This paper draws from comparative rhetoric and contrastive stylistics and, by resorting to corpus linguistics methodologies, looks into possible quantitative descriptions of rhetorical deviations that are present in source texts within different language pairs and genres, in an attempt to map patterns of deviation, actual or expected, in translated texts, as a possible method to circumvent the epistemological limitations of one-to-one (language-to-language or text-to-text) comparisons. An experiment on the rhetorical use of metonymies and enthymemes within persuasive discourse in English and Spanish is discussed. Guidelines are proposed for the development of potentially comprehensive methodologies.