Exploitation of stress information in spoken-word recognition in Dutch

Abstract

A fragment-priming experiment addressed the question of how stress information is exploited in the process of recognizing spoken Dutch words. Materials were 20 pairs of Dutch words such as octopus/oktober, with two initial syllables which were segmentally identical but different in stress; both begin okto- (N.B.: in neither case is their vowel reduction, in Dutch), but octopus is stressed on the first syllable, oktober on the second. Exploiting the known effect that visual lexical decisions are made more rapidly if subjects have just heard the word, or part of it, we compared the lexical decision response times of 56 native speakers of Dutch to the target words (e.g., OKTOBER), presented visually immediately following a spoken presentation of a neutral sentence ending with a partial word. This partial word had (a) a correctly stressed beginning (e.g., okTO-), (b) an incorrectly stressed beginning (e.g., OKto-), or (c) a control beginning (e.g., eufo-). Responses were significantly faster (than the control condition) only after correctly stressed primes; that is, OKto- primed OCtopus but not okTOber, okTO- primed okTOber but not OCtopus. Thus Dutch listeners can exploit stress information effectively at a relatively early stage in the spoken presentation of a word.