Narrator language and character language in Thucydides: A quantitative study of narrative perspective

Abstract

We investigate the distribution of content words in Thucydides to shed light on his famous ability to construct a perspectivally rich narrative. Following de Jong's narratological study of Homer we define as character language words that occur predominantly in character text (direct speech reports). When such words do - infrequently - appear outside character text, they are often to be interpreted as conveying a character's perspective. We apply the same reasoning to Thucydides, but proceed in a highly automated way: we look at the relative frequencies of all words in character and narrator text and we identify character language as those words whose distribution is the most skewed. Our lemmatizer Glem makes it possible to do this at the level of the lemma. It turns out that there indeed is such a class of character language in Thucydides, many words of which are evaluative. Manually investigating the rare occurrences of character language in narrator text reveals that Thucydides uses character language not only as a perspectivization device, but also for authorial comments. In a second stage we identify chapters in narrator text where character language expressions cluster and show them to be passages with interesting perspectival shifts or special dramatic intensity. Interestingly, the passage ranking highest on character language density is chapter 84 of book 3, the authenticity of which has been long doubted. Our study provides a novel approach to narratological questions, and also contributes to a linguistic debate on the nature of perspective dependence.