Arbitrariness at the agreement-classifier boundary
- Probal Dasgupta (Indian Statistical Institute)
- Thursday 29 September 2016
2311 BD Leiden
Classifiers may initially strike an agr[eement] language speaker as expressing Number plus obscure content. Once we put Gender in the picture, however, new questions emerge. Gender harbours arbitrariness; classification reflects tau [taxonomic] properties. Gender systems are austere; classifier systems are rich. Just how are we to handle languages like Bangla, Odia, Assamese that mix classifiers and agr? What sense can we make of the interaction of tau properties and phi-features more generally?
In the context of drafting tentative answers to these tough questions, we seem to have to appeal to semiotic principles. To introduce such principles gently, the presentation begins not with these obviously recalcitrant interactions between agr and classification phenomena, but with relatively straightforward, in-your-face facts about ‘titular’ classifiers. Titulars carry meanings like ‘big sister’ or ‘mister’ or ‘(your) excellency’; their properties await cross-linguistic exploration. It turns out that a set of titular facts in Bangla fall out neatly if we combine an unconstrained recursive syntax with sensitive principles of semiotics. These relatively easy-to-defend proposals give us the confidence to open up the much less tractable domain of agr-classifier interaction, where we seem to run into semiotic principles of a different order – at the level of characterizing feature ‘interpretability’ itself.