False memories for morphologically simple versus complex words in English

Abstract

In tasks such as lexical decision, people respond differently to morphologically complex words compared to morphologically simple ones (e.g. in English, lies vs. rise). These divergent responses could conceivably arise from differences in activation levels, or alternatively, from the additional steps required to decompose complex words. To investigate this issue, we used the Deese-Roediger-McDermott (DRM) false memory paradigm, which probes activation of lexical representations by measuring the probability of recalling or recognizing a word (such as lies) after listening to a list of its phonological neighbors (such as wise, lose, lime, etc.). Our results showed a significant false memory effect for complex words, which demonstrates that similar-sounding words can activate representations for stem-plus-affix combinations. Our results also showed no significant difference between false memory rates for complex versus simple words, which suggests that complex stem-plus-affix representations activate at levels equivalent to those of simple stem representations. These findings indicate that differences in activation level probably do not lie at the source of divergent responses to complex and simple words, and that decomposition is the more likely origin.