Proceedings of the Annual Meeting of the Cognitive Science Society

We investigated whether Chinese readers learn to segment words automatically while reading unspaced sentences throughstatistical learning. Experiment 1 replicated Saffran et al.s (1997) study using Chinese monosyllables presented auditorilyto foreign learners of Chinese. The learning outcome was .57 on a two-alternative forced-choice test, statistically betterthan guessing (.5). Experiment 2 repeated Experiment 1 but presented the Chinese monosyllable string visually as acharacter string. Experiment 3 repeated Experiment 2 but doubled the exposure. Experiment 4 repeated Experiment 2with characters of fewer numbers of strokes. The learning outcomes were .53, .52, and .52., not significant when testedindividually, but was significant when the data were combined. At least 60% of the participants in each experiment showedthe effect. We conclude that visual statistical learning does contribute to automatic word segmentation in Chinese reading.