ABSTRACT
The recognition and parsing procedures described and implemented in previous chapters have all been versions of a top-down, depth-first, left-to-right strategy (with backtracking). With regards to the phrase structure grammars of chapter 10 and the DCGs of chapter 11, these strategies were seen to be a consequence of the in-built proof procedure of Prolog. However, as noted in section 10.1, a context-free phrase structure grammar (and, by extension, a DCG) is compatible with a number of different procedural interpretations. If Prolog were only suitable for expressing the one strategy, this would severely restrict its usefulness as a tool for NLP research. However, one of the strengths of Prolog is the ease with which it is possible to devise other control strategies. This is the subject matter of the present chapter.
