A Breadth-First Parsing Model

Recent attempts at modeling humans' abilities at processing natural language have centered around depth first parsing algorithms, and control strategies for making the best choices for disambiguation and attachment. This paper proposes a breadth-first algorithm as a model. The algorithm avoids some of the common pitfalls of depth-first approaches regarding ambiguity, and by using more pre-ccmputed information about the grammar, avoids same of the usual problems of parallel parsing algorithms as well. In the study of computational models of human language processing, cognitive scientists seem to have given little attention to all-paths parsers, focusing instead on depth-first algorithms. This restriction is imposed so that the models will be consistent with the fact that people do not generally perceive ambiguities. In addition, it is an attempt to stay in line with the hypothesis that people parse sentences in linear time. The idea is that the fewer alternatives considered, the faster the parse time should be. The fastest way, of course, would be a depth first parse which made the right choice at every step of the way, hence the interest in deterministic parsers. The attempt to find principles which would guide a parser correctly through a depth-first search, led (Kimball, 1973) to formulate the principles of Right Association and Closure. propose the principles of Lexical Preference and Final Arguments. All of these principles try to account for how a top-down depth-first parser could get the preferred readings of sentences. However, the point I would like to make here is that for each choice point in an ambiguous example, there is a second alternative which the parser needs to be able to get at least some of the time. As (Crain and Steedman, 1981) point out, the fact that people generally only perceive one reading of a sentence is perfectly consistent with a parsing model which finds all the possible syntactic ramifications from looking at a word, does some contextual filtering to decide which alternative(s) to keep, and then looks at a new word and repeats. Given this, it is not obvious that the breadth-first approach is inferior. Since (Earley's, 1970) and (Pratt's, 1975) demonstrations of how a parser can work both bottom-up, and top-down, there have been several proposals for how this information might be used to good effect in a psychological model. (Chester, 1980) proposes a depth-first left-corner parser which uses top-down information. (Martin, Church, and Patil, 1981) propose an …