Lexical decision for open- and closed-class words: failure to replicate differential frequency sensitivity.

Bradley and her colleagues (D. C. Bradley, Computarional distinctions of vocabulary type, unpublished Ph.D. thesis, Massachusetts Institute of Technology, 1978; D. C. Bradley, M. E. Garrett, & E. B. Zurif, in D. Caplan (Ed.), Biological studies ofmenra/ processes, Cambridge, MA: MIT Press, 1980) have reported a marked difference in frequency sensitivity between openand closedclass words on a lexical decision task. This effect was obtained with normal subjects, but not with Broca’s aphasics. Their results have already influenced experimental and theoretical investigations of syntactic processing. However, in three lexical decision experiments with normal subjects, modeled on those of Bradley et al., we failed to find such a theoretically interesting difference between the two classes. Instead, both classes showed similar reaction time frequency sensitivity for word frequencies less than approximately 316/million (H. Kucera & W. N. Francis, Computational analysis of presenf-day English, Providence, RI: Brown Univ. Press, 1967, count); above 399/million, the closed class had an almost-flat function of reaction time versus the logarithm of the frequency, while the open class may have had too few members for meaningful assessment. Because reaction time may be a nonlinear function of log frequency, and because there is relatively little overlap between the frequency ranges of the two classes, comparisons of the members of the two classes which might straddle the function’s inflection point must be made with extreme caution.