Abstract
This study investigated the relationship between writing systems and sound symbolic associations. This was achieved by evaluating the results of an experiment, which measured distributional properties of participants’ writing and the congruence of their responses with the previously established sound symbolic association patterns. The participants were grouped by the primary writing system of their first language (L1) to allow for the cross-script comparison. Three groups of N = 20 participants were included, one for users of each: Latin scripts, Chinese simplified characters and Thai script. The experiment conducted involved two tasks. The participants first completed a task involving writing down sounds of non-word audio stimuli using their respective writing systems. The second task followed an established model of matching images to audio stimuli. The individual results obtained were analysed to derive two distributional metrics: the average token count measuring the number of written representations per speech sound (for Latin and Thai scripts) or speech sound sequence (for Chinese simplified characters) and the entropy of the distribution of those tokens. These two metrics were then tested for correlation with scores in the sound symbolic association task. The statistical analysis of the results confirmed negative correlations of both average token count and calculated entropy with scores obtained in the sound symbolic association task, both with the strength of approximately r = - 0.52 and the significance p = 0.02, in the Latin script group only. Additionally, the negative trend of increase in distributional properties’ scores accompanying decrease in sound symbolic associations test scores was observed in cros-group comparison. Taken together, these results confirm the previously observed association between the writing system itself and sound symbolism, and provide a new insight into the possible (conditionally) correlated factor of distributional properties. The study concludes that the chosen metrics, along with other L1- and usage-based factors constitute good candidates for further examination in terms of relationship with sound symbolism and broader iconicity.