The Time Course of Activation of Semantic Information during Spoken Word Recognition

Abstract

We investigated the time course of activation of the mental representatio ns of word meanings in a series of three cross-modal priming experiments. In Experiment 1, we showed a signi cant priming effect for semantically related targets presented at the "isolation point" of the prime word, con rming earlier evidence for the activation of multiple word meanings before the point at which a word can be recognised. The use of non-associated prime-target materials ruled out the possibility that this could be an artifact of form-based associative priming. Experiment 2 demonstrated that certain semantic properties are made available more rapidly than others in the duration of a spoken word and, speci cally, that for words referring to man-made objects, information about their function and design is activated more quickly than information about their physical form. However, this experiment did not reveal any limit on the activation of a word's meaning as a function of the number of simultaneously activated competitor candidates. Finally, Experiment 3 showed that the results of the rst two experiments could not be explained in terms of backwards priming from target to prime. We interpret the data with respect to both localist and distributed implementations of the cohort model.

Looking at anything that is green when hearing “frog”: How object surface colour and stored object colour knowledge influence language-mediated overt attention

“…But when two types of competitor are copresent one type of information may be weighed more strongly than the other. Categorical/functional knowledge, for instance, is a particularly salient aspect of lexical knowledge (e.g., Moss et al, 1997); it is not difficult to imagine a model in which the strength of the activation of particular features in a given situation translates into the probability of attending towards whatever shares those features.…”

“…Pecher, Zeelenberg, and Raaijmakers (1998), however, found perceptual priming only if participants were first given practice in categorizing primes and targets in a perceptual categorization task. Moss, McCormick, and Tyler (1997), using the lexical decision task, found a significant priming effect for functional properties of words early during the duration of the word but priming for perceptual targets (e.g., the colour overlap between aspirin and white) only at the offset of the prime word. Kellenbach, Wijers, and Mulder (2000) obtained robust perceptual priming as indexed by the event-related potential (ERP) N400 component but contrary to Moss et al (1997) observed no effect for the same materials from the ERP study when used in a lexical decision task.…”

“…Moss, McCormick, and Tyler (1997), using the lexical decision task, found a significant priming effect for functional properties of words early during the duration of the word but priming for perceptual targets (e.g., the colour overlap between aspirin and white) only at the offset of the prime word. Kellenbach, Wijers, and Mulder (2000) obtained robust perceptual priming as indexed by the event-related potential (ERP) N400 component but contrary to Moss et al (1997) observed no effect for the same materials from the ERP study when used in a lexical decision task. Similarly, we (Huettig & Altmann, 2004 observed that our participants started to shift overt attention to a cable well before the offset of the spoken word "snake", suggesting that shape information can be accessed before word offset (in contrast to Moss et al, 1997, who did not find priming of perceptually related word pairs before the offset of the prime word).…”

Modeling the relation between the production and recognition of spoken word forms

“…While Dell and Reich (1981) found a strong lexical bias in their corpus of errors in spontaneous speech, Garrett (1976) found no such effect and Stemberger (1985) found only a weak effect. In an analysis of the errors in picture naming of fifteen aphasic speakers, Nickels and Howard (1995) found no evidence for lexical bias. A feedback account of the lexical error bias is provided by the DSMSG model.…”

“…Evidence for the multiple activation of lexical candidates during word recognition comes from cross-modal semantic priming experiments (e.g., Moss, McCormick, and Tyler 1997;Zwitserlood 1989). For example, Zwitserlood (1989) asked participants to listen to spoken words (e.g., CAPTAIN) or fragments of these words (e.g., CAPT).…”