UC Berkeley

When we say that a dress looks blue, or that a musical instrument sounds off-key, or that a soup smells like it contains nutmeg, what do we communicate? With claims about appearances like these we seem to communicate both about the objective world and about our subjective experience. This comes out in two puzzling features of appearance claims.

Faultless disagreement arises when speakers disagree, and yet neither seems to be mistaken. Speakers may faultlessly disagree about appearances, for instance if one holds that a dress 'looks blue', and the other that it doesn't. There is a felt incompatibility here, just as with disagreement over objective claims. But assuming the speakers have different visual experiences of the dress, neither seems to be mistaken. Which appearance claim a speaker correctly makes depends not just on the objective world, but on their subjective experience as well. Faultless disagreement thus precludes viewing appearance claims as straightforwardly objective or subjective.

The acquaintance inference is the inference from an utterance to the conclusion that the speaker has relevant first-hand acquaintance. For instance, if a speaker says that the dress looks blue, one will infer that they have seen it. The utterance is infelicitous if they haven't. However, this inference is not an ordinary entailment. Just because I haven't seen the dress, doesn't mean it doesn't look blue. Again, this phenomenon precludes taking appearance claims to be straightforwardly objective or subjective.

I defend an expressivist analysis of appearance claims, on which they are used to express speakers' experiential states. On this view, faultless disagreement arises when speakers express incompatible experiential states, while nonetheless expressing experiential states they are in fact in. And the acquaintance inferences arises because when a speaker makes an appearance claim, one can infer that they are in an experiential state of the sort expressed by the utterance.

My analysis covers not only appearance language, but experiential language more generally, which encompasses both appearance language and the evaluative language of personal taste (e.g. 'tasty', 'interesting'). Indeed, both faultless disagreement and the acquaintance inference have been associated primarily with evaluative vocabulary. I argue, however, that these features are not especially associated with evaluative language. The language of personal taste falls in the intersection of the evaluative and the experiential; but these puzzling features are due to experientiality. The investigation of appearance language is crucial for identifying the source of these features, for it includes claims that are experiential but not evaluative.

Appearance discourse also offers insight into epistemic notions, like adequacy of evidence. This comes out in my investigation of the acquaintance inference with appearance claims, which examines behavior with no analogue in the more widely-discussed evaluative cases. Some appearance claims (e.g. 'Tom looks like he's cooking') require acquaintance with a specific stimulus (Tom), while others (e.g. 'Tom looks like he's well organized') just require acquaintance with something evidentially-relevant (like Tom's clutter-free office). Making use of experimental work, I argue that these two forms of acquaintance inference display our sensitivity, in discourse, to fine-grained evidential distinctions, for instance between transient properties (like cooking) and standing ones (like being well-organized). Appearance claims can thus serve to express our evidence. This integrates smoothly with the expressivist analysis I offer, as experience is a source of evidence about the world. Thus, in expressing experiential states, we can at the same time express our evidence.