The reversible grammar, together with the notion of derivation that underlies the uniform algorithm, constitute the grammatical competence base of a natural language system. The grammar declaratively describes the set of all possible grammatical well-formed structures of a language and the uniform algorithm is able to find all possible grammatical structures for a given input - at least potentially.
The monitoring and revision methods are designed to improve a system's performance in order to obtain an effective and flexible use of language. Thus they belong to the performance part of a natural language system.
The item sharing approach can be seen as a kind of mediator between the competence and performance methods because it is a straightforward extension of the competence base, particularly designed to support efficient interleaving of parsing and generation.
Since the competence base plays an important role for realizing these methods, the approach followed in this thesis should be seen as a step towards a competence-based performance model of natural language.
Clearly, additional mechanisms will be necessary for increasing the robustness, efficiency and flexibility of natural language systems. For example, machine learning methods as well as statistical or preference-based methods have to be developed and integrated into such a model in order to improve the system's performance by experience, to handle ill-formed or under-specified input, or to realize specific control strategies (see, for example, [Uszkoreit1991, Barnett1994, Samuelsson1994, Neumann1994, Wirén1992]).
The results presented in this thesis - especially the uniform tabular algorithm and the item sharing method - are important foundational contributions for a competence-based performance model and they should be seen as part of a long-term scientific program for achieving such a model of natural language. In the final chapter of this thesis we outline how our new model can fruitfully be combined with the above mentioned high-level performance methods.