Function Words in Authorship Attribution. From Black Magic to Theory?

Abstract

This position paper focuses on the use of function words in computational authorship attribution. Although recently there have been multiple successful applications of authorship attribution, the field is not particularly good at the explication of methods and theoretical issues, which might eventually compromise the acceptance of new research results in the traditional humanities community. I wish to partially help remedy this lack of explication and theory, by contributing a theoretical discussion on the use of function words in stylometry. I will concisely survey the attractiveness of function words in stylometry and relate them to the use of character n-grams. At the end of this paper, I will propose to replace the term 'function word' by the term 'functor' in stylometry, due to multiple theoretical considerations.

Not All Character N-grams Are Created Equal: A Study in Authorship Attribution

“…Kestemont (2014) conjectured that their utility was in capturing function words and morphology. Koppel et al (2009) suggested that they were capturing topic information in single domain settings, and style and syntactic information in cross-domain settings.…”

“…Various hypotheses have been put forth to explain the "black magic" (Kestemont, 2014) behind the success of character n-gram features in authorship attribution. Kestemont (2014) conjectured that their utility was in capturing function words and morphology.…”

CIC-FBK Approach to Native Language Identification

“…They are considered one of the most important stylometric features (Kestemont, 2014). Function words can be seen as indicators of the grammatical relations between other words.…”

Authenticating the writings of Julius Caesar

“…Much research nowadays therefore concerns ways to effectively extract stylistic characteristics from documents that are not affected by a text's specific content or genre (Argamon & Levitan, 2005;Kestemont et al, 2012;Efstathios, 2013;Sapkota et al, 2015;Seroussi et al, 2014;Sapkota et al, 2014). This has not always been the case: historical practitioners in earlier centuries, commonly based attributions on a much looser defined set of linguistic criteria, including, for instance, 45 the use of conspicuous, rare words (Love, 2002;Kestemont, 2014). Naturally, an expert reader's subjective intuitions (Gelehrtenintuition, connoisseurship) would play a much larger role in studies than would nowadays be acceptable.…”

“…This modelling strategy has the advantage that it can also capture morphemic information at the subword level, and is thus potentially sensitive to functional morphemes that are not realised as individual words (e.g. word endings) (Kestemont, 2014;Sapkota et al, 2015;Stamatatos, 80 2009b). Similarly, certain n-grams also pick up word stems and research increasingly demonstrates that text representations based on function words can be supplemented with information from lowerfrequency strata in languages (Burrows, 2007), such as word stems (Koppel et al, 2009).…”