What makes them different: the extraction of distinctive linguistic patterns for the protagonists of Molière's plays

Quantitative approaches to the study of style in literature are far from a modern novelty. They have however recently gained more and more popularity, not only among computer scientists and corpus linguistics, but also among some influential literary critics. The present panorama of quantitative techniques is very rich, but often confusing, with a plethora of denominations and methodologies often difficult to reconcile; computer scientists classify their work as stylometry or computational stylistics, while linguists may use the label corpus stylistics, and finally critics like Franco Moretti will talk about macro-analysis and distant reading. This talk will try first to identify the differences between these trends, distinguishing between corpus based and corpus driven approaches on the methodological side (Quiniou et al 2012), and (following Ramsey 2011) between experimental and hermeneutical approaches. Finally we will present ongoing work conducted at Labex OBVIL on syntactic pattern extraction from theatrical characters. The proposed approach, using correspondence analysis to extract distinctive traits for each character, is imagined rather as an hermeneutical tool, in the sense that it does not seek to demonstrate that two different characters have been endowed with significantly different stylistic traits by the playwright, but it does enable the visualisation of their relative distances and the extraction of those elements that make them distinct.