The effects of semantic relatedness on reaction time in a lexical decision task



Journal Title

Journal ISSN

Volume Title



This study tests two types of models that have been proposed to explain recent findings on the time needed to classify or compare strings of letters and words. Network models emphasize the processes that occur while retrieving stored information about letter strings. Set-theoretic models emphasize the processes that occur while comparing the semantic attributes of words. The two types of models make divergent predictions as to reaction time in comparison tasks in which the semantic relatedness between words is varied. Twelve subjects were instructed to respond "yes" if all three words in a vertical array were members of the same semantic category (animal or plant) , but to respond "no" if some were members of different semantic categories (animal and plant). For "no" responses, the reaction time was faster when words from the different categories were in the top or middle rather than bottom positions, suggesting that the words were processed serially from top to bottom. Both "yes" and "no" responses were faster when words in an array were semantically similar (from the same subcategory, e.g., all mammals). When one word within an array was from the same category (e.g., animal) as a semantically similar pair but was semantically different (in a different subcategory, e.g., bird instead of mammal), its placement between the two similar words did not slow responses, these results support a network model in which retrieval of words is facilitated by spread of excitation from semantically similar words already retrieved. The results do not support the set-theoretic models.