Stylometry for dummies: understanding how Stylo package works on Romanian texts
Stylometry for dummies: understanding how Stylo package works on Romanian texts

Summary/Abstract: In the absence of clear strategies, such as “thematic foregrounding” (a means of highlighting the thematic aspects and of backdropping the stylistic aspects of a text), Stylo Package should be used with care, as a diagnosis means. It is, thus, useful to evaluate, through statistical lens, traditional assumptions on periods, subgenres, authors’ co-influencing and so forth, but these results should be credited only as prelims. Beside studies that could replicate on Romanian novel corpora previously proposed workflows, we believe that evaluation of the tool’s performance as far as a precise language is concerned (in our case Romanian) should be done by means of multilingual stylometry, which has yielded very nice results for “lesser-resourced” languages too. We entitled the present contribution, “stylometry for dummies”, in a self-ironical mood, but also for several other reasons, such as to indicate the experimental level of our approach to both the tool’s and the testing material’s features, as well as to point at the perplexity that the novice researcher experiences in front of large, easy-to-get and easy-to-discard sets of “results” produced by the Stylo Package. After spending some time with combining features, changing parameters, optimizing processing and other tasks, we believe that beginners should keep a diary in which self-observation while using the tool and progress with understanding the tool’s computational premises should mix.

• Details
• Contents