Gripla - 2020, Qupperneq 25

GRIPLA24
latter: both methods are premised on the notion that writers, consciously
or unconsciously, employ patterns in their use of language. For a given doc-
ument, then, it should be possible to identify these patterns and compare
them with other documents to determine how tendencies appear across
the corpus. While it is certainly possible to do this sort of investigation
manually, as Erichsen and Magerøy did, the stylometric method allows us
to perform this sort of analysis in a statistically-robust, computer-assisted
manner.53 Stylometric methodologies are supported by a growing body of
research conducted by scholars operating in an interdisciplinary manner
at the intersection of language, literature, statistics, machine learning, and
corpus linguistics. When done correctly, it also allows us to reduce bias,
since the selection of features is generally not conducted by the human
investigator (though the human investigator can still very well introduce
bias into the research design, as discussed below).
The advent of stylometry has not changed the fact that the original
premise of stylistics (that writers have particular habits) is not without its
complications. While it is mostly uncontroversial to talk about the exist-
ence of style, scholars undertaking the analysis of style must constantly in-
terrogate whether identified “patterns of language use” should be explained
by style, or by something else entirely. There are, in fact, a large range of
possible explanations for a given pattern. It may be that a particular pat-
tern emerges because of circumstances arising due to a text’s manuscript
transmission, thematic content, genre, setting, narration, editorial practice,
or something else. Stylometry has provided us with powerful methods for
identifying patterns in our texts and measuring the similarity between
documents based on the frequencies of these patterns. But we must be
careful not to get carried away by these advancements and neglect to inter-
53 At least four published studies in the Old Norse field have applied stylometry to ill-
uminate old debates surrounding mainly questions of authorship. These studies are:
Rosetta M. Berger and Michael D.C. Drout, “The relationship between Víga-Glúms
saga and Reykdæla saga: Evidence from new lexomic methods,” viking and Medieval
scandinavia 11 (2015): 1–32; Jón Karl Helgason et al., “Fingraför fornsagnahöfunda:
Fráleiðsla í anda Holmes og stílmæling í anda Burrows,” skírnir 191 (2017): 273–309;
Haukur Þorgeirsson, “How Similar are Heimskringla and Egils saga? An Application of
Burrows’ delta to Icelandic Texts,” European journal of scandinavian studies 48.1 (2018):
1–18; and Michael MacPherson, “Samdi Bjarni biskup Málsháttakvæði? Glímt við drótt-
kvæði með stílmælingu,” són 16 (2018): 35–58.