This course will begin with an introduction to the intriguing phenomenon of allocutive agreement, where a verb or functional element in the clause agrees not with the subject or some other argument, but with the addressee. For example, in certain dialects of Tamil, if you want to say ‘I’m leaving’ to a single person who you use informal pronouns with, it would be “Naan varreen”. But if you say it to a group of people or an individual who you would use formal pronouns with, it’s “Naan varreen-nga”. Patterns like this are found in a number of languages, but remain relatively understudied and have only come to be extensively discussed in the theoretical literature very recently. I will give an overview of the phenomenon based on data I have collected for dialects of Tamil, comparing these with what has been reported for Basque, Japanese and a handful of other languages.
I will then consider the implications that allocutive agreement has for our understanding of the grammatical representation of the speech act and for the theoretical treatment of agreement, once we’ve convinced ourselves that it really is agreement with the addressee. If we accept the idea that agreement relations are established in the morpho-syntax, then this means that there must be a representation of the addressee at a level of the grammar that the morpho-syntax has access to. This allows us to construct an argument for a version of the neo-performative hypothesis, according to which the speech act and its participants are represented by syntactic material in the left periphery of at least certain types of clause. This also means that we should find interesting interactions between allocutive agreement and other phenomena that have been argued to involve this clause-peripheral material, e.g. indexical shift, (embedded) root phenomena, imperatives and questions. I will show that, indeed, we do.
Once we’ve posited a representation of the addressee in the clausal left periphery and hypothesized that allocutive agreement involves agreement with this by some functional probe, we still have to ask ourselves how this compares to other types of agreement and how it fits in with our general theory of Agree. Where is the allocutive probe, and why does it target the addressee as opposed to the speaker or some argument? One could imagine a simplistic theory where the allocutive probe differs from a classic subject/argument-agreement probe in that it is located somewhere in the C domain rather than in T. It then picks out the addressee rather than the subject due to some sort of locality. Here the comparison with other types of agreement phenomena that seem to involve the clausal periphery becomes extremely important. Specifically, we already have fairly well-analyzed cases of agreement probes in the C domain going by the name of complementizer agreement. Here, however, the probe agrees not with the representation of a speech act participant, but with a local subject, either that of the clause in which the complementizer appears (e.g. in various West Germanic varieties) or that of the matrix clause (e.g. in various Bantu languages). It has also been convincingly argued that switch-reference marking, at least in some languages, involves agreement by a clause-peripheral probe with arguments of two distinct clauses (see Karlos and Emily’s course!). The course will conclude with a comparison of these various types of clause-peripheral agreement and explore ideas about how they might all be accommodated under a unified account, and what consequences this would have for our general theory of Agree.