Nonlinguistic influences on lexical selection: Evidence from double entendres

Abstract

Contemporary models of lexical selection tend to be based upon spreading activation. Lexical units are supposedly organized such that semantic (and perhaps phonological and syntactic) associates are interconnected. During the encoding of a message lexical units are activated, and the activation reverberates with associate units. Those units which accumulate the greatest levels of activation have the greatest probability of being selected for the speech plan. The present study challenges two assumptions common to these models. One of these is the assumption that lexical activation may operate within only one semantic network at a time; the other is that lexical activation is instigated only by the linguistic parameters of the encoded message. The present experiments demonstrate that during the coding of neutral messages, sexually‐aroused respondents encode more sexual double entendres—i.e., words semantically appropriate to both the message and the arousal state—than do non‐aroused control respondents, and that the effect is more pronounced for high sex‐anxiety respondents than for others. These results are taken as evidence against the challenged assumptions, since lexical selection appears to have been operating simultaneously on two semantic networks, one of which is linguistically independent of the message.

Notes

Michael T. Motley is Associate Professor of Rhetoric, University of California, Davis, and Carl T. Camden is Assistant Professor of Communication, Cleveland State University. The authors wish to acknowledge the assistance of Becky Steveson, Linda Borla, and Georgia Saks. This study was partially funded by Faculty Research Grant #D‐1733 from the University of California, Davis.