To understand eponymous verb phrases such as "do a John Travolta," readers
cannot merely select a sense out of a mental lexicon (sense selection). The
y must create new senses (sense creation) by retrieving salient information
from memory. We conducted two experiments to test the hypothesis that thes
e processes of memory retrieval parallel those used for ordinary lexical am
biguities. To prepare for Experiment 1, we gathered readers' interpretation
s of eponymous verb phrases like "do a John Travolta" to establish dominant
and subordinate interpretations. We then wrote story contexts that biased
comprehension toward one or the other interpretation. In Experiment 1, para
phrase judgment times were used to demonstrate that dominant meanings are p
rivileged in the sense that they are accessible even when the story creates
a subordinate bias. In Experiment 2, this privilege faded somewhat when th
ere was a delay before the paraphrase judgment. We discuss the results with
respect to the distinction between sense selection and sense creation.