•  
  •  
 

Abstract

This paper is concerned with a computational linguistics analysis of Role and Reference Grammar [RRG] (cf. Van Valin and LaPolla 1997; Van Valin (2005) and introduces research work in progress aiming to analyze the computability of RRG. The concept of computational adequacy is introduced as an important external principle from a philosophy of science perspective to sharpen the scientific principles of the area of functional computational linguistics. In addition, a pseudo-code-based meta-language is developed in order to semi-formalize the linking algorithm from semantics to syntax. This paper will show that RRG in its current fashion is not executable on an abstract machine model - called Random Access Machine - and is therefore not computationally adequate. It is highlighted that the semantics to syntax linking algorithm as proposed in Van Valin (2005) is in fact too coarsely grained to account for the variable undergoer linking in English three-place predicates. Also, the concept of intelligent software agents is introduced in order to account for the functional linguistic approach used in RRG. It will be shown that it is possible to account for variable undergoer linking in three-place predicates using constructional schemas as developed in Nolan (2011). Based on the development of typed feature structures of thematic relations it is possible to show that semantic macroroles as developed in Van Valin (2005) are epiphenomenal. They are an unnecessary concept set on top of thematic relations, which is in conflict with the principle of economy as discussed in Van Valin and LaPolla (1997). It is shown that thematic relations are stored in inheritance networks in the mental lexicon and that they interact with constructional schemas for transfer verbs as they are developed in this paper. The concept of discourse representation structures is also of crucial importance in this paper. It will be shown that variable undergoer linking in English is based on information structure considerations. In order to develop a computationally adequate version of RRG, a revised version of the semantics to syntax linking algorithm is developed.

Share

COinS