Frequency effects in reading are powerful – But is contextual diversity the more important variable?

For decades word frequency has been one of the most important variables in psycholinguistics. Frequent words are more easily recognized and processed more efficiently than rare words. In the fields of word recognition and psycholinguistics, all researchers are reminded to statistically control for word frequency. But is that advice still correct? Are other variables which are correlated with word frequency more important for human language processing? These questions have arisen because of the recent construction of text corpora of billions of words. Also important is the growing practice of archiving word recognition data in databases accessible for anyone to mine. A key result is that words that typically appear in restricted contexts are processed less efficiently than words appearing in diverse contexts. But the new variable of contextual diversity hasn't simply replaced word frequency. This paper traces the history of contextual diversity findings, including the twists and turns towards a more sophisticated understanding of what makes words easy to learn and process. Myriad findings of the last 20 years are discussed: the rational theory of memory, spacing effects in learning, phrase frequency effects, the neural basis of repetition suppression, and why reverse frequency effects are observed in semantic aphasia. Methods reviewed include artificial language learning, event-related potentials, and eye movement studies. The result is a new appreciation that word processing skills emerge from complex brain networks which include information about words' typical contexts of occurrence.