Pragmatic reasoning through semantic inference

Abstract

A number of recent proposals have used

Negotiating Lexical Uncertainty and Speaker Expertise with Disjunction

“…The model finds its conceptual origins in Lewis's (1969) work on signaling systems and builds on ideas from iterated best response models (Jäger 2007(Jäger , 2012Franke 2009) and more thoroughly probabilistic variants of them (Camerer et al 2004;Frank & Goodman 2012;Russell 2012). The crucial feature of our model is that it lets discourse participants communicate, not just about the world, but also about the language they are using (Bergen et al 2012(Bergen et al , 2014. From the speaker's perspective, this means that one's intentions in production are characterized in terms of both world information and linguistic information.…”

“…The pattern is similar for β, but the target becomes exchanging information about the language itself. If we set β = 0, then the agents places no value on communicating about the lexicon, and the model reduces to a variant of the lexical uncertainty model of Bergen et al (2014). As β rises, communicating about the language itself becomes more important.…”

“…However, the data in Section 2 show that the lexicon is not known precisely, but rather negotiated during interactions. We now bring that insight into our model using the lexical uncertainty technique first introduced by Bergen et al (2012Bergen et al ( , 2014. The first step is to define a space of lexica L and a probability distribution over them P L :…”

“…Our presentation is somewhat compact. Readers wishing to more fully explore the model are referred to the website for this paper, which provides implementations of this model (as well as those of Frank & Goodman 2012, Bergen et al 2014, and Smith et al 2013) and includes code for calculating all of the examples we review here.…”

“…Introducing 0 also helps ensure that various calculations in the model are well-defined. (For alternative methods of ensuring definedness, see Jäger 2012 andBergen et al 2014 …”