Compact storage of syntax trees for ambiguous sentences

We discuss the computation of parse forests, i.e., compact representations of all syntax trees for a given sentence. Conventional parsing algorithms can be adapted in a simple way to produce parse forests. Recently, the subject of parse forests was brought up in the context of generalized LR(O) parsing, a new parsing method for general context-free grammars. It is widely held that this parsing method is particularly suited to producing parse forests. The contrary is true: generalized LR(O) parsers are poor tools for producing compact parse forests.