Why Stylometrics Is Not a Convergence Line
Computer-assisted authorship attribution was considered and excluded: the methodological reasons
Stylometrics is not a convergence line on this site. That decision requires more than a sentence; the method has genuine capabilities, and a page that simply dismissed it would be less than honest. What follows explains what stylometrics can establish, what it cannot, and why its results (including results that cut against the uncertainty this site documents) do not settle the question.
What stylometrics can establish
Computer-assisted authorship analysis has made genuine contributions to Early Modern literary scholarship. Its most powerful application is collaborative authorship: identifying where stylistic patterns shift within a text attributed to a single author, suggesting a second or third hand. The finding that Marlowe contributed to the Henry VI plays, that Fletcher co-wrote Henry VIII, that Middleton revised Macbeth; these are stylometric findings that have entered mainstream scholarship, and they are incorporated into this site’s collaborative authorship framework. In this application the method is not circular: it identifies a shift in stylistic signature and tests it against the signed works of known contemporaries.
The same collaborative-authorship methodology also supports one finding that this site accepts without qualification: the canon (minus the identified late collaborations) is overwhelmingly the work of a single mind. The stylistic consistency across the uncontested plays is not an artefact of a circular corpus argument; it is a positive finding about internal coherence. Fletcher and a small number of other late collaborators are identifiable precisely because they interrupt a signature that is otherwise remarkably stable across three decades and multiple genres. Whatever the authorship question ultimately resolves to, it resolves to a single primary author for the great bulk of the work. Stylometrics established that, and this site accepts it.
The circular corpus problem
The most common stylometric claim in the authorship debate takes a different form: measuring the plays against a “Shakespeare corpus” to confirm that they sound like Shakespeare. This is circular. The canon is defined by the traditional attribution; a method trained on that canon and used to confirm it is measuring consistency with itself.
The Barber finding is the sharpest version of this problem. Ros Barber’s peer-reviewed analysis (Digital Scholarship in the Humanities, Oxford Academic, 2021) showed that the Zeta test methodology (the basis for the most widely cited stylometric claims in the debate) cannot correctly attribute Hamlet to Shakespeare. A test that fails on the canon’s most certain case cannot carry weight on contested ones.
More broadly, stylometric methods lack a coherent probabilistic framework for assessing probative value: they produce a similarity score without the likelihood ratios needed to say how much that score should shift prior belief.
The cross-candidate comparison, and what it shows
There is a version of stylometrics that avoids the circular corpus problem: train models on candidates’ signed works and compare those models to the Shakespeare plays. This is methodologically legitimate, and it has been done. Craig and Kinney’s Shakespeare, Computers, and the Mystery of Authorship (Cambridge University Press, 2009) and MacDonald Jackson’s comparative analyses produce results more consistent with the traditional attribution than with the major alternative candidates. This is evidence worth acknowledging rather than avoiding.
The largest study of this kind is the Claremont Shakespeare Clinic, run by Ward Elliott and Robert Valenza from 1987 to 2010, and worth naming precisely because Elliott began the project sympathetic to the Oxfordian theory. Comparing the canon’s countable stylistic habits against the signed works of thirty-seven proposed claimants, the Clinic found the Shakespeare plays internally consistent (the profile of a single writing hand, not a committee) and found that no tested claimant’s surviving work matched it: Oxford, Bacon, and Marlowe were all excluded as the canon’s author on stylometric grounds. Cited by orthodox surveys as decisive, it is reported here at full strength, subject to the same two qualifications as every result on this page.
The honest qualification is the thin corpus problem. Most alternative candidates left too little signed writing in comparable genres for a reliable stylometric model. De Vere’s authenticated poetry is a small corpus of lyric verse, a different register from stage drama. Florio’s signed works are prose translations. Marlowe’s signed dramatic works are the most plausible comparison, but he died in 1593 with a fraction of the canon written. A negative stylometric result, when the comparison corpus is this thin, is largely uninformative: the absence of a match does not establish that the candidate did not write the plays. The method cannot distinguish “the candidate didn’t write it” from “the candidate’s corpus is too small for a reliable model.”
What cross-candidate stylometrics can say is this: the plays are more consistent with writing attributed to Shakespeare than with the signed works of the major alternatives, given current corpora. What it cannot say is: therefore the man from Stratford wrote them. Stylistic consistency with an attributed corpus is not biographical identity. The name on the corpus and the person who wrote the plays are precisely what the question is about.
Why it is not a convergence line
Two reasons together are sufficient. First, the thin corpus problem means that negative candidate comparisons are uninformative for most candidates on this site. A test that cannot separate “didn’t write it” from “corpus too small” cannot be scored as evidence. Second, even a positive finding (a stylistic match between a candidate’s signed work and the plays) cannot establish biographical identity. It establishes a stylistic profile match, which is consistent with authorship but not equivalent to it.
The convergence framework requires evidence that is both discriminating and primary-source-grounded. Stylometrics, in its current state for the authorship question as a whole, is neither. What it does contribute (the collaborative authorship findings) is incorporated. The rest is noted where relevant but not scored.
No investigative or scholarly sub-branch treats stylometrics as a necessary condition for establishing authorship. Forensic linguistics, corpus linguistics, and literary attribution studies across other fields all present it as one tool among several, and its practitioners consistently describe its outputs as probabilistic rather than definitive. In Early Modern drama conditions specifically (where collaboration, scribal transmission, and printing house variation confound the signal), its reliability is further reduced. The exclusion of stylometrics from this framework is not an eccentric position: it is consistent with how authorship attribution operates in comparable fields.