Abstract The present research examined the influence of polysemy on word naming. Naming was faster to polysemous than to nonpolysemous words. Moreover, polysemy interacted with word frequency such that the facilitative effects of polysemy were isolated to naming of low-frequency words. These findings are discussed with reference to the distributed models of Masson (1991, 1995; Borowsky & Masson, 1996; Masson & Borowsky, 1995) and Kawamoto, Ferrar, and Kello (1994). Resume La presente recherche examinait l'influence de la polysemie sur la designation des noms. Pour la designation, on choisissait plus rapidement des mots polysemiques plutot que des mots non polysemiques. En outre, la polysemie interagissait avec la frequence des mots de telle sorte que les effets facilitateurs de la polysemie etaient isoles a la designation de mots a basse frequence. Ces conclusions sont discutees en reference aux modeles repartis de Masson (1991, 1995; Borowsky & Masson, 1996; Masson & Borowsky, 1995) ainsi que Kawamoto, Ferrar et Kello (1994). The goal of reading is to extract meaning from printed text. Hence, it is reasonable to assume that semantics plays a role in the processing of single words. One semantic variable that has been used to investigate this assumption is polysemy. Polysemy (or ambiguity) refers to the number of meanings associated with a word. For example, the polysemous word chest can mean a body part or a piece of furniture, whereas the nonpolysemous word lake has a single meaning referring to a body of water. Kawamoto, Ferrar, and Kello (1994) and Masson and his colleagues (Borowsky & Masson, 1996; Masson, 1991, 1995; Masson & Borowsky, 1995) have implemented distributed models that provide explicit accounts of how polysemy should affect processing of single words. In both of these models, polysemy is predicted to facilitate lexical decisions, a prediction that is generally supported by the empirical evidence. For example, in recent papers, both Hino and Lupker (1996) and Borowsky and Masson (1996) reported that lexical decisions are faster to polysemous than to nonpolysemous words. For naming, however, the models diverge. The Kawamoto et al. model leads to the prediction that polysemy should facilitate naming, a prediction that has been supported with evidence from Hino and Lupker who found that polysemous words were named faster than nonpolysemous words. In contrast, Masson's model predicts that polysemy does not affect naming. This prediction has been supported by evidence from Borowsky and Masson showing that polysemy did not affect naming performance. Given the divergence of the models concerning polysemy effects in naming, and given the contradictory evidence of Hino and Lupker versus Borowsky and Masson, the goal of the present research was to provide a further empirical investigation of polysemy effects in naming. KAWAMOTO ET AL. (1994) MODEL The Kawamoto et al. (1994) model is built on the notion that for a polysemous word there is one orthographic pattern (or entry) but more than one semantic pattern (i.e., more than one meaning). The existence of more than one meaning is such that when a polysemous word is presented, activation at the semantic level may be unstable and inconsistent. To compensate for this inconsistency and uncertainty at the semantic level, the model's learning algorithm was tailored to give a relatively larger influence to weights at the orthographic layer than to those at the semantic layer for polysemous words. In contrast, nonpolysemous words activate only single patterns at both the orthographic and semantic layers and thus the learning algorithm generates weights that are similar at the orthographic and semantic layers. Kawamoto et al. assume that lexical decisions are based primarily on orthography. Accordingly, the model generates polysemy effects in lexical decisions because of the greater influence of the orthographic weights for polysemous words as compared to those for nonpolysemous words. …
Read full abstract