The present study examined the viability of a model of biased language use (e.g., Maass et al., 1989; Maass et al., 1996) as a measure of anti-lesbian prejudice among heterosexual women. As a matter of methodological and psychometric exploration, four alternative methods of scoring linguistic abstraction were employed.One hundred eighty-one female university undergraduates provided abstraction ratings for a series of seven fictitious news clippings, adapted from von Hippel et al. (1997), four of which depicted either heterosexual or lesbian women engaging in behaviors that were either socially favorable or unfavorable and lesbian stereotypic or counterstereotypic. Participants also completed a "Memory Test" devised for this study, a demographic questionnaire, and the ATL Subscale of the ATLG (Herek, 1988, 1994).The prediction that the linguistic expectancy bias (LEB; e.g., Maass, 1999) would be observed in the present intergroup context was not supported. The overall pattern of results suggested some stability across scoring methods. A number of implications of the findings are considered, with a focus on methodological concerns and applications to practice. / Department of Counseling Psychology and Guidance Services