Processing Scalar Inferences in Face-Threatening Contexts

Processing Scalar Inferences in Face-Threatening Contexts Jean-Franc¸ois Bonnefon (bonnefon@univ-tlse2.fr) CNRS and Universit´e de Toulouse Toulouse, France Wim De Neys (deneys@univ-tlse2.fr) CNRS and Universit´e de Toulouse Toulouse, France Aidan Feeney (a.feeney@qub.ac.uk) Queen’s University Belfast Belfast, Northern Ireland Abstract contexts, for reasons of politeness; but they did not elucidate how individuals adopted this interpretation. This is the ques- tion we will address. Depending on politeness considerations, the quantifier ‘some’ can receive a broad interpretation (some and possibly all) or a narrow interpretation (some but not all). Face-threatening statements such as ‘some people hated your speech’ encour- age the broad interpretation that everyone hated the speech. Because previous research showed that broad interpretations are normally faster and easier, politeness should be easy to process, since it would encourage what is normally the eas- ier interpretation of the statement. Using response time mea- sures and a cognitive load manipulation, this research shows that just the opposite is true: Face threatening contexts encour- age the broad interpretation of ‘some’ while making it longer and more difficult to reach. This result raises difficulties for current cognitive theories of pragmatic inferences. Keywords: Scalar inference; politeness; processing; response time; cognitive load. Face-Threatening Contexts Face is a sense of public self-esteem that all of us project, and are motivated to support, in social interactions. Many actions, called face-threatening acts, can induce a loss of face (e.g., apologizing, or criticizing). Performing such an action often requires the use of a politeness strategy that mitigates the face threat (Brown & Levinson, 1978/1987). One of these strategies is to hedge by means of a scalar term. For example, instead of bluntly asserting (2-a), speak- ers may politely hedge as in (2-b): The Inference from ‘Some’ Experimental pragmatics engages in an experimental inves- tigation of the mental processes involved in inferring what people mean from sentences, contexts, and the implicit prin- ciples that govern the use of sentences in context (Noveck & Reboul, 2008). The drosophila of experimental pragmatics is the contextual inference from ‘some’ to some but not all. Although ‘some’ is semantically consistent with the broad in- terpretation some and possibly all, it is commonly given the narrow interpretation some but not all. For example, most adult speakers of English would assume the assertion of (1-a) to convey that the speaker believes (1-b): a. b. Some of the guests brought wine. Not all of the guests brought wine. The inference from (1-a) to (1-b) is a scalar inference (Horn, 1984), stemming from the ordered informativeness scale . From a Gricean perspective (Grice, 1989), if a cooperative speaker were in a position to assert (1-b), he or she would do so, and would not use the less informative (yet logically consistent) wording (1-a). A key theoretical issue within experimental pragmatics is to explain why and how people adopt the narrow interpreta- tion of ‘some’ in some contexts but not in others. Bonnefon, Feeney, and Villejoubert (2009) in particular showed that broad interpretations were more likely in face-threatening a. b. You are wrong. You are possibly wrong. When they hear (2-b), individuals tend to interpret ‘possibly’ as denoting a high probability, because they do not construe this term as a genuine expression of uncertainty, but rather as a polite way to preserve the face of the listener (Bonnefon & Villejoubert, 2006; Pighin & Bonnefon, in press). In a sense, people usually adopt a narrow interpretation of ‘possibly’ that eliminates high probabilities, but they switch to a broad inter- pretation (that includes high probabilities) in face-threatening contexts. Comparably, instead of bluntly asserting (3-a), speakers may politely hedge as in (3-b): a. b. All of the guests hated the meal you cooked. Some of the guests hated the meal you cooked. When they hear (3-b), individuals tend to disregard the scalar inference attached to ‘some’ and to reach a broad interpreta- tion. That is, when X in ‘some X-ed’ threatens the face of the listener, individuals find it likely that the speaker might have meant that all X-ed, but used the scalar term nonethe- less, out of politeness (Bonnefon et al., 2009). As we will now explain, the current article aims at elucidating the time and effort it takes to reach that interpretation, and thus at elu- cidating whether politeness is easy or hard to process.