Â鶹ÒùÔº

When reading a passage, readers may pause at a particular word or return to reread it. Studies using eye-tracking, which record eye movements during reading, have suggested that word length, frequency, and predictability from context are the three primary factors influencing word processing during English reading. While these factors are known to affect non-native English readers as well, it has remained unclear which is the most decisive.

Researchers at the University of Tsukuba have analyzed an eye-tracking database of Japanese undergraduate and graduate students reading English texts. Researchers applied machine learning (i.e., random forest) to examine the relative importance of word length, frequency, and predictability during word processing.

The results, in Studies in Second Language Acquisition, showed that word length was the most critical factor in determining whether words were skipped and in total reading time. However, word frequency and predictability were more influential in initial processing (the time spent when first encountering a word) and in rereading. In addition, individual differences were sometimes more significant than lexical features.

The findings not only clarify the factors that make word processing difficult for English learners during reading comprehension, but also hold promise for applications across a wide range of fields, such as improving computational models of eye movements during reading.