Facts and opinions: Effects of expertise and social consensus on how we think about perspective-sensitive language
(Partly based on joint work with Deniz Rudin, USC Linguistics)

The information that we encounter on a daily basis involves both objective facts about the world and people’s subjective opinions. This distinction is also reflected in language: Words that express opinions (e.g. fun, amazing) differ from words conveying more objective facts (e.g. wooden, Californian): Subjective adjectives are perspective-sensitive and reflect someone’s opinion/attitude. Indeed, it has been widely noted that when two people disagree about matters of taste, neither is in the wrong: There is nothing contradictory when one person says "The rollercoaster was scary!" and the other responds "No, it was not scary" – in contrast to disagreements about objective facts. This phenomenon – central for theories of subjective language – is called Faultless Disagreement. It is assumed to stem from subjective adjectives having a special semantics. However, in this talk, I present a series of experiments (joint work with Deniz Rudin) showing that people’s judgments of faultless disagreement – in other words, how willing we are to accept two divergent opinions as both being valid – are modulated by contextual and social considerations that go far beyond the adjective itself.

First, I show that the extent to which people accept two divergent views as both being right depends on the prevalence of opinions in a community. For example, disagreements about divisive foods (e.g. anchovies, blue cheese in a U.S. context) are rated more faultless than disagreements about widely-liked foods (e.g. donuts, pizza). Our results suggest that unpopular opinions (e.g. "donuts are disgusting") are viewed as more 'wrong' than more prevalent opinions (e.g. "anchovies are disgusting"). Second, I provide evidence that people yield to expertise: When individuals who respect expertise in a given domain (e.g. wine) encounter a disagreement between two wine experts, they are more likely to say both people are right compared to a disagreement between an expert and a non-expert. Thus, people's reactions to subjective disagreements cannot be reduced to the adjective's semantics or even the lexical content of the sentence, but reflect more complex contextual and social factors. Third, time permitting, I will present recent work on how linguistic packaging influences our ability to recognize subjective information.

As a whole, these results suggest that (our ability to perceive) the distinction between ‘opinion’ and ‘fact’ is less clear than one might expect, and is sensitive to multiple sources of information.