ABSTRACT
This chapter addresses the following questions concerning the processing of words with more than one meaning. How many meanings are computed and in what manner? At what point and how does prior context have a selective effect? What role does consciousness play? It is proposed that all meanings are accessed in a first preconscious stage, irrespective of context. However context determines what interpretation will be represented in consciousness.
The criterial index of whether a particular meaning had been accessed was facilitation of lexical decision to a subsequent word related to that meaning. Three letter-strings (LSs) were presented successively. Lexical decision was required to LS1 and LS3. On critical trials LS2 was a polysemous word (e.g., PALM) and LS3 was related to one of its meanings (WRIST). LS1 was either related to the same meaning (HAND), a different meaning (TREE), or unrelated to the following words (CLOCK). Control conditions included association of only LSI and LS3 and lack of any association. LS2 was (1) left unmasked to allow awareness; (2) pattern masked to prevent awareness but allow semantic access; or (3) energy masked to prevent both awareness and lexical processing.
When subjects were aware of LS2, prior context determined which meaning was chosen (indicated by facilitation of LS3 decision latency). But when LS2 was pattern-masked such that subjects were unaware of its presence, apparently both meanings were accessed, irrespective of context. It is argued that unconscious perception, including lexical/semantic access, is of unlimited capacity and precedes conscious perception, which is limited to one interpretation of an event at a time. Prior context is only used selectively after semantic access to determine what enters consciousness. This second stage of comprehension is 436constructive, and context is used inferentially as a consistency criterion. All semantic interpretations of a word accessed in the first stage automatically activate associated lexical entries. Once a conscious interpretation is arrived at, only entries associated with that interpretation receive activation; entries associated with incompatible interpretations receive inhibition.
