ABSTRACT
Words and concepts are obviously useful units for the analysis of language. A number of representational mechanisms have been proposed to capture their meaning, such as feature systems (Katz & Fodor, 1963; E. E. Smith, Shoben, & Rips, 1974) or semantic networks (Collins & Quillian, 1969). There are, however, other levels of analysis that have proved useful in the study of language, both finer grained units (such as syllables or phonemes) and coarser grained ones. A particularly important level of analysis has been the propositional level. The term derives from logic but has been widely used in linguistics and psychology to designate units consisting of one or more concepts that function as arguments (with specific semantic roles) and a relational term. In one form or another such units played an important role in the work of linguists like Fillmore (1968), Bierwisch (1969), or van Dijk (1972). Psychologists, too, felt a need for units of this kind, either talking informally about "idea units," or using propositional analysis, following the lead of linguists (e.g., Kintsch, 1974). Indeed, the addition of propositional analysis to the toolbox of psychologists played a major role in their efforts to understand discourse comprehension and to model it.
