Stylistics and questioned authorship

Forensic stylistics is the application of the science of linguistic stylistics to forensic contexts. The focus of forensic stylistics is author identification of questioned writings.

There are at least three types of questioned authorship problems. First, one may want to determine if one author wrote all the writings in a questioned set, i.e., if a particular writing, which may or may not be already accepted as part of a body (canon) of known writings, is consistent with the rest of the known writings. Second, one may be asked to compare a questioned writing with the writings of a large number of possible authors, if there are no obvious suspect authors. Third, the most common type of forensic problem is to assess the resemblance of a questioned writing to that of one author or a small number of candidate authors, if in fact possible suspect authors can be identified by external (nonlinguistic) means.

Linguistic stylistics uses two approaches to authorship identification: qualitative and quantitative. The work is qualitative when features of writing are identified and then described as being characteristic of an author. The work is quantitative when certain indicators are identified and then measured in some way, e.g., their relative frequency of occurrence in a given set of writings. Certain quantitative methods are referred to as stylometry. Qualitative and quantitative methods complement one another and are often used together to identify, describe, and measure the presence or absence of style-markers in questioned and known writings.

Typical cases of questioned authorship present a questioned writing to be compared or contrasted to the known reference writings of questioned writing one or more candidate authors. Such an analysis is accomplished by examining the writing style of all available questioned and known writings. The writing style is exhibited in underlying linguistic patterns internal to the habitual language used by the author. Results of this analysis may be 1) determination of resemblance of questioned writings to a common canon of known writings, 2) elimination or identification of one or more suspect authors, or 3) inconclusive with respect to data that support neither elimination nor identification.

Author identification is based on two well-documented facts: author-specific linguistic patterns are present in unique combination in the style of every writer, and these underlying patterns are usually established enough to be empirically described and measured by careful linguistic observation and analysis, making author identification possible.