Reply to: Large-scale quantitative profiling of the Old English verse tradition

TL;DR

This paper critically examines a previous large-scale quantitative study of Old English poetry, highlighting methodological flaws and demonstrating that stylistic differences in Beowulf are significant and detectable.

Contribution

It challenges prior claims by identifying errors in data and code, and shows that reliable stylometric analysis reveals heterogeneity in Beowulf's style.

Findings

Stylometric methods detect significant stylistic heterogeneity in Beowulf.

Previous data and code errors undermine earlier conclusions.

Reliable analysis supports the presence of multiple stylistic features in Old English poetry.

Abstract

In Nature Human Behaviour 3/2019, an article was published entitled "Large-scale quantitative profiling of the Old English verse tradition" dealing with (besides other things) the question of the authorship of the Old English poem Beowulf. The authors provide various textual measurements that they claim present "serious obstacles to those who would advocate for composite authorship or scribal recomposition" (p. 565). In what follows we raise doubts about their methods and address serious errors in both their data and their code. We show that reliable stylometric methods actually identify significant stylistic heterogeneity in Beowulf. In what follows we discuss each method separately following the order of the original article.