Abstract
This paper describes how intelligent agents modeled within the OntoAgent cognitive architecture treat the linguistic phenomenon of modal scope ellipsis. The approach offers two innovative aspects with respect to cognitive modeling: (1) ellipsis treatment is distributed across processing modules in a cognitively plausible way and (2) agents are prepared to incorporate calculations of utility into their ellipsis resolution efforts. The latter means that agents can evaluate their confidence in each ellipsis resolution decision and, in cases of low confidence, determine whether or not it is worthwhile to pursue a clarification. Endowing agents with such decision making capabilities about language processing creates an environment in which it is feasible and useful to attempt even the more sophisticated aspects of language processing in the near term.
