Published Paper Details:

Lever handles are an important product category in architectural hardware. It is a challenging to conceptualize and its design has been popularly undertaken by many famous designers. Surface level aesthetics as well as the semantic meaning are vital concerns when designing lever handles. Here, sound-shape correspondence is an important determinant of the appropriateness of semantic fit between a product name and its form. In this study, 5 handle designs were assessed for the extent of their sound-shape correspondence by 21 university-level undergraduate design students. Then, using ChatGPT 4o and Midjourney tools, new handle designs were generated to semantically represent the original handle names. These new designs were also assessed for the extent of their sound-shape correspondence. Finally, the correspondence values of original and new handle designs were compared using paired-sample Student's t-test. Only 1 original handle form was found to have "good correspondence" with its actual name (Dana). However, all the 5 new handle designs were found to have "good correspondence" with their corresponding names (i.e. Axor, Dana, Primo, Torso, Zenith). It was found that the average ratings of correspondence between all the new handle designs and their original names were higher than that of the original designs. This difference was found to be statistically significant as well (p < 0.05). An important observation was that out of the 5 handles, the original design of Zenith was found to have good correspondence with 4 out of the 5 handle names; original Dana with 3 out of the 5, and new Axor with 3 out of the 5 handle names. In the results, certain sound-shape perceptions were perceived better than the others. Importantly, it was established that the use of generative AI tools augmented with expert human decision making could be successfully used as a process to conceptualize product designs with good sound-shape correspondences.