Lexical frequency in morphology: is everything relative?

While it is widely assumed that high-frequency morphologically complex forms tend to display characteristics of noncompositionality, models of morphological processing do not predict a direct relationship between absolute frequency and decomposition. Rather, they predict a relationship between decomposition and the relative frequency of the derived form and the base. This paper argues that such a relative frequency effect does, indeed, exist. First, the results of a simple experiment demonstrate that subjects perceive derived forms that are more frequent than their bases to be significantly less complex than matched counterparts that are less frequent than their bases. And second, dictionary calculations reveal that derived forms that are more frequent than their bases are significantly more likely to display symptoms of semantic drift than derived forms containing higher-frequency bases. High-frequency forms, however, are no more prone to semantic drift than low-frequency forms. These results provide evidence that it is relative frequency, rather than absolute frequency, that affects the decomposability of morphologically complex words A low-frequency form is likely to be nontransparent if it is composed of even-lower-frequency parts. And a high-frequency form may be highly decomposable if the base word it contains is higher frequency still.