Quantitative measures of lexical complexity in modern prose fiction

Abstract

Abstract The complexity of literary language proves stubbornly difficult to isolate, for reasons that we can label as both absolute and relative. On the one hand, complexity is commonly understood as an intrinsic property of literature, which distinguishes it from whatever we take to be ‘normal’ or ‘natural’ language. Literature is complexity, whether through vocabulary, metaphors, syntax, narrative structure, temporality, etc. It would be hard enough to capture any or all of these features, even were they to remain constant across time. But this is not the case. For both the quantitative nature and the qualitative effect of literary complexity change throughout history. While a reader may naturally find Homer’s Iliad or Shakespeare’s King Lear heavy going, it is really only in the 19th century—and then more powerfully still with the access of modernism—that a common perception emerges of literature, or at least a specific subset of literature, as characteristically ‘difficult’. The differing expectations that a modern reader brings to bear upon a text change the nature of that text. Faced with the dual character of complexity, even the massively more powerful computational tools that now exist can come to seem like blunt instruments. This article argues that we can usefully model literary complexity, notwithstanding these significant challenges. We do not believe that our own suggestions—which develop a new measure, and bring it into contact with existing approaches—tell the whole story; far from it. But we do believe that our methodology might usefully signpost future possible work in the field.