Parsing incomplete sentences revisited

We describe a context-free parsing algorithm to deal with incomplete sentences, including unknown parts of unknown length. It produces a finite shared-forest compiling all parses, often infinite in number, that could account for both the error and the missing parts. In contrast to previous works, we derive profit from a finer dynamic programming construction, leading to an improved computational behavior. We also introduce a deductive construction, which has the advantage of simplifying the task of description.