“Undoing” a Rhetorical Metaphor: Testing the Metaphor Extension Strategy

ABSTRACT

Political metaphors do more than punch up messages; they can systematically bias observers’ attitudes toward the issue at hand. What, then, is an effective strategy for counteracting a metaphor’s influence? One could ignore or criticize the metaphor, emphasizing strong counterarguments directly pertaining to the target issue. Yet if observers rely on it to understand a complicated issue, they may be reluctant to abandon it. In this case, a “metaphor extension” strategy may be effective: Encourage observers to retain the metaphor but reinterpret its meaning by considering other, less obvious implications. The current studies support this claim. Under conditions where participants gained a strong (versus weak) epistemic benefit from a rhetorical metaphor, they were more persuaded by a rebuttal that extended (versus ignored or criticized) that metaphor. The studies use converging operational definitions of epistemic benefit and offer insight into how political attitudes are made and unmade.

Funding

This material is based on work supported by the National Science Foundation under Grant Number BCS-1222047.

Notes

¹ That’s why we put “undo” in quotes: Technically, the extension strategy retains the original metaphor at a general level but changes the implications of its mapping.

² We examined the percentage of participants in the mapping salience condition who made at least one error in the mapping task (i.e., incorrectly answered one of the four items). The error rate in the Preliminary Study was n = 14 (9%); in Study 1, n = 13 (8%). All the predicted between-condition differences remained significant regardless of whether these participants’ data were included in the analysis (all ps < .05). That is, the primary predicted results were unaffected by the inclusion of participants who made at least one error on the mapping task.

³ As we anticipated, participants frequently chose this “Not sure/Uncertain” option when faced with challenging questions. That is, within the high target uncertainty condition (n = 121): only six participants (5%) never selected this option; 99 (82%) selected this option for more than half of the questions; and almost half (48%) chose it for all or all but one question.

Additional information

Funding

This material is based on work supported by the National Science Foundation under Grant Number BCS-1222047.