Mitterer, H., Schuerman, W. L., Reinisch, E., Tufvesson, S., & Dingemanse, M. (2012). The limited power of sound symbolism. Talk presented at the 18th Annual Conference on Architectures and Mechanisms for Language Processing (AMLaP 2012). Riva del Garda, Italy. 2012-09-06 - 2012-09-08.
In defiance of the assumed design principle of language of arbitrariness between sign and signified, many languages
use ideophones, which are depictive words for sensory imagery. The form-meaning mappings in ideophones have been
variably hypothesized to be language-specific, universal, or a mixture of both. We test the claim of universality, and in
particular, the claim that ideophones “do the work of representation by phonetic means” (Tedlock, 1999). In support of
this claim, recent research shows that naive listeners can consistently map certain sounds to certain meanings in nonce
words, leading to claims that such mappings may underlie the evolution of language (Ramachandran & Hubbard, 2001).
Given the theoretical weight ascribed to sound-symbolism in language, it is important to know whether ideophones
could live up to this promise. The "recognizability" of these mappings may be due to both segmental and suprasegmental
properties of the stimuli. While the segmental properties tend to be singled out, prosodic aspects have not been
investigated yet. To critically evaluate the power of lexicalised sound-symbolism in ideophones, we recorded over 200
ideophones from five semantic categories (Sound, Motion, Texture, Visual Appearance, and Shape) and from five
languages (Japanese, Korean, Semai, Siwu, Ewe, representing four language families).