Townsend, D.

Montclair State University

People form mental models of situations in the world and in language (Johnson-Laird,1983; Radvansky & Zacks, 2014). In language, they may form these models incrementally (Paczynski et al., 2014; Poppolo et al., 2016) or they may underspecify them (Pickering et al., 2006).

Two eye tracking experiments show that:

(a) Telic predicates increase first pass time on durative modifiers (climbed a mountain for several days). This difficulty disappears when the likelihood of a situation is controlled or when a result state model is available (halted a class for several days) (Pinon, 1999), indicating that incremental processing occurs when situational likelihood facilitates finding a plausible temporal distribution of a situation (Deo & Pinango, 2011) or an available state model reduces prediction error (Zacks et al., 2007).

(b) Atelic predicates increase total time but not first pass time on frequency modifiers (admired a mountain every day), indicating an underspecified model until the processor matches sub-intervals to distinct occasions of the unbounded situation (Deo et al., 2012; Rothstein, 1995).

Thus, the timing of commitments depends on the availability of information about a plausible temporal distribution of the situation. This information includes knowledge of situational likelihood and the meanings of verbs and modifiers.