Running head : FEATURE FREQUENCY EFFECT Feature Frequency Effects in Recognition Memory Kenneth

Rare words are usually better recognized than common words, a finding in recognition memory known as the word-frequency effect. Some theories predict the word-frequency effect because they assume that rare words consist of more distinctive features than common words (e.g., REM, Shiffrin & Steyvers, 1997). In this study, recognition memory was tested for words that vary in the commonness of their orthographic features, and we found that recognition was best for words made up of primarily rare letters. In addition, a mirror effect was observed: words with rare letters had a higher hit rate and a lower false-alarm rate than words with common letters. We also found that normative word frequency affects recognition independently of letter frequency. Therefore, the distinctiveness of a word's orthographic features is one but not the only factor necessary to explain the word-frequency effect. FEATURE FREQUENCY EFFECT 3 Rare words are better recognized than common words (Schulman, 1967; Shepard, 1967; but see Wixted, 1992), a finding in recognition memory known as the wordfrequency effect (WFE). For single-word old-new recognition, hit rates (HRs, correctly responding “old” to an old word) are higher and false-alarm rates (FARs, incorrectly responding “old” to an new word) are lower for low-frequency words than for highfrequency words (Glanzer & Adams 1985). Examples of some of the accounts for the advantage for low-frequency (LF) words include: differences in the distribution of attentional resources (e.g., Glanzer & Adams, 1990; Maddox & Estes, 1997), multiple retrieval processes (e.g., Joordens & Hockley, 2000), the number of different contexts in which words appear (e.g., Dennis & Humphreys, 2001), and differences in encoding variability (e.g., McClelland & Chappell, 1998). No consensus yet exists regarding how many -if any -of these accounts is correct, but it is quite clear that normative word frequency is correlated with a large number of variables that could theoretically produce the WFE (cf. Gillund & Shiffrin, 1984; Shiffrin & Steyvers, 1997). Here, we directly test an hypothesis other than those listed above, that differences in the distinctiveness of the features that comprise words of differing frequency produce the WFE (e.g., Shiffrin &