolutionary function of expressive behaviors.[22]  Although naturally expressive behaviors aren’t in general instances of intentional communication by expressers, they are not simply reliable indicators of their states of mind; this is part of what is entailed by saying that such behaviors show expressed states.[23]  A dog’s cowering demeanor, or a play bow, a monkey’s agitated alarm call, and so on – are Janus-faced: they point inward – to the animal’s expressed state of agitation, fear, anger, etc. – at the same time as they point outward – toward the object or event at which the state is directed.  They also reveal the relevant behavior’s cause or motivation at the same time as they foretell the expresser’s impending behavior and move others to respond appropriately.[24]  Though expressive signals lack compositional structure, they inherit a certain complexity from the states of mind they express; they have referential and even predicative dimensions (though not components), and may be said to exhibit a measure of intentionality or subjective directedness, even when not produced intentionally.  And though animals’ expressive repertoires are not in general learned, and their expressive displays are not in general deliberate, expressive behaviors go well beyond merely instinctive or reflexive ones.

We can think of animals’ expressive behavior as designed by nature (as opposed to individual intention or culture) to show the kind, degree, and object of an expresser’s state of mind to a suitably endowed audience (typically, conspecifics), as well as to move the audience to an appropriate response to relevant objects or states of affairs (perhaps via social-emotional contagion or empathy).  As vehicles, natural expressions are not symbolic, and they do not s-express propositions.  But they can still exhibit various degrees of what we may call ‘proto-propositional’ content, in virtue of showing various aspects (object, degree, quality) of the expressed states.  If this is right, then, contrary to prevalent prejudice, animals’ expressive behaviors have features that prefigure significant psychological, semantic, and pragmatic features of human linguistic communi­cation.  The task of explaining the emergence of meaningful speech is, then, not the impossible one of crossing the Rubicon that separates purely instinctive, reflexive-reactive signals, on the one hand, and reflective-creative, symbolic and compositional linguistic signs, on the other.  Instead, it is the more tractable task of identifying processes that could transform complex (though semantically inarticulate) expressive vehicles, used by social, nonlinguistic creatures to a-express their states of mind in communicating with others, into increasingly more articulate vehicles that eventually take on semantic life.

3.2 Neo-Expressivism: First-Person Authority and Motivational Internalism

Earlier, I pointed out that, even once linguistic vehicles capable of autonomous, articulate s-expression are in place, language users continue to engage in acts of giving vent to the full array of states of minds they possess – both affective and cognitive – using both linguistic and nonlinguistic devices.  In general, mastery of acquired expressive vehicles requires apprehending the link to the states of mind that can be a-expressed through their use.  However, in the case of expressive acts that use articulate linguistic vehicles, we can (and should) separate what a given sentence s-expresses from the mental states speakers use the sentence to a-express, on a given occasion, or even characteristically.  (After all, as we saw earlier, two sentences with obviously different meanings can, on a given occasion, be used to a-express one and the same sentiment.)  This, I have argued elsewhere,[25] would allow us to explain both the epistemic asymmetry between avowals and other ascriptions and the motivational contrast between ethical and ordinary descriptive claims, without compromising relevant semantic and epistemic continuities. Very briefly:

Re First-Person Authority:  Like such other English terms as ‘building’, ‘painting’, or ‘peeling’, terms such as ‘judgment’, ‘statement’, ‘report’, and ‘claim’, are ambiguous between a process or act of some sort – the claiming, or reporting, etc. – and the product of that act.  To make the claim or issue the report “It is raining” is to perform an action of a certain sort; and what is produced by that action is a sentence-token.[26]  It’s true that in making the claim, a state of mind gets expressed (typically – but not invariably! – the belief that it is raining).  But in light of the above distinction between a-expression and s-expression, we can see that it is the person, and not the sentence, that a-expresses the belief.  Sentences are simply not in the business of a-expressing states of mind.  Instead, sentences (indicative ones, at any rate) s-express propositions.[27]

The term ‘avowal’, too, exhibits the act/product ambiguity.  In keeping with the expressivist insight, we can explain avowals’ special status in terms of the distinctive expressive character of acts of avowing.  Consider, first, the intuitive contrast between acts of expressing a given state of mind and acts of merely telling about its presence.  Anyone can say truly, and some can even tell reliably, that I’m feeling sad.  But presumably only I am in a position to express my sad feeling itself – for example, by letting tears roll down my cheeks, or saying “This is so sad”.  When you say “DB is feeling sad”, you are employing a sentence that s-expresses the propo­si­tion that DB is feeling sad, and, if you are sincere, you are a-expressing your belief that DB is feeling sad.  My tears, on the other hand, s-express nothing (though I may be a-expressing my sadness by letting them roll down my cheeks); and the sentence “This is so sad” s-expresses a proposition that describes something as sad, but in uttering it, one would typically be a-expressing her sadness.

Now, the neo-expressivist account I advocate seeks to explain the epistemic asymmetry between avowals and reports of the same states without compromising semantic continuities.[28]  On this account, the asymmetries hold not between types of sentences with certain semantic contents, but rather between acts that directly express one’s mental state and reports of that state, whoever produces them, and however reliably.[29]  Avowals are taken to be acts in which we a-express the very state that the proposition s-expressed by the sentence we use ascribes to us.  So, considered as acts, avowals – like grimaces and groans, and unlike evidential reports – directly a-express the avowed mental states; consequently, they are immune to epistemic criticism.  However, inasmuch as such acts involve using as expressive vehicles sentence tokens that s-express self-ascriptive propositions, avowals considered as products are importantly different from other kinds of expressions of the relevant states.[30]

Re Motivational Internalism:  When we turn to the ethical case, the neo-expressivist proposal is that the motivational contrast between ethical discourse and ordinary descriptive discourse can be captured by appeal to the expressive character of acts of making ethical claims.[31]  The key idea is that ethical claims – understood as acts – are ‘in the business of’ a-expressing motivational states (whatever those turn out to be, according to our best moral psychology).[32]  However, ethical claims often employ as their expressive vehicles indicative sentence- (or thought-) tokens.  And we can take these tokens to s-express propositions.  We can thus explain the motivational contrast without compromising the palpable logico-semantic continuities between ethical claims and ordinary descriptive claims.  Thus, to use Ayer’s example, when one affirms that tolerance is a virtue (in speech or in thought), the sentence- or thought- token one produces can be said to s-express a truth-evaluable proposition.  (Of course, it’s a substantive – and difficult – question what, if anything, would make the proposition true or false, but it’s not a question to be settled by semantics).  This enables ethical claims – understood as products – to participate in logical inferences, embed in negation and other truth-functional contexts, etc.  However, as with avowals, what is s-expressed does not automatically settle what mental state is (even characteristically) a-expressed by acts of making the claim.  We can perhaps agree with motivational internalists that ethical claims – as acts – are ‘in the business of’ expressing, specifically, motivational attitudes.  This is not because those making such claims report that they have the relevant attitudes.[33]  Nor, I submit, does it require taking the attitudes (a-) expressed to constitute the semantic contribution of ethical terms to what sentences containing them s-express.  Instead, neo-expressivism proposes that, just as it’s the way a given self-ascription is issued – and not its semantic content – that determines whether the act of issuing it is an avowal, so it’s the way a sentence containing ethical terms is put forward – not the sentence’s content – that determines whether the act of producing it is one of (genuinely) making a ethical claim.  If motivational internalism is right, properly issuing ethical claims requires giving voice to – that is, a-expressing – the very states whose presence is required for understanding their perceived motivational force.[34], [35]

4. Expression and ‘Normative Language’

A key question for expressivists in various domains has been how best to represent the contribution of expressed states of mind to discourse.  It has been assumed by both expressivists and their opponents that being an expressivist requires locating that contribution in the literal semantics of the relevant discourse – i.e., in the linguistic meanings of sentences used to make claims in that discourse (or in the semantic content of relevant thoughts).  As we just saw, neo-expressiv­ism rejects this assumption, and maintains that the expression of states of mind or attitudes should be located in acts (e.g. of making claims).  It is in the performance of such acts that states of mind and attitudes get expressed – in the use of language, not in the semantics of the language used.  In the time that remains, I’d like to offer some broad conside­rations in support of this rejection, and briefly reflect on the proper role of expression in accounting for the distinctive character of so-called normative language (continuing to focus on the ethical case).  I shall reserve fuller development of these reflections for another occasion.

4.1 Expressivism’s Two Tasks

I’ve been arguing that we should keep separate the expression of states of mind and attitudes as something done with language, on the one hand, from expression as a semantic relation holding in language, between meaningful tokens and their contents.  This, in order to take proper account of action-theoretic continuities in acts that employ diverse kinds of expressive vehicles, while accommodating logico-semantic continuities among specifically linguistic[36] vehicles of expression.  But, at least in the ethical case, it has seemed important for expressivists to insist on keeping together – in fact, identifying – that which captures the expressive character of ethical claims and that which provides the meanings of ethical sentences.

By way of diagnosis, I suggest that this is because, historically, ethical expressivism has undertaken two different tasks that can – and, I’d argue, should – be decoupled.  In response to the challenge posed by Moore’s ‘open-question argument’, expressivists have tried to preserve the linguistic significance of ethical terms without (as Moore was apparently prepared to do) embracing nonnatural, sui generis properties for such terms to denote.  Call this ‘the ontological task’.  But the unanalyzability of ethical terms that Moore’s argument highlights has another crucial aspect.  That is the ostensible normativity of ethical language.  Precious few atomic terms of any natural language admit of analytic definition.[37]  (Try ‘love’, ‘nice’, ‘sad’, ‘weight’, ‘hill’.)  And even when we’re satisfied with a proposed dictionary definition of a term, we would not in every case expect the definition to reveal transparently the underlying property or properties (if any) shared by all and only things to which the term applies.  In the case of ethical (and other evaluative) terms, though, what seems to require explanation, beyond identifying some properties (if any) that the terms denote, is their normativity.  Now, normative language, it is thought, is language that prescribes rather than describes, that commends/condemns or otherwise evaluates, rather than purely reports, that directs, rather than neutrally represents, that signals commitments rather than registers observations.  So any proposed analysand for ethical terms that fails to partake in these features is bound to fall afoul of some version of the ‘open question’ argument.  (This is so – NB – regardless of whether the analysans invokes natural or nonnatural properties.)  One advantage expressivism can be thought to have over other major responses to Moore’s argument is that it directly addresses the normativity of ethical claims. And this is its second, ‘normativist task’.

Commitment to the dual agenda set by Moore’s argument, together with difficulties associated with the Frege-Geach problem, may provide one explanation for the relatively recent emergence of a variety of expressivist proposals for overhauling of traditional semantics.[38]  For example, according to “ideationalist” views, semantics should not assign truth-evaluable propositions as the meanings of any sentences; instead, all sentences mean what they do in virtue of the ‘idea’ (read: mental state type) that they express.[39]  Along these lines, Mark Richard has recently argued that absorbing the consequences for semantics of recognizing the normative character of sentences such as “Hunting for sport is good” requires a radical departure from ‘Grandfather’s semantics’.  A correct semantic theory, Richard proposes, should proceed by, first, determining, for each simple sentence, “what attitude the sentence is a means to expressing”.  This, he says, “will often be determined by vocabulary – sentences of the form a is expensive express beliefs, sentences of the form a is good express (one or another form of) approval”.  Next, what needs to be determined is the kind of commitment that is “characteristic of the attitudes expressed”. This is what the semantics will then assign as the sentence’s meaning.[40]

Recent critics of ideationalist (including so-called hybrid) theories have focused on persisting difficulties in working out in a satisfactory way a semantics built (even in part) on an attitudinal foundation; the conclusion they think should be drawn is that even the more sophisticated versions of expressivism are ultimately hopeless.[41]  But, if neo-expressivism is right, it is possible to preserve the positive expressivist intuition concerning an ‘internal’ connection between ethical claims and expressed evaluative attitudes without at the same time undertaking to secure the negative onto­lo­gi­cal claim.[42]  Indeed, the burden of earlier and forthcoming work on ethical neo-expressiv­ism has been to show how the distinction between a- and s-expression, coupled with a suitably modest under­standing of propositional content (as well as a separation between truth-evaluable proposi­tions and their truth-makers) can liberate expressivism from the need to undertake a thorough­going overhaul of traditional semantics – a tall order by anyone’s lights.  This modest under­standing of propositional content can be independently motivated by considerations of language mastery, structure, and acquisition.  Adopting it requires accepting that the ontological task – in the case of ethical discourse, as in all other cases – is not to be handled by one’s semantic theory; but for all that, it does not require the expressivist to abandon ontological parsimony or naturalism (see below).  So both tasks can be tackled – though not with a single theoretical stroke.

4.2 The Autonomy of Meaning: the Attitude/Content Distinction

There is much to be said for taking the sentences with which we express states of mind to have autonomous – yet relatively ‘lean’ – propositional content, and to think of the expressive dimension of claims made in various domains as pertaining to acts of making those claims.  Traditionally, the notion of propositional content has been invoked by theories of meaning to accommodate preservation of content in inter-linguistic translation, as well as intra-linguistic commonalities of content across different gram­matical contexts and propositional attitude attributions.  More ambitiously, it has also been supposed that propositional semantics would afford the pairing of components of truth-evaluable propositions with the objects, properties, and relations that would make the propositions true.  The neo-expressivist approach does not embrace this ambition.  The notion of propositions it invokes when claiming that ethical claims s-express propositions is a relatively innocuous notion of a proposition, one that does not require analytic paraphrase and does not commit us to specific truth-makers.  I now want to marshal some considerations in support of this relatively modest notion.

Consider again “Yuck!”.  Mastery of “Yuck” requires recognizing its specific expressive function.  Plausibly, “yuck” originated from the nonverbal gag vocalization, and its expressive function (to express one’s disgust) is near enough exhaustive of its use in the language.  Yet, through a familiar and productive process of language development, “yuck” has acquired adjectival form, so we now have “yucky”; and by a simple application of a syntactic rule, we get ‘‘This is yucky’’, which is semantically continuous with “This is blue”.  (We can also get “You have an owi” from the purely expressive “Ow!”.)  Similarly for other conventionally expressive sentences, such as ‘‘[It’s] Nice to meet you,’’ ‘‘[I’m] Much obliged.’’

Reflection on these and other cases of purely expressive uses of language suggests that semantic continuity comes with acquired propositional form, and does not require descriptive origins or grounding in objective facts.  At the same time – it should by now be clear – acquired propositional form (involving nouns, adjectives, and subject-predicate form) does not imply loss of expressive function in use.  Moreover, having factual grounding is by no means sufficient for passing the standard ‘tests’ for propositionality.  For example, “What a long journey that was” has uncontroversially descriptive or factual import.   But, for all that, it is no more fit for embedding in well-formed conditionals or for being entered as a premise in a Modus Ponens than the exclamation “Long journey!” produced in an exasperated tone of voice.  The link between descriptiveness and propositionality thus seems relatively superficial.

Now consider a simple example of a structured sentence such as “John loves Mary”.  A competent speaker of English will know that, e.g., “John loves Mary” s-expresses the proposition that, well, John loves Mary.  A proper semantic analysis will, we hope, yield an account of how the proposition (s-)expressed is a systematic function of the semantic values of the parts of the sentence and how these are composed in its logical form.  It may (if one adopts an inferential semantics) reveal various inferential relations that reflect competent speakers’ understanding of the sentence.  So we can accept that the job of a semantic theory is not exhausted by providing a disquotational specification of s-expressed propositions.  What shouldn’t be regarded as part of its job description, I’ve suggested, is to provide in some way a substantive characterization of the conditions that must obtain for the sentence to be true.[43]^, [44]

4.3 ‘Normative Language’

What about the normativitist task (accounting for the normative character of ethical claims)?  It has seemed tempting to address this task by focusing attention on the meanings of terms characteristic of a given discourse and identifying ‘normatively-laden’ states of mind that are to be semantically associated specifically with those terms.  This, I think, is a mistake.  It’s a mistake analogous to the mistake of trying to capture what is unique about first-person thoughts by focusing exclusively on the linguistic behavior of the pronoun “I”, or the mistake of trying to understand the nature of speech-acts by focusing on the analysis of special vocabulary items – explicit force indicators (like Frege’s assertion sign) – and identifying the illocutionary force that is to be semantically associated with the relevant words.

Thus, imagine (following a suggestion of Anscombe’s) a language lacking the first-person pronoun.  Or imagine (following some suggestions of Frege’s) a ‘force-innocent’ language in which there are neither mood indicators nor words for indicating or making explicit the illocutionary force with which an utterance is put forward.  (In a similar vein, Kripke imagines a semantically innocent language that lacks the truth predicate.)[45]  The lack of the pronoun “I” does not prevent speakers from putting forward claims ‘in the first-person mode’ (as we might put it), let alone having first-person thoughts.  Similarly, the lack of explicit force indicators does not stand in speakers’ ways of performing various speech-acts – just as the lack of a truth-predicate doesn’t prevent Kripke’s imaginary language speakers from speaking truths.  (This is not to deny that the introduction of devices of making explicit enables speakers to do some things they couldn’t do without the devices.)

And now consider an ‘attitude innocent’ language – one lacking the resources for making explicit, or encode, the attitude or state of mind a speaker is expressing with a given utterance.  Speakers in an attitude-innocent language can clearly express various states of mind and attitudes – including evaluative attitudes – for which they have no labels or other linguistic indicators.  Equally, as Davidson pointed out: “Once a feature of language has been given conventional expression, it can be used to serve many extra-linguistic ends; symbolic representation necessarily breaks any close tie with extra-linguistic purpose” (1979: 113).   And just as an acquired interrogative grammatical mood indicator does not stand in the way of a speaker uttering an interrogative sentence with the force of an assertion (as in the case of, e.g., rhetorical questions), so an acquired indicative form and propositional content (as we’ve already seen) does not stand in the way of a speaker a-expressing an evaluative/motivational attitude in uttering a declarative sentence.

We’ve seen (in connection with two of our puzzles), that, once language is in place, with a full suite of expressive vehicles that do s-express individual concepts and structured propositions ‘all on their own’, it remains to be determined what speakers do with such vehicles – what states of mind they a-express when using various linguistic vehicles.  This is not a question that gets answered simply by pairing up words and sentences with their linguistic meanings.  For, as we saw, a speaker may use sentences with different such meanings, as well as use nonlinguistic means, to a-express the same state of mind.  And a single self-ascriptive sentence (“I am feeling sad”) could be used to express the sad feeling, as when one uses it to avow her sadness, or to issue an evidence-based report, as when one merely affirms something of which a therapist has convinced him (not to mention, e.g., using it ironically, sarcastically, and so on).  Similarly, a speaker can express approval or disapproval and various evaluative attitudes using sentences with different meanings, as well as non-linguistic means (by rolling one’s eyes, for just one example).  And sentences using terms such as “good/bad”, “right/wrong” or “ought/ought not” can be used to express a variety of attitudes, and not necessarily to make, specifically, ethical claims.[46]  Moreover, some sentences name explicitly the state of mind that uttering them is designed to betray.  (Avowals are a case in point.)  Others do not.  (Such naming, however, can take different forms; in addition to “I am sad that she left the company”, we have “Sadly, she left the company”, or the more impersonal “It is sad that she left the company”, yielding sentences that are subject to different syntactic and semantic constraints.)  Other sentences – like one word expletives or exclamations, sentences containing slurs, and sentences containing ostensibly normative terms – do not wear the states they are used to a-express on their linguistic sleeve (in this, they are not so different from nonlinguistic gestures and tones of voice that are acquired through enculturation).

Where does that leave the expressivist’s positive contribution to understanding the normativity of various discourses?  If we are to understand the normativity of this or that discourse by appeal to the expression of relevant states of mind or attitudes, we should seek to understand what constitutes putting forth sentences of the discourse in a distinctive, normatively-laden mode.  What is distinctive of normative language is not the vocabulary used, but the way sentences employing the vocabulary are put forth.  In the case of ethical discourse, it might be suggested that the connection to motivation and evaluation, and thus to ethical normativity, is forged when a claim is issued in the practical (as opposed to the theoretical) mode.  But to do so, we have seen, it is neither necessary nor sufficient to use ostensibly ethical vocabulary.  A claim such as “What John did is good/bad” or “Mary ought/not to help her aunt” can be issued in any number of ways that sever the connection – for example, it can be produced as a hypothetical, as in the course of characterizing someone’s ethical theory, or entertained as a possible logical consequence of one’s own or someone else’s view.  None of these examples, I submit, impugns the core expressivist claim that it’s the connection to the evaluative that accounts for the normativity of ethical claims.  (And this claim, I’ve argued is consistent with seeing ethical claims as using sentences with propositional contents as the linguistic vehicles with which ‘normatively-laden’, motivational attitudes are expressed, but which nevertheless possess auto­no­mous meaning – meaning that, moreover, does not settle what in the world – if anything – would make the sentences true.)

* * *

The understanding I have here urged of the role expression plays in both the emergence of meaningful language (whether in phylogeny or in ontogeny), and in its use by linguistic creatures, reveals that the domains of the expressive and the linguistic interact in myriad, complicated ways.  The expressive behaviors of many nonlinguistic animals, and those of prelinguistic children, are social, communicative behaviors, with discernible proto-semantic and proto-pragmatic dimensions, while paradig­matically linguistic behaviors themselves have important expressive dimensions (avowals and ethical discourse being but special cases of the much broader phenomenon). Neither type of behavior falls neatly into one of two mutually exclusive categories: the category of the purely expressive – understood as wholly natural, inadvertent, merely responsive, non-propositional – on the one hand, and the category of the properly linguistic – understood as governed by convention, reason-based, intentional, flexible, reflective, and propositional – on the other hand.  In the various ways I’ve outlined, language emergence, development, and ordinary use seem importantly continuous with nonlinguistic expressive communication, diachronically speaking, and intertwined and integrated with it, synchronically speaking.  What begins its life as purely expressive can acquire stable independent semantic content; and what has autonomous meaning can acquire regular expressive function.

In discussing the puzzle regarding the origins of meaningful speech, I argued that even nonlinguistic creatures, insofar as they are capable of engaging in expressive behavior, can show aspects of their states of mind through complex, though semantically inarticulate expressive vehicles – ones that do not s-express propositions.  Indeed, I suggested that this capacity is directly relevant to the explanation of the emergence of linguistic vehicles possessed of powers of s-expression.  This gives a relatively clear sense to the idea that our language as we find it has expressive origins; s-expression, we might say, is borne of a-expression.  But it would be a mistake to find support in this for the idea that a semantic theory for a mature natural language – a theory that pairs up sentences with what they s-express – should directly reflect these expressive origins.  The work of expression – as well as force – is done not in language but with language.  In this I concur with one early author discussing metaethics’ role in expounding the nature of normative language:

To take our subject matter as we find it, without retouching, is to recognize that normative language is not comparable to the language of chemistry … It does not have a distinctive technical vocabulary … Normative language … is no less rich and various than our total linguistic resources. (Mothersill, 1955: 408)[47]

 

 

REFERENCES*

(* Note: the reference list below is still incomplete.)

Alston, W. (1965). Expressing. In Black, M. (ed.). Philosophy in America. Ithaca, NY: Cornell University Press, pp. 15-34.

Anderson, S. (2004). Doctor Dolittle’s Delusion. New Haven: Yale University Press.

Ayer, A.J. (1936/1946). Language, Truth and Logic. Harmondsworth: Penguin.

Bar-On, D. (2015). Expression: Acts, Products, and Meaning, in Meaning Without Representation: Essays on Truth, Expression, Normativity, and Naturalism (Steven Gross et. al. eds.), Oxford UP, pp. 180-209.

————— (2013a). Origins of Meaning: Must We ‘Go Gricean’?, Mind & Language

————— (2013b). Expressive Communication and Continuity Skepticism, Journal of Philosophy

————— (2013c). Varieties of Expressivism (with Jim Sias), in Philosophy Compass

————— (2012). Expression, Truth, and Reality: Some Variations on Themes from Wright. In Coliva, A. (ed.). Mind, Meaning, and Knowledge: Themes from the Philosophy of Crispin Wright. Oxford: Oxford University Press.

—————  (2010). Expressing as Showing What’s Within: On Mitchell Green’s Self-Expression. Philosophical Books 51(4): 212-227.

————— (2004) Speaking My Mind: Expression and Self-Knowledge, Oxford: Clarendon Press

Bar-On, D., Chrisman, M., and Sias, J. (2015). (How) Is Ethical Neo-Expressivism a Hybrid View? In Ridge, M., and Fletcher, G. (eds.). Having it Both Ways: Hybrid Theories and Modern Metaethics. Oxford University Press.

Bar-On, D., and Chrisman, M. (2009). Ethical Neo-Expressivism. In Shafer-Landau, R. (ed.). Oxford Studies in Metaethics, Vol. 4. Oxford: Oxford University Press, pp. 133-166

Bar-On, D., and Green, M. (2010). Lionspeak: Communication, Expression, and Meaning. In O’Shea, J., and Rubenstein, E. (eds.). Self, Language, and World: Problems from Kant, Sellars, and Rosenberg. Atascadero, CA: Ridgeview.

Bar-On, D., and Priselac, M. (2011). Triangulation and the beasts. In Amoretti, C., and Preyer, G. (eds.). Triangulation: From an Epistemological Point of View. Ontos Verlag.

Bar-On, D., and Sias, J. (2013). Varieties of Expressivism. Philosophy Compass.

Bennett, J. (1976). Linguistic Behavior. Cambridge: Cambridge University Press.

Blackburn,S. (1998). Ruling Passions: A Theory of Practical Reasoning. Oxford: Clarendon Press.

Burge, T. (2009). Primitive agency.

Burling, R. (2005). The Talking Ape: How Language Evolved. Oxford: Oxford University Press.

Cheney, D., and Seyfarth, R. (2007). Baboon Metaphysics. Chicago: University of Chicago Press.

Davis, W. (2003). Meaning, Expression, and Thought. Cambridge, Mass: Cambridge University Press.

Fitch, T. (2010). The Evolution of Language. Cambridge: Cambridge University Press.

Fodor, J. (1998).

Gallagher, S. (2008). Direct perception in the intersubjective context. Consciousness and Cognition 17, pp. 535–543.

Gibbard, A. (2003). Thinking How to Live. Cambridge, Mass.: Harvard University Press.

——————- (1996).

Gómez, J. C. (2004). Apes, Monkeys, Children, and the Growth of Mind. Cambridge, MA: Harvard University Press

——————- (1996). Ostensive Behaviour in Great Apes: the Role of Eye-contact. In Russon, A., Bard, K., and Parker, S., (eds.). Reaching into Thought: The Minds of the Great Apes. Cambridge: Cambridge University Press.

—————– (1998). Do Concepts of Intersubjectivity Apply to Non-human Primates? In Braten, S. (ed.). Intersubjective Communication and Emotion in Ontogeny. Cambridge: Cambridge University Press.

Green, M. (2007). Self-Expression. Oxford: Oxford University Press.

Grice, P. (1957). Meaning. The Philosophical Review 66: 377-88.

Hauser, M. (1999). The Evolution of a Lopsided Brain: Asymmetries Underlying Facial and Vocal Expressions in Primates. In Hauser, M., and Konishi, M. (eds.). The Design of Animal Communication. Cambridge, MA: MIT Press

Hurford, J. (2007). The Origins of Meaning: Language in the Light of Evolution. Oxford: Oxford University Press.

Jackendoff, R. (2002).

———————-  (1999).

Leavens, D., and Hopkins, W. (1998). Intentional Communication by Chimpanzees: A Cross-Sectional Study of the Use of Referential Gestures. Developmental Psychology 34: 813-822.

Mackie, J. L. (1977). Ethics: Inventing Right and Wrong. London: Penguin

Maestriperi, D. et al. (2011). Familiarity affects the assessment of female facial signals of fertility by freeranging male rhesus macaques. Proc. R. Soc. B 278: 3452–3458.

Maynard-Smith, J., and Harper, D. (2003). Animal Signals. Oxford: Oxford University Press.

Millikan, R. (2004a). On reading signs: some differences between us and the others. In Oller, D. K., and Griebel, U. (eds.). Evolution of Communication Systems: A Comparative Approach. Cambridge, MA: MIT Press.

——————- (2004b). Varieties of Meaning: The 2002 Jean-Nicod Lectures. Cambridge, MA: MIT Press.

Pack, A., and Herman, L. (2006). Dolphin Social Cognition and Joint Attention: Our Current Understanding. Aquatic Mammals 32(4): 443-460.

Parr, L. (2011). The evolution of face processing in primates. Philos Trans R Soc Lond B Biol Sci. 366(1571): 1764-77. 38

Parr, L. et al. (2011). Effect of familiarity and viewpoint on face recognition in chimpanzees. Perception 40(7): 863-72.

Pepperberg, I. (2002). The Alex Studies. Cambridge: Harvard University Press.

Pettit, D. (2010). On the Epistemology and Psychology of Speech Comprehension. The Baltic International Yearbook of Cognition, Logic, and Communication, 5: 1-43.

Progovac, L. (2010). Syntax: Its Evolution and Its Representation in the Brain. Biolinguistics 4.2-3: 233-254.

Progovac, L. and Locke, J. (2009). The Urge to Merge: Ritual Insult and the Evolution of Syntax.

Biolinguistics, 3.2-3: 337-354.

Putnam, Hilary. (1975). The Meaning of ‘Meaning’.  In Putnam, Mind, Language and Reality. Cambridge; Cambridge University Press, pp. 215-71.

Recanati, F. (2002). Does linguistic communication rest on inference? Mind and Language 17: 105-126.

Richard, M. (2015). “What Would an Expressivist Semantics Be?” in Meaning Without Representation. S. Gross, M. Williams, and N. Tebben (eds.), Oxford University Press, pp. 137-159.

—————— (2008).  When Truth Gives Out.

Ridge, M. (2006a). Ecumenical Expressivism: Finessing Frege. Ethics 116(2): 302-336. Ridge, M. (2006b). Ecumenical Expressivism: The Best of Both Worlds? In Shafer-Landau, R. (ed.). Oxford Studies in Metaethics, Vol. 2. Oxford: Oxford University Press, pp. 51-76.

Schroeder, M. (2008). What is the Frege-Geach Problem? Philosophy Compass 3/4: 703-720

———————-  (2009). Hybrid Expressivisms: Virtues and Vices. Ethics 119: 257-309

———————-  (2013).

Sellars, W. (1969). Language as Thought and as Communication. Philosophy and Phenomenological Research 29: 506-27

Tomasello, M. (2008). Origins of Human Communication. Bradford Books.

Tormey, A. (1971). The Concept of Expression. Princeton, NJ: Princeton University Press.

 

Notes

[1] For a recent survey of the relevant literature, see Fitch (2010), especially ch. 4.  See also Anderson (2004).

[2]  See, e.g., Burling (2005: 16), Fitch (2010: 392), Hurford (2007: 184ff.), and Tomasello (2008: 17, and passim).

[3] See Anderson (2004: Ch. 2 and passim).  For an extensive discussion of the puzzle of the origins of meaning, see Bar-On (2013a).

[4] For simplicity, I here focus on avowals made in speech.  However, the puzzle – as well as the solution I canvass later – applies to avowals made in speech or in thought.

[5] To use an example from Richard (forthcoming), whose view will be discussed briefly below.

[6] On Ayer’s view, the motivational attitudes all belonged to the emotional, or the conative side of the cognitive/conative divide (hence the epithet “emotivism”).

[7] Perhaps the best-known version of this objection comes under the title “the Frege-Geach problem” (for some discussion, see Bar-On and Chrisman 2009, Bar-On and Sias 2013, Bar-On, Chrisman, and Sias, forthcoming, and Schroeder, 2008.)  More sophisticated expressivist proposals have been made that attempt to address the problem, to which more sophisticated versions of the objection from semantic continuities have been raised.  (For review, see Schroeder 2009.)

[8] See Green (2007: 140-3).

[9] I here set aside, for the most part, what Sellars (1969) calls (misleadingly, I think) ‘expression in the causal sense’ – e.g., uncontrolled facial expressions or gestures that reveal one’s state of mind.  This is because the expressive behaviors relevant to my concerns here – avowals and ethical claims, but also (I’d argue) communicative expressive behaviors that are potential candidate precursors of linguistic expressions – belong in the realm of a-expression.  (See Bar-On 2004: 216f., 249ff., 289, 315.)  These are not nonvoluntary or reflexive bodily happenings, but rather things that are done by an individual (as opposed to a sub-system, or module, within the individual), over which the individual exercises a certain kind of central, executive control.  The individual is an agent doing things as opposed to a mere patient undergoing various processes. (For the relevant notion of agency, see Burge 2009.)      

[10] A word on propositions, relevant to the later discussion of ethical expressivism: to say that sentences express propositions is not to commit to any particular semantic ontology.  ‘Proposition’ is used here as a relatively innocent stand-in for what is sometimes referred to as a (declarative) sentence’s literal meaning: what a good translation of a declarative sentence into other languages aims to preserve, what serves to specify the content that diverse propositional attitudes can share (as in “John thinks but I only hope that the Democrats will win”), and what can be abstracted as shared across sentences in different moods (e.g., “Is it raining?” “It’s raining” “Let it be raining”).   In particular, one can maintain that sentences in a given discourse express propositions, and can thus partake in logical infe­ren­ces and stand in systematic logico-semantic relations to other sentences – and, in particular, can be embedded in negations, conditionals, intensional contexts, etc. – without supposing that the terms in the sentence refer to objects or properties.

[11] See e.g. work by Maestriperi (2011) and Parr (2011); see also Parr et al. (2011).

[12] As an aside: Of course, on a given occasion, an animal can bare its teeth without being angry.  This can happen for any number of reasons, and not necessarily because the animal is trying to deceive.  In such cases, of course, the teeth-baring does not make the anger perceptible.  Showing and seeing are both factive.  But that doesn’t tell against perception.  Compare: We can see a tree by seeing a characteristic component of it – say, one of its branches – even though on occasion, if the branch were severed, we might be seeing the branch without seeing the tree.  If the tree is there, attached to the branch, I can see it by seeing its branch.  Likewise, if our animal is angry, we can see its anger by seeing a characteristic component of it.  (A characteristic component need not be an essential component, nor do we need to suppose that being angry necessarily requires showing one’s anger through behavior.  So the present proposal can be divorced from logical behaviorism.  Moreover, the proposal does not rule out the possibility of expressive failures: e.g., someone showing sadness, yet failing to show her sadness.  For discussion, see Bar-On 2004: 240ff., 310ff., and 410ff.)

[13] As I explain elsewhere (2004, 2010, forthcoming), to express a state of mind is not merely to signal its presence, or to provide information about it, or simply to tell one’s audience about it.  (See, e.g., Alston (1965), and the earlier quotations from Ayer 1936/1946).  If my neo-expressivist account of avowals is right, acts of avowing use vehicles that say (in the sense of s-expressing propositions to the effect) that the avower is in the expressed state.  But even so, this doesn’t mean that the expresser is using the relevant sentence (merely) to tell someone that they are in the state.  It’s not just a matter of the behavior’s being apt to impart knowledge (as opposed to merely providing evidence) about the expresser’s state of mind; it also has to do with the specific way knowledge is imparted.  For our purposes, the crucial contrasts can be put in action-theoretic and epistemic terms.  On the expresser’s side, the relevant contrast – to be explained by an appropriate theory of action – is between engaging in spontaneous behavior that springs immediately from the expressed state of mind, on the one hand, and behavior that the agent engages in only for some specific reason, or with some specific (communicative or other) purpose in mind.  (I thus here depart from Green’s account of showing 2007, aspects of which I criticize in Bar-On 2010.)  On the observer’s side, the relevant contrast is an epistemic one – between, on the one hand, behavior that affords direct recognition of the expresser’s state of mind, and, on the other hand, behavior that can only be treated as mere observational evidence from which the state’s presence and character can be inferred.   An individual engaging in expressive behavior uses some expressive means to show (as opposed to hiding) the presence and character of a state of mind she is in, in the sense of producing spontaneous, uncalculated but intentional behavior that is designed to allow suitably attuned receivers to recognize it immediately.

[14]  This, I’d argue, can also shed light on some of the most successful language-learning protocols – of Gray Parrots, Chimpazees, and Bonobos – and the acquisition by animals of human gestures such as pointing.  It’s at least in part because nonhuman animals are able to show their trainers some of their affective and cognitive states – what holds their attention, where their focus lies, what they want or need, whether they are bored, excited, tired, agitated, alarmed, and so on – that trainers are able to offer them appropriate labels that they can then incorporate into their expressive repertoires.  For one especially striking example, consider Pepperberg’s training of Gray Parrots (see e.g. Pepperberg 2002).  Capitalizing on these birds’ keen interest in various items in the lab, Alex (and later Griffin) were taught (among other things) to say “paper”, “cork”, “corn” – and later “want paper/cork/corn” to request the relevant items and label them correctly, as well the template “wanna x/y/z” (e.g. “wanna go back/eat”) to make various action requests.  The parrot’s productions of English word sounds were not instances of rote, purposeless mimicry; they were goal-directed, novel, and referential.  So they appear to meet standard current definitions of imitation (see Fitch 2010: 162).

[15] When someone blurts out an expletive, or an exclamative, or makes a fully explicit snide remark, or a spontaneous comment on ‘the passing show’, we are put in a position immediately to recognize what state of mind he is giving voice to, with the linguistic vehicles being used to articulate various aspects of the state.

An interesting case to consider, in this context, is the use of emoticons, which Wikipedia helpfully characterizes as follows:  “An emoticon is a metacommunicative pictorial representation of a facial expression which, in the absence of body language and prosody, serves to draw a receiver’s attention to the tenor or temper of a sender’s nominal verbal communication, changing and improving its interpretation. It expresses – usually by means of punctuation marks – a person’s feelings or mood …”.

^[16]  Indeed, it has been argued by a number of authors in recent years that the best model for understanding ordinary use and uptake of language – and not just that of linguistic expressives – is that of perception, as opposed to propositional-inferential information processing.  See e.g. Recanati (2002), Gallagher (2008), and Pettit (2010).

[17] Thus, to go back to our earlier discussion (in 2.2), as far as I can see, nothing in the transition to verbally articulated expression requires retreating to the idea that verbal expressions enable knowledge of expressed states only in virtue of speakers intending to be – or their witnesses interpreting them as – providing conclusive evidence for the presence of the expressed state.  For a criticism of a proposal along these lines (following Green 2007), see Bar-On (2010).

[18]  Barely ‘inside language’, we have e.g. “Pssst”, “Shush”, “Ugh!”.  Then:

• inarticulate interjections (“Oh!”, “Ow!”, “Ooh!”, “Yeah!”, “Yuk!”, “Yum!”, “Blah”, “Ouch!”, “Huh?”)
• expletive words or phrases (“A-hole”, “Damn it!”, “What the heck!”, “Screw you!”, “He is a total sissy”)
• slurs and derogatory terms and phrases, which are often thought to have some descriptive content (“kike”, “kraut”, “moron”, “douche bag”, “piece of sh-t” )
• exclamations (“Super!”, “How cool is that!”, “What a jerk!”)
• intensifiers (“This house is so big!”, “She’s very/really tall”)
• expressive adjectives (“huge”, “enormous”, “tiny”, “gross”, “lovely”, “puny,” “pathetic”, “unbelievable”, “outrageous”; but also “yucky”, “yummy”, “kinda blah”)
• mental-state adjectives (“amazing”, “surprising”, “painful”, “boring”, “sad”, “tiring”, “annoying”, “disgusting”, “gross”)
• exclamatives such as “What a fantastic performance that was!”, “How cleverly he dodges the ball!”
• adverbial modifiers (“She’s amazingly smart”, “He’s surprisingly quiet”, “Hopefully/Unfortunately he won’t get there on time”)
• constructions such as “It’s amazing/sad/surprising that p”, or “It’s painful/tiring/annoying to Ҩ”, and “Ҩ’ing is annoying/sad/….

As has been increasingly recognized in recent literature, proper understanding of the role of such devices requires appreciating their expressive dimension. On some views, such understanding can potentially shed light on a hypothetical earlier stage of human language – so-called protolanguage.  For, at least some linguistic expressives can be regarded as ‘fossils’ of nonlinguistic vocalizations or gestures that may have preceded language (phylo­genetically speaking).  See e.g. Jackendoff (1999) and (2002).  See also Progovac (2009) and (2010).

[19] Materials in this section overlap parts of Bar-On (2013a) and (2015).

[20] For some descriptions and analyses of facial, vocal, postural and gestural expressions, in addition to Darwin’s classic work, see e.g. Juan Carlos Gómez (1996, 1998, 2004), Leavens and Hopkins (1998), Hauser (1999), Pack and Herman (2006), and Cheney and Seyfarth (2007).  (Note that at least some of the behaviors on the list can be plausibly characterized as only expressing in (Sellars’) ‘causal’ sense.  See footnote 9.)

[21] See, for example, Grice (1957), Alston (1965), and Bennett (1976).  But see also, in the animal communica­tion literature, e.g. Maynard-Smith and Harper (2003: ch. 7), Anderson (2004: ch. 2), and Fitch (2010: ch. 4).

[22] For the account of expressive behavior that follows, see Bar-On (2004).  See also Green (2007) (though see Bar-On (2010) and Bar-On & Sias, in progress, for some reservations), and Bar-On and Green (2010).

[23] For relevant references and discussion, see Bar-On and Green (2010), Bar-On and Priselac (2011), and Bar-On (2013b), and see below.

[24] For an early occurrence of the idea that expressive behavior shows what’s within while pointing to what’s without, see Tormey (1971: 27f.).

The Janus-face character discussed here is different from the dual force ascribed by Millikan to ‘pushmi-pullu’ representations; see Millikan (2004a, 2004b).

[25] And see 3.3 below for summary.

[26] Or thought-token, if the claim is made in thought rather than speech.  In what follows, I will drop this qualification.

[27] While it seems natural to speak of sentences as the products of expressive acts performed in speech, it may seem less natural to speak of facial expressions, bodily demeanor, etc. as products (though it could be argued that an analogous type-token distinction could be drawn there, too).  Even so, sentences, like facial expressions, etc. can be thought of as vehicles of expression.  So we can perhaps tentatively think of the act/product distinction as a special case of the more general act/vehicle distinction expounded earlier.  (Thanks to Peter Hanks for pressing me on this point.)

[28] Bar-On (2004b), (2007), (2012).

[29] Notably, the sentence “I am feeling sad” can be used by me at the conclusion of a therapy session as a mere evidential report of my state.

[30] For a full development of this account and an explanation of the strong presumption of truth governing avowals, see chs. 6-8 of Bar-On 2004.  In Ch. 9, the status of avowals as articles of knowledge is discussed at length.

[31] Cf. Bar-On and Chrisman (2009), Bar-On and Sias (2013), and Bar-On, Chrisman, and Sias (forthcoming).

[32] Though it’s not conceptually impossible for someone to make a mental self-ascription and not be in the self-ascribed state, it is a propriety condition on acts of avowing that one is in the self-ascribed mental state.  The neo-expressivist explanation of the motivational contrast between ethical claims and descriptive claims can be seen as analogous.  Someone who is making an ethical claim (as opposed to producing a descriptive report on some state of affairs) is a-expressing the relevant motivational attitude – the very attitude whose presence would explain why she is suitably motivated.  Even if it’s not concep­tual­ly impossible for someone to make an ethical claim without having the relevant attitude, having the attitude can be seen as a propriety condition on making ethical claims.  So someone who makes the claim while lacking the attitude is violating a propriety condition on acts of ethical claim making.   And adequate mastery of ethical discourse and ethical concepts requires grasp of this propriety condition.  In Bar-On and Chrisman 2009, we argued that this provides resources for capturing a fairly strong “internal” connection between ethical claims and action, as well as providing a more nuanced array of diagnoses of different ways the connection between making an (apparently) ethical claim and motivation can be broken.

[33] As Ayer already saw (1946: 104–8), the expressivist insight is best captured without supposing that the vehicles used in making ethical claims s-express propositions that self-ascribe those states.  So it’s important to note that the neo-expressivist is not claiming that ethical claims are themselves avowals!

[34]  Thus, ethical neo-expressiv­ism breaks away from an assumption shared by traditional expressivists and their opponents, namely, that expressivism requires the expressive function of ethical claims to be somehow executed through the literal semantics of ethical sentences.  Instead, it maintains that the relevant expression occurs in acts of making ethical claims – in their use or function to (a-)express motivational attitudes.

Note that neo-expressivism does not purport to settle the question which psychological states qualify as motivational.  Humeans will insist that they must be noncognitive; others may demur.  Moreover, even if one sides with the Humeans, and insists that one who makes an ethical claim is a-expressing a noncognitive motivational attitude, it’s still possible to allow that one is also a-expressing a belief whose content is given by the proposition that is s-expressed by the sentence used.  For some discussion, see Bar-On and Chrisman (2009).

[35] Expressivism is often applied to a particular area of discourse to allow the expressivist to avoid unwanted ontological commitments.  So for instance, ethical expressivists are often motivated by a desire to avoid commitment to (what they see as) “spooky” irreducibly normative properties (see e.g. Mackie 1977).  But it’s worth noting that, despite its common association with the negative ontological thesis, even traditional expressivism does not entail anti-realism.  Just like traditional expressivism, neo-expressivism does not entail anti-realism, but it can still accommodate anti-realism in virtue of its ontological neutrality.

To be sure, there are many who will want to couple the idea that sentences in a given area s-express truth-evaluable propositions with a metaphysically inflated conception of truth, thereby interpreting the neo-expressivist as committed to the idea that there are ways the world might be that would make the relevant propositions true.  But this is not essential to the propositional-compositional semantics that the neo-expressivist endorses.  Maintaining that a sentence expresses a proposition is quite different from maintaining that any of the terms in that sentence refer.  In the case of avowals, the neo-expressivist will want to say that mental-state terms refer (see Bar-On, 2012); but in other domains, neo-expressivists may be anti-realists.  Whether or not the propositions s-expressed by claims in a given domain have ontologically problematic truth-makers is a matter to be settled by metaphysicians, not semanticists.

[36] Or language-like.

[37] At most only ‘single-criterion’ terms, if Putnam was right (see Putnam 1975).  Likewise precious few concepts admit of decomposition into more basic concepts (see, e.g. Fodor 1998.)

[38] See e.g., Gibbard (1996 & 2003), Blackburn (1998), Ridge (2006a, 2006b), and Richard (2008 & forthcoming).

[39] When it comes to non-ethical sentences, ideationalists often claim that a sentence such as “It’s raining” expresses the belief (type) that it’s raining without expressing the belief (token) of any particular individual.  But this doesn’t seem plausible.  What is in common among attitude ascriptions like: (i) John believes that it’s raining, (ii) John hopes that it’s raining, (iii) John fears that it’s raining, (iv) John doubts that it’s raining, (v) John suspects that it’s raining, it seems, is some content, and not a type of belief (or any other attitude) with that content.  (Moreover, one might wonder: why single out belief as the relevant type of attitude that is held constant across attitudes?)  What one wants is a notion of content that is attitude neutral, one that abstracts away from attitude type (as well as, relatedly, abstracting from speech-act type) – precisely something like the conventional notion of a proposition understood as Austinian locutionary content, or a Fregean thought (minus the Platonist ontological commitment).  (Thanks to Dean Pettit.)

It is revealing that philosophers otherwise drawn to ‘mentalist’ conceptions of meaning have nonetheless found reason to resort to essentially abstract, non-mental notions of content (see e.g. Davis 2003).

[40] See Richard (2015 p. XX).

[41] See, e.g., Schroeder (2008, 2009).  However, Schroeder (2013) considers what he takes to be a more promising line for expressivists to take, aspects of which closely resemble the neo-expressivist proposal expounded in Bar-On and Chrisman (2009).

[42] Without, for example, excluding the possibility that ethical claims have their own proprietary (natural or nonnatural) truth-makers by maintaining that ethical sentences are not truth-evaluable in the first place, simply in virtue of their semantics.  (For discussion, see Bar-On and Sias 2013.)

[43] Relatedly, on pain of strong commitment to verificationism, it should not be assumed to be part of the semantic competence of the average English speaker to be able (even tacitly, or in principle) to offer such paraphrases.  Note that, to insist that semantic analysis should adhere to ontological innocence is only to say that it cannot be expected to tell us what in reality sentences in any given domain are about (beyond disquotational specification).

[44] For discussion, see Bar-On et al. (2015), and Bar-On (forthcoming).

[45] In a different vein, Millikan hypothesizes that, evolutionarily speaking, purely descriptive (like beliefs) and purely directive representations (like desires), are downstream from more primitive ‘pushmi-pullu’ representations (e.g., a food call of a hen to its brood is a primitive, which is at once a description – ‘food here’ – and a directive (‘come get it!’).  Discrete, ‘pure’ representations, Millikan argues, “require more sophisticated cognitive apparatus” (166).

[46] At times, possibly not even to make any normative claim.

[47] Very special thanks to Carol Voeller for many discussions and comments on several drafts.  Thanks, too, to Simon Blackburn, Matthew Chrisman, Mitch Green, Ram Neta, Dean Pettit, and Jim Sias, as well as to audiences at the Arché workshop on Normative Language in St. Andrews (October 2013), the conference on Global Expressivism, Szczecin, Poland (August 2014), and colloquia at Stockholm, Leipzig, and Potsdam Universities (2015-16).