Abstract

The tools of linguistics and discourse analysis can produce an overwhelming amount of data on the language of a text. Given this richness, it is easy for analysts to select features that support a favored interpretation. How can stylistic analysis overcome selection bias, especially when examining the language of argument? Five methods for designing stylistic studies are worth considering. The first three involve making a principled selection from the text under consideration: If different versions or drafts of a text exist, the analyst can compare the changes the author considered critical. If particular phrases or passages are widely quoted or re-circulated, they can provide evidence of noticed stylistic choices. If similar texts exist, they can be examined for consistent features. The final two methods triangulate using constructs from rhetorical theory: The analyst selects and interprets language choices based on rhetorical features such as the line of argument used. The analyst starts from a device identified in rhetorical stylistics and sees how this device is used across examples. This last option will help to ground studies conducted under the other four.