UConn Ling Lunch: Peter Baumann (Northwestern) (Feb. 4, 2014)

Peter Baumann (Northwestern University) will give a ling lunch talk titled Overspecification in referential language games, starting at 12:30 on Feb. 4, 2014.

In language comprehension, pragmatic inference can be effortful: in reading, sentences involving implicatures take longer to process than sentences without implicatures (e.g. Hamblin & Gibbs, 2003), and in pragmatic language games, target identification is less accurate when the message involves an implicature than when it does not (e.g. Degen & Franke, 2012).

In language production, on the other hand, speakers have the possibility to reduce their effort of pragmatic inference by over-specification, i.e. saying more than is strictly necessary or optimal. And indeed, there is plenty of empirical evidence that speakers do make use of over-specification (e.g. Koolen, Gatt, Goudbeek, & Krahmer, 2011)

However, in empirical research on pragmatic reasoning about referring expressions, in particular in referential language games (e.g. Degen & Franke, 2012; Frank & Goodman, 2012), the speakers’ possibility of over-specification is often ignored and instead the theoretical assumption that overly specific (referring) expressions incur a cost for being longer than an ‘optimal’ expression is built into the experimental tasks by artificially restricting the set of possible referring expressions available to identify a referent.

In this talk, I will present a simple referential language game, in which there was no restriction for the speakers, and I will show that speakers make use of overspecification to avoid the effort of pragmatic inference. Based on this (preliminary) evidence, I propose that in addition to a cost for longer expressions, the speaker also incurs a cost for pragmatic inference, and that a speaker will choose over-specification if the cost for pragmatic inference outweighs the cost for articulatory effort (or longer expressions).