The goal of the proposed research is to develop a theory of how people use knowledge of their language to comprehend sentences they read or hear. Experiments are proposed that examine the cognitive processes of normal, intact adults during reading or listening. The experiments are designed to uncover the principles that govern readers' and listeners' use of grammatical information in sentence parsing, and is predicated on the assumption that these principles will form a distinct subset of the principles that govern cognition in general. The primary innovation of the present proposal is its claim that two different types of phrases are processed quite differently. "Primary relations" (arguments and obligatory constituents of main predicates, and their arguments and obligatory dependents) are processed according to principles we have advanced earlier. They are attached into a single, determinate syntactic structure which is then interpreted semantically. "Nonprimary relations" (phrases which cannot be primary relations) are associated with a thematically-defined domain of a sentence, and semantic and nonlinguistic sources of information can be used (following a process we call "construal") to determine their specific association within this domain. The proposed research uses a variety of techniques of experimental cognitive psychology to test these new claims about construal and to continue our previous research on syntactic processing principles. It is designed to contribute to the development of psycholinguistic theory. Recent advances in psycholinguistic theory have had substantial impact on the analysis of language disorders, especially aphasia. We anticipate that these new analyses will lead to increased understanding and, hopefully, improved treatment of such disorders.