Document Type


Degree Name

Master of Arts (MA)




Faculty of Science

First Advisor

Keith Horton

Advisor Role

Thesis Supervisor


The study addresses a number of issues related to the effects of biasing semantic contexts on the processing of words with more than one meaning (homographs). Biasing contexts have been taken to either constrain “lexical access” to a contextually relevant meaning of a homograph (selective access), or to exert a selective effect only after access to all, or some subset of, the meanings of a homograph (multiple access). Recent findings based on the two-factor theory of attention (Posner & Snyder, 1975a) suggest that lexical access occurs in two stages, where the first stage involves automatic activation of all meanings and the second involves a rapid attentional selection of the contextually relevant meaning. A three word priming paradigm (Schvaneveldt, Meyer, & Becker, 1976) was employed to test the stages hypothesis. Subjects were required to name only the final target word, and their reaction time was the dependent variable. The critical trials involved presentation of the two word primes, where the first prime was a word related to one meaning of the second prime, which was a homograph. The comparison of most interest was between targets that were semantically congruent or incongruent with the biased homograph (e.g., oar-row-PADDLE and oar-row-COLUMN, respectively). These conditions were compared to two baselines: One employing two neutral primes (e.g., xxxxx-xxxxx-PADDLE), and one employing the biased homograph followed by an unrelated target (e.g., oar-row-GREEN). The stimulus onset asynchtrony (SOA) of the homograph was varied, as well as the strategies that subjects were instructed to use in attending to the context stimuli. Some evidence was found for the stages view of ambiguity resolution: At brief SOAs, congruent and incongruent targets were facilitated, whereas at a longer SOA, facilitation was significantly reduced for incongruent targets. Attentional strategies had less effect than anticipated. Also, results with the neutral baseline were discrepant with earlier findings. Discussion focused on the research hypotheses and characteristics of the naming task that might account for the discrepant findings. A brief theoretical overview concluded.

Convocation Year