The structured definition of language
Trevor Chorvat · The University of Queensland · 1996
What it does ... This thesis extends language technology by applying the principle of separation of concerns to language definition. The motivation for doing this is to make language technology more accessible to language users and to bring language definition into synchrony with software engineering principles. As a byproduct this work extends the effective scope of formality and predisposes language processing to a parallel and distributed computational model. Why it is needed ... Language definition has traditionally been broken up into sequential and nested sub-languages, but not hitherto into parallel sub-languages. By “parallel” it is meant that a number of sublanguages are recognised concurrently as the single input stream (i.e. putative sentence) is looked at from a number of different views. The motivation for this approach is that the different views would provide for the separation of concerns, thus being able to manage the complexity of a definition in a structured way. In accord with the goal of making the overall definition comprehensible, a conscious decision has been made to keep each sub-language within manageable complexity. This work accepts context-free grammars as empirically comprehensible and adds only those notions and notations necessary to support the extension to parallel sub-languages, or facets. What it is ... The resulting Multifaceted Grammars (MFGs) can primarily be seen as extending context-free grammars with the concepts and mechanisms required to support multiple context-free grammars and the communication between them. This is achieved by augmenting context-free productions with a mechanism to specify the communication of tokens and productions to and from other parsers. Mapping of information from one vocabulary into another is also supported. Fundamental linguistic capabilities such as symbol tables and scoping are directly provided for. What it has been used for ... In this thesis Multifaceted grammars are applied to the static semantic definition of Modula-2. This application demonstrates the feasability of using Multifaceted grammars for practical language definition and illustrates the more declarative style of definition attainable. The experience also highlighted, however, that separation, by itself, does not bring a high degree of re-use.