This dissertation by Mitchell Philip Marcus investigates the hypothesis that natural language syntax can be parsed by a left-to-right deterministic mechanism without parallelism or backup. The hypothesis, explored in the context of English grammar, leads to a simple grammar interpreter with several key properties: 1. **Simple Grammar Rules**: The interpreter can write simple rules that capture generalizations behind linguistic phenomena like passives, yes/no questions, and imperatives, despite the intrinsic difficulty of capturing such generalizations in a processing model. 2. **Constrained Grammar Operation**: The structure of the interpreter constraints its operation such that only very complex, ad hoc grammar rules can parse sentences that violate Chomsky's proposed constraints on grammatical rules. This is achieved through the use of traces and Annotated Surface Structure, derived from Chomsky's work. 3. **Resolution of "Garden Path" Sentences**: The interpreter provides a simple explanation for the difficulty caused by "garden path" sentences, such as "The cotton clothing is made of grows in Mississippi." Special diagnostic rules can resolve local structural ambiguities, but their power depends on a parameter of the mechanism. Most structural ambiguities can be resolved, but those causing garden paths cannot. The paper also demonstrates that the determinism hypothesis necessitates semantic/syntactic interaction to test the comparative semantic goodness of different structural possibilities for an input. The author argues that the properties of the grammar interpreter, which reflect deep properties of natural language, follow from the determinism hypothesis, providing indirect evidence for its truth. The dissertation includes a detailed discussion of the methodology, the notion of "strictly deterministic," and the structure of the parser, PARSIFAL, which embodies these principles.This dissertation by Mitchell Philip Marcus investigates the hypothesis that natural language syntax can be parsed by a left-to-right deterministic mechanism without parallelism or backup. The hypothesis, explored in the context of English grammar, leads to a simple grammar interpreter with several key properties: 1. **Simple Grammar Rules**: The interpreter can write simple rules that capture generalizations behind linguistic phenomena like passives, yes/no questions, and imperatives, despite the intrinsic difficulty of capturing such generalizations in a processing model. 2. **Constrained Grammar Operation**: The structure of the interpreter constraints its operation such that only very complex, ad hoc grammar rules can parse sentences that violate Chomsky's proposed constraints on grammatical rules. This is achieved through the use of traces and Annotated Surface Structure, derived from Chomsky's work. 3. **Resolution of "Garden Path" Sentences**: The interpreter provides a simple explanation for the difficulty caused by "garden path" sentences, such as "The cotton clothing is made of grows in Mississippi." Special diagnostic rules can resolve local structural ambiguities, but their power depends on a parameter of the mechanism. Most structural ambiguities can be resolved, but those causing garden paths cannot. The paper also demonstrates that the determinism hypothesis necessitates semantic/syntactic interaction to test the comparative semantic goodness of different structural possibilities for an input. The author argues that the properties of the grammar interpreter, which reflect deep properties of natural language, follow from the determinism hypothesis, providing indirect evidence for its truth. The dissertation includes a detailed discussion of the methodology, the notion of "strictly deterministic," and the structure of the parser, PARSIFAL, which embodies these principles.