ABSTRACT
The frequency with which words are used has long been known to influence performance in reading tasks. Such effects are robust, often large, and sufficiently universal that no serious account of the processes underlying reading can fail to incorporate an explanation of frequency effects. However, the task of formulating such an account has been made difficult by controversy about where frequency effects arise in the chain(s) of mental processes connecting printed word to measured performance in reading tasks used in the laboratory. The debate about localization of frequency effects will be one theme of this chapter. Another will be a motive for interest in frequency effects that has tended to be neglected: their relation to learning. Frequencies of occurrence, as listed in the various text-based counts (e.g., Kucera & Francis, 1967), provide a rough estimate of the differential amount of experience an average adult reader of English will have had with particular orthographic patterns (and their interpretation). Experience is the stuff of which all learning is made. My argument will be that if our models of word recognition do not deal with how we learn to recognize new words, they are unlikely to deal adequately with effects of frequency upon the recognition of old words.
