Does language rule perception? Testing a radical view of linguistic relativity.

Abstract

To investigate whether language rules the visual features that can be discriminated (a radical assumption of linguistic relativity), we examined crosslinguistic differences between native Korean and German speakers during liminal perception of a target disk that was difficult to perceive because its visibility suffered from masking by a ring that followed and enclosed the target disk (metacontrast-masking). Target-mask fit varied, with half of the masks tightly and the other half loosely encircling the targets. In Korean, such tight versus loose spatial relations are semantically distinguished and thus highly practiced, whereas in German, they are collapsed within a single semantic category, thus are not distinguished by language. We expected higher sensitivity and greater attention to varying spatial target-mask distances in Korean than in German speakers. This was confirmed in Experiment 1, where Korean speakers consistently outperformed German speakers in discriminating liminal metacontrast-masked stimuli. To ensure that this effect was not attributable to generic differences in attention capture or by language-independent differences between participant groups, we investigated stimulus-driven attention capture by color singletons and conducted a control experiment using object-substitution masking, where tightness of fit was not manipulated. We found no differences between Korean and German speakers regarding stimulus-driven attention capture or perceptual sensitivity. This was confirmed in Experiment 3, where we manipulated types of masking within participants. In addition, we validated the tightness-of-fit manipulation in a language-related task (Experiment 4). Overall, our results are consistent with linguistic relativity, namely its assumed generalized language influences in nonlinguistic perceptual tasks.