A Corpus-Driven Approach to Stylistic Analysis of a Lexical Richness Curve