Statistical Learning in Non-Chinese Speakers Exposed to a Sequence of Words without Spaces

Abstract

A Chinese sentence contains no extra spaces between words. We hypothesize that implicit word segmentation by Chinese readers is developed via the statistical learning mechanism that is universal. Twenty non-Chinese speakers were exposed to a sequence of 3600 characters constructed from six disyllabic words with 6 different characters. The transitional probabilities between any two characters were .46 to 1 within words, and 0 to .29 between words. The sequence was presented one character every half a second. Occasionally, the presentation rate doubled. The participants’ task was to detect the instances of double presentation. Upon completing the task, a surprise test followed that consisted of a word and a nonword (reversals of the characters in a word) from the sequence. The participants had to decide which one had appeared before. The averaged accuracy rate was greater than chance (.53). This suggests the statistical learning mechanism is available to all language learners.


Back to Table of Contents