New paper - The anatomy of iconicity✨
Our new paper on unpicking the iconicity of ideophones is out in Open Mind!
You can find and cite the paper here:
Punselie, S., McLean, B., & Dingemanse, M. (2024). The Anatomy of Iconicity: Cumulative Structural Analogies Underlie Objective and Subjective Measures of Iconicity. Open Mind, 8, 1191-1212.
This was part of a really wonderful BA project by Stella Punselie, supervised by Mark Dingemanse, which I was lucky to be invited to collaborate on just for the final push. The figure below (Figure 2 from the paper) shows the general idea. We combine three different measures of iconicity: two behavioural (iconicity ratings and guessing accuracies, panels B and C) and one qualitative (Stella and Mark’s analyses of iconic mappings in the stimuli, which draws heavily on Structure-Mapping theory (Gentner, 1983), panel A). The stimuli in question for this study were ideophones from five languages representing four language families: Japanese (Japonic), Ewe (Kwa, Niger-Congo), Korean (Koreanic), Semai (Aslian, Austroasiatic) and Siwu (Na-Togo, Niger-Congo).
Figure 2. Overview of the triangulation method. Structure mapping grounds the notion of iconicity, ratings capture subjective form-meaning fit, and guessability provides a baseline of guessing accuracy. For instance, Korean tuˈgɯndugɯn ‘heartbeat’ features at least three levels of iconic form-meaning correspondences (A). This predicts it should be rated as highly iconic (B) and should be more guessable than words with fewer detected correspondences (C). Consistent with this, it was rated 4.8 for subjective iconicity and guessed at 80% accuracy. Arrows indi- cate how measures predict (black) and inform (grey) each other.
Rating tasks and guessing experiments are pretty well established methods for putting a numeric value on iconicity. The novel contribution in this paper is to triangulate these two quantitative measures with qualitative analyses of iconic mappings in the stimuli using Structure-Mapping theory.
The table below, table 1 from the paper, shows how the iconicity as structure-mapping analysis works:
There are 9 form features, which map to 7 semantic features (there are more formal than semantic features because three of the formal features relating to consonant, vowel, and tone weighting map to a single semantic feature of magnitude). Ideophones are coded first for their formal features, and then separately for their semantic features. When the formal and semantic features align as shown in the table, we consider this an instance of iconicity. I really love this analytical framework for the following reasons:
- It’s explicit about how exactly a particular mapping is iconic.
- These iconic mappings can be informed by data independent to the data under study, and created prior to seeing it. The iconic mappings in Table 1 were informed by prior work on ideophones by pioneers like Diedrich Westermann (for West-African ideophones), Gerard Diffloth (Southeast Asian ideophones), and Roman Jakobson (for Indo-European languages). They do not come from analysis of the ideophones in our study.
- It creates testable hypotheses.
- It leaves room for revision and expansion.
Qualitative analyses of iconicity are often criticised for being vague or wishy-washy, and lacking empirical grounding. After all it’s very easy, after the fact, to come up with ways in which a particular form-meaning mapping could be construed as iconic. We hope to address these issues with our new approach.