Invited talk

Tanmoy Bhattacharya

Title: Code-Switching without Grammar? Evidence from Language Mixing in Meiteilon-English and Bangla-Hindi

November 11, 2016, 2.40 pm

About the Speaker: Tanmoy Bhattacharya is an Associate Professor of Linguistics in the department of Linguistics , University of Delhi. His work mainly focuses on syntactic features of various Indian languages, Indian sign language, language and gender issues among others.

Abstract of the talk: There is a substantial Sociolinguistics literature arguing against the need for grammar in accounting for Code-Switching (CS); this is also implied in other statistics-based specialised approaches, including computational accounts. I will argue that most of these arguments are in fact short-sighted and incomplete. It will be argued that cases of language mixing cannot be passed off as a cases of the so-called “Third forms”, and that existing models of grammar are more than well-equipped to deal with such mixed forms. The unnecessary mystification surrounding these mixed forms in the name of innovation and creativity, will be dispelled by way of illustrating that the patterns found in the mixed forms all fall within the basic principles of universal grammar.

By this logic however, it seems that language mixing cannot really give us any new insight, since, apparently all mixing phenomena would fit neatly into existing formal models.

I will however, make the argument that this is not entirely true. The anti-grammar attitudes in their disproportionately adverse appraisal of applicability of grammar to language mixing miss out on the potentiality of language mixing data to help tease out a certain fundamental distinction between functional and content (or lexical) words – a distinction that a monolingual corpus has no way to furnish evidence in favour.

I will additionally demonstrate that such a distinction between functional and lexical items finds a natural home in non-lexical theories of grammar since such a distinction is the raison d’être for non-lexical family of models. Thus, I show that a potential positive contributory aspect of language mixing data towards grammatical theories can only be exploited if we take a turn towards non-lexicality in general.

Once this is established, it is an easy passage from there to the demonstrable inevitability of a non-lexical model like the Exoskeletal model to in fact provide the best fit for any language mixing data; I will show this for Meiteilon-English and Bangla-Hindi mixing cases.