Ideophone mechanics: a cross-linguistic study of iconic meaning mapping

Ideophones are an open word class found in many of the world's languages whose members depict sensory imagery using non-arbitrary association between sound and meaning (Dingemanse, 2012; Voeltz and Kilian-Hatz, 2001). These associations are based on general iconicity (relationships of resemblance) and language-internal systematicity (statistical correspondences).

In English, ideophones usually 'imitate' sounds, visual patterns or movement - 'woof-woof', 'higgledy-piggledy' and 'zig-zag' - and are used sparingly. In other languages such as Japanese, they have a much more important role in the language and are used to depict smell, taste, bodily feelings, emotions, colours and texture as well.

This project aims to (1) improve and increase our typological data on ideophones and (2) understand the structure and mechanism of their iconic meaning mapping. The hypothesis, outlined in my MSc thesis, is that the sound-meaning association used to create and interpret ideophones is a non-arbitrary mapping between contrastive (i.e., phonologically accessible) features and sensory meaning.

Previous studies on ideophones and iconic mapping assume that phonemes - not phonological features - are the sound units associated with meaning. This would mean that, for instance, the phonological units associated with ideophone 'moo' should be the two phonemes 'm' and 'oo'. My hypothesis, on the other hand, is that meaning is mapped onto the phonological features of 'm' and 'oo': involvement of the lips word-initially, nasal sound, lower pitch vowel quality and word-final long vowel.

Though similar hypotheses have been used to explain general iconicity in signed languages (Taub, 2001; Emmorey, 2014), there have only been two attempts to understand iconicity mapping in ideophones (Nuckolls, 2019; Thompson and Do, 2019) - neither of which used cross-linguistic comparison.

According to several authors (Gentner and Markam, 1997), construal domains - in this case between phonological sound and sensory images - have a one-to-one mapping in the brain. Therefore, if sensory features were mapped onto phonemes there could only be two meaning associations for 'moo': one for 'm', one for 'oo'. On the other hand, if phonological features were the basic unit of meaning mapping, 'moo' could have up to eight.

This would better account for the diversity of ideophones expressing similar sensory images found in languages with large ideophone lexicons.

This study aims to analyse and compare ideophone systems across the world's languages to develop the first comprehensive model of their iconic meaning mapping. To achieve this, the project has two main components: (1) fieldwork documentation of previously uninvestigated ideophone systems and review of existing ideophone research and (2) comparative and theoretical analysis of iconic meaning mapping.