An inconclusive study comparing the effect of concrete and abstract descriptions of belief-inconsistent information.

Linguistic bias is the differential use of linguistic abstraction (as defined by the Linguistic Category Model) to describe the same behaviour for members of different groups. Essentially, it is the tendency to use concrete language for belief-inconsistent behaviours and abstract language for belief-consistent behaviours. Having found that linguistic bias is produced without intention or awareness in many contexts, researchers argue that linguistic bias reflects, reinforces, and transmits pre-existing beliefs, thus playing a role in belief maintenance. Based on the Linguistic Category Model, this assumes that concrete descriptions reduce the impact of belief-inconsistent behaviours while abstract descriptions maximize the impact of belief-consistent behaviours. However, a key study by Geschke, Sassenberg, Ruhrmann, and Sommer [2007] found that concrete descriptions of belief-inconsistent behaviours actually had a greater impact than abstract descriptions, a finding that does not fit easily within the linguistic bias paradigm. Abstract descriptions (e.g. the elderly woman is athletic) are, by definition, more open to interpretation than concrete descriptions (e.g. the elderly woman works out regularly). It is thus possible that abstract descriptions are (1) perceived as having less evidentiary strength than concrete descriptions, and (2) understood in context (i.e. athletic for an elderly woman). In this study, the design of Geschke et al. [2007] was modified to address this possibility. We expected that the differences in the impact of concrete and abstract descriptions would be reduced or reversed, but instead we found that differences were largely absent. This study did not support the findings of Geschke et al. [2007] or the linguistic bias paradigm. We encourage further attempts to understand the strong effect of concrete descriptions for belief-inconsistent behaviour.