Dynamic semantics for a controlled natural language

In this paper I present a dynamic approach for constructing an unambiguous semantic representation for a text written in a controlled natural language called PENG. The semantic representation is built up incrementally - in left-to-right order - while the user of the PENG system writes the text. For each word form that the user types, the language processor creates a partial logical structure that is analysed in the context of a given information state and then either updates that information state or delays this process if necessary. After each processing step, the language processor reports what kind of syntactic categories can follow the current input. These predictive grammatical hints guide the writing process of the user and enforce the rules of the controlled language. The result is an unambiguous text in controlled natural language that has first-order equivalent properties and that can be further processed by a computer.