In two experiments, we tested the hypothesis that some portion of the liter
al meaning of ironic remarks is processed automatically, along with the int
ended meaning. In experiment 1, subjects took longer to judge the evaluativ
e tone (positive or negative) of utterances used ironically than used liter
ally, demonstrating that the literal meaning of the ironic utterances was a
ccessed, and interfered with the nonliteral meaning judgment. In experiment
2, subjects took longer to judge the evaluative tone of the literal meanin
g of utterances used ironically than used literally, demonstrating that the
nonliteral meaning of the ironic utterances was accessed, and interfered w
ith the literal meaning judgment. These results support a multiple meaning
model of irony processing in which both literal and nonliteral meanings are
obligatorily processed. Results are also consistent with social-functional
studies of irony demonstrating that the evaluative tone of the literal mea
ning of irony serves to mute the perceived tone of the intended, nonliteral
meaning (Dews and Winner, 1995). (C) 1999 Elsevier Science B.V. All rights
reserved.