Changes in the length of speeches in the plays of William Shakespeare and his contemporaries: A mixed models approach

Abstract

Since 2007 a number of investigators have compiled statistics on the length in words of speeches in plays by William Shakespeare and his contemporaries, focusing on a change to shorter speeches around 1600. In this article we take account of several potentially confounding factors in the variation of speech lengths in these works and present a model of this variation in the period 1538-1642 through Linear Mixed Models. We confirm that the mode of speech lengths in English plays changed from nine words to four words around 1600, and that Shakespeare's plays fit this wider pattern closely. We establish for the first time: that this change is independent of authorship, dramatic genre, theatrical company, and the proportion of verse in a play's dialogue; that the chosen time span can be segmented into pre-1597 plays (with high modes), 1597-1602 plays (with mixed high and low modes), and post-1602 plays (with low modes); that some additional secondary modes are evident in speech lengths, at 16 and 24 words, suggesting that the length of a standard blank verse line (around 8 words) is an underlying unit in speech length; and that the general change to short speeches also holds true when the data is viewed through the perspective of the median and the mean. The change in speech lengths is part of a collective drift in the plays towards liveliness and verisimilitude and is evidence of a hitherto hidden constraint on the playwrights: whether or not they were aware of the fact, playwrights as a group were conforming to a structure for the distribution of speech lengths peculiar to the era they were writing in. The authors hope that the full modelling of this variation in the article will help bring this change to the attention of scholars of Shakespeare and his contemporaries.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 2 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.