Compound Frequency Effect in Word Production: Evidence from Anomia
Yanchao Bi,Zaizhu Han,Hua Shu
DOI: https://doi.org/10.1016/j.bandl.2007.07.042
IF: 2.781
2007-01-01
Brain and Language
Abstract:The issue of what kind of frequency affects compound word (e.g., ‘‘butterfly’’) production has recently received special attention in the context of distinguishing two major types of speech production models. The frequency effect is widely assumed to originate from the lexical system (Jescheniak & Levelt, 1994), and one model (IN, Caramazza, 1997) predicts that the frequency of the compound words themselves (‘‘butterfly’’) determines naming performance and another model (WEAVER++, Levelt, Roelofs, & Meyer, 1999) predicts the frequencies of the constituents (/butter/ and /fly/) are the effective frequency. Such predictions are deduced from the following combinations of assumptions of the two models. WEAVER++ assumes that compounds are decomposed into its constituents at a lexeme level where phonological (or orthographic) forms are represented, and this is also the level at which frequency effect locates. IN assumes that a compound is represented by one node (‘‘butterfly’’) in its only lexical layer, which is also the locus of the frequency effect. Chronometric studies have yielded inconsistent results (e.g., Bien, Levelt, & Baayen, 2005; Janssen, Bi, & Caramazza, submitted for publication). Here we present convergent evidence from the perspective of aphasia. It has been assumed that the same mechanism underlie the frequency effect in predicting the naming latencies of unimpaired speakers and the frequency effect in affecting the accuracy of anomic patients (see Miozzo, Jacobs, & Singer, 2004; Nickels, 2001). We report two Chinese patients with whom we studied oral compound production and written compound production respectively. The question is whether compound frequency or the frequencies of the constituents best predict the naming accuracy in the two modalities of output.