English written texts were produced by a group of monolingual speakers, as well as Chinese-English and Spanish-English bilinguals. These were randomly presented to another set of participants from the same three language groups for rating. The raters were unable to identify the language background of the authors of the transcripts, yet they were found to prefer the way the arguments were presented in the transcripts of their own language group. In contrast, there was no preference for the content of the arguments of the three language groups. A discourse analysis identified several aspects of the texts that might have led to the own-language preferences for rhetorical structure. The study provides empirical support for the notion of contrastive rhetoric.