Contributing to a New Database on Valency Classes with Latin Data. Challenges and Perspectives

This paper describes the first steps toward adding Latin to PaVeDa, a typological database designed after the ValPaL database supporting cross-linguistic and diachronic analysis of valency patterns, and shows the perspectives of this endeavour through a case study. Concerning data selection for PaVeDa, we point out methodological challenges related to Latin being a “corpus-language” by discussing new criteria (e.g., frequency, morphological complexity, continuity of attestation) and relevant issues. We then show that the PaVeDa methodology allows for studying valency changes over time by discussing the diachrony of the valency patterns of the experiential verb doleo ‘feel pain’. Our analysis of basic and alternat- ing valency patterns shed light on the argument structure variations of doleo ‘feel pain’ in different stages of Latin. Most remarkably, pattern alternation in Classical Latin is meaningful. Early Latin’s basic valency pattern with dative experiencers is retained only to express bodily pain, whereas a transitive pattern with nominative experiencers and accusative stimuli specializes to encode mental pain.