Word Frequency Measures Research Articles

First in, first out: word learning age and spoken word frequency as predictors of word familiarity and word naming latency.

This study is concerned with recent claims that subjective measures of word frequency are more suitable than are standard word frequency counts as indices of actual frequency of word encounter. A multiple regression study is reported, which shows that the major predictor of familiarity ratings is word learning age. Objective measures of spoken and written word frequency made independent contributions to the variance. It is concluded that rated familiarity is not an appropriate substitute for objective frequency measures. A multiple regression study of word naming latency is reported, and shows that rated word learning age is a better predictor of word naming latency than are spoken word frequency, written word frequency, rated familiarity, and other variables. Possible theoretical explanations for age-of-acquisition effects are discussed and it is concluded that early-learned words have a more complete representation in a phonological output lexicon. This conclusion is related to relevant developmental literature.

Word Frequency and Age of Acquisition as Determiners of Picture-Naming Latency

In multiple-regression analysis of picture-naming latencies from an experiment modelled on Oldfield and Wingfield's (1965), with 94 stimuli and 37 adult subjects, two word frequency measures had insignificant beta weights, while two measures estimating age at which the word was learned had highly significant weights. Objects whose names were learned early were named faster. This result may have important implications for the interpretation of studies using word frequency as a critical variable. It is suggested that word retrieval may be a one-stage process that depends upon the age at which a word was learned.

The subjective estimation of relative word frequency

Experiments reported here demonstrate the relationship between measures of relative word frequency in terms of (a) frequency counts of textual material and (b) subjective scaling procedures. With two separate subjective scaling techniques—magnitude estimation (ME) and multiple rank orders (MRO)—it was found that, with respect to the relationship of the subjective scales to direct criterion measures, the ME approach yielded scales for which the power law provided the best fit. For the MRO scales, however, a logarithmic formulation was most appropriate.