[PS-1.1] A picture is worth two thousand words: Visual complexity in morphographic word recognition

Miwa, K. 1 , Hendrix, P. 1 , Allen, D. 2 , Dijkstra, T. 3 & Baayen, H. 1, 4

1 Eberhard Karls Universitaet Tuebingen, Germany
2 Ochanomizu University, Japan
3 Radboud University Nijmegen, Netherlands
4 University of Alberta, Canada

In this study, we took a new perspective to study recognition of morpholographic words. Namely, Japanese kanji characters were considered to be pictures as well as words. In addition to stroke counts considered in past studies as a measure of feature-level complexity, various other types of visual complexity were studied: the number of sub-character constituents, JPEG picture complexity, character prototypicality based on a composite image of 2136 kanji characters, and individual characters? deviation from the composite image.
In a character complexity rating task with native and non-native speakers of Japanese, only native speakers were sensitive to sub-character constituents and character prototypicality. A stroke effect was found to be positive decelerating, perhaps following the Weber-Fechner law.
Analyses of response times in a progressive demasking task and first subgaze duration in a lexical decision task similarly revealed non-linear stroke effect and prototypicality effects.
The results overall indicate that the classical measure of character strokes is insufficient to capture the full visual complexity relevant in Japanese morphographic word recognition, in which words are dynamically perceived both as an ordered set of strokes, as constituents, and as an image, with reference to a kanji prototype in memory.