Grounding language in the neglected senses of touch, taste, and smell

ABSTRACT

Grounded theories hold sensorimotor activation is critical to language processing. Such theories have focused predominantly on the dominant senses of sight and hearing. Relatively fewer studies have assessed mental simulation within touch, taste, and smell, even though they are critically implicated in communication for important domains, such as health and wellbeing. We review work that sheds light on whether perceptual activation from lesser studied modalities contribute to meaning in language. We critically evaluate data from behavioural, imaging, and cross-cultural studies. We conclude that evidence for sensorimotor simulation in touch, taste, and smell is weak. Comprehending language related to these senses may instead rely on simulation of emotion, as well as crossmodal simulation of the “higher” senses of vision and audition. Overall, the data suggest the need for a refinement of embodiment theories, as not all sensory modalities provide equally strong evidence for mental simulation.

KEYWORDS:

• Embodiment
• grounded language
• mental simulation
• touch
• taste
• smell

Acknowledgements

We thank Shirley-Ann Rueschemeyer, Bodo Winter, and one anonymous reviewer for comments on the manuscript.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 The interaction between modality and switch-type (switch vs. no-switch) was marginally significant, F(2, 148) = 2.55, p = .08, $n_p^2$ = .033. Follow-up t-tests found a difference between switch and no-switch trials for auditory, t(74) = 2.34, p = .02; and visual t(74) = 2.75, p = .007 items, but not tactile ones, t(74) = −.4, p = .69. Data kindly shared by the authors.

2 The interaction between modality and switch-type (switch vs. no-switch) was marginally significant, F(2, 118) = 2.43, p = .09, $n_p^2$ = .040. Follow-up t-tests found a switch cost for auditory items, t(59) = 3.17, p = .002; but not visual t(59) = .64, p = .53 items, or tactile ones, t(59) = −.24, p = .81. Data kindly shared by the authors.