Word Embedding Distance Does not Predict Word Reading Time

Abstract

It has been claimed that larger semantic distance between the words of a sentence, as quantified by a distributional semantics model, increases both N400 size and word-reading time. The current study shows that the reading-time effect disappears when word surprisal is factored out, suggesting that the earlier findings were caused by a confound between semantic distance and surprisal. This absence of a behavioural effect of semantic distance (in the presence of a strong neurophysiological effect) may be due to methodological differences between eye-tracking and EEG experiments, but it can also be interpreted as evidence that eye movements are optimized for reading efficiency.


Back to Table of Contents