Stylistic constancy and change across literary corpora: Using measures of lexical richness to date works

The measure of the lexical richness of literary texts as a tool in the comparative analysis of literary style has been hampered by the problem of the inequality of text lengths within and between literary corpora. This paper proposes an empirical method of description of lexical richness by averaging measures on multiple chunks of text of a standard length within a literary work or corpus. A work's average vocabulary richness, average portion of hapax legomena of the corpus from which it derives, and average repetition of frequently appearing vocabulary may then characterize that work relative to other works partitioned along with it. This method reveals the possibility of significant variance of these measures of vocabulary among works of a single author's corpus and warns against the notion of some absolute authorial stylistic character. We apply this method of vocabulary averaging to the corpora of three playwrights from classical antiquity whose works are chronologically rankable: Euripides, Aristophanes, and Terence. We look for trends in vocabulary richness over time, which we posit functions as an indicator of progressively changing authorial ability or inclination. This method then holds the potential of predicting dates for undateable or tenuously dated works within a corpus of otherwise securely dated texts. From the results derived, a relatively late date for the composition of the redrafted version of Aristophanes' Clouds appears likely; we predict an early composition date for the redraft of Terence's Hecyra (and thus are inclined to think that the playwright did very little redrafting); and finally we find Euripides' Electra and Supplices exhibiting vocabulary characteristics of extremely late composition and we predict dates much later than those assigned based on metrical considerations.