High variability orthographic training: Learning words in a logographic script through training with multiple typefaces

We tested whether naturally occurring visual variability—specifically, typefaces—would help people generalize word learning to typefaces they had never seen before. In Chinese, thousands of unique written characters must be learned item by item, and differentiated from similar-looking characters. Participants (n = 190) with no previous Chinese experience learned 24 Chinese characters in one of two training groups: the Single-Typeface group trained using only one of three Chinese typefaces; the Variable-Typeface group trained using all three. Everyone completed two training and testing phases. During Definition Training, they saw each character six times and learned to associate it with an English definition (水–water). After training, participants were tested on their accuracy in providing definitions for the characters. During Form Training, participants chose the characters they had previously learned from a display that included a trained character and a visually similar distractor (水 vs 永). After training, they were tested on their speed/accuracy in choosing the learned characters. At testing in both phases, half of the words were presented in a familiar typeface; half in a novel typeface. Results showed significant interactions between training and testing conditions in both phases, with a significant effect of training in the Form Testing phase: Single-Typeface training resulted in faster responses for familiar typefaces, but much slower responses for novel typefaces; in comparison, Variable-Typeface training resulted in better generalization to novel typefaces. These results suggest that typeface variability can influence how effectively people generalize knowledge during the initial stages of learning a logographic script.