The contribution of semantic transparency to the morphological decomposition of prefixed words

Linguists typically assume that a word is lexically decomposed into its constituent parts even when the root morpheme cannot be ascribed any clear meaning (e.g., sub-mit, de-cide, precise) (cf. Hockett 1954, Aronoff 1976). To date, the psychological evidence supporting the decomposition of these words is conflicting and controversial. In fact, some word recognition research supports the opposite view, that semantically non-compositional complex words do not pattern behaviorally with semantically compositional complex words (e.g., insincere or unfasten) (Marslen-Wilson et al. 1994). This paper presents data from two segment shifting studies (Feldman & Fowler 1987) which investigate the representations of bound root words. Experiment 1 contrasts segment shifting times for bound root words (e.g., receive), free stems words (e.g., reheat), and morphologically simple words with pseudoprefixes (e.g., religion). Response times after complex words with free stems were faster than after morphologically simple words, replicating the pattern of results previously found for suffixes using this task. This result suggests that speakers do encode a morphological boundary between prefixes and free stems. However, response times for complex words with bound roots were not statistically different from the morphologically simple words, suggesting that these words are not represented as morphologically complex. Experiment 2 continues the investigation of bound root words by focusing on the semantic contribution of prefixes to bound root words. I contrast bound root words with semantically transparent prefixes (e.g., recede) to bound root words with semantically opaque prefixes (e.g., receive). The results of Experiment 2 suggest that when the semantic transparency of the prefix is considered, bound root words are distinguished from morphologically simple words. Specifically, while response times after morphologically simple words and bound root words with semantically opaque prefixes were equivalent, the response times after bound root words with semantically transparent prefixes were significantly slower than after the simple condition. This result suggests that the semantic contribution of the prefix is relevant to whether a word is represented as morphologically complex or not. The results of these two studies suggest that speakers do decompose words that are derived from a semantically opaque bound root provided that the prefix is semantically transparent. Furthermore, these results suggest that the contrast between morphologically complex and morphologically simple is gradient, not dichotomous. The salience of the morphemic units and the boundaries between them depends on the evidence available to support their representations.