On the relations over representations of linguistic structure and grammars

A familiar and fairly well-known distinction exists between a derivational type of grammar (for example, mainstream Generative Grammar) and a representational type of grammar (Lexical-Functional Grammar, Head-Driven Phrase Structure Grammar, etc.). As far as the derivational type is concerned, the derivational process goes on in an incremental manner through a series of structural alterations involving structural constraints, whereas in the case of the representational type as soon as the linguistic structure is built, it is checked for conformity to certain representational constraints. This paper argues that the two types are not simply mutually exclusive choices for the representation of linguistic structure. Rather, they constitute and also reflect two distinct and yet parallel modes of knowledge representation of language vis-à-vis the abstract (axiomatic) system of language from a metatheoretical perspective. That they are sometimes equivalent in expressing linguistic facts and sometimes divergent in descriptions of other linguistic structures is explained by appealing to the idea that the formal representation of language is bimorphic but not in terms of the same morphism. There exists a morphism instantiating an epimorphism (in the category-theoretic sense) that maps between categories of objects designating linguistic entities and procedures, establishing the divergence, while the case for equivalence can be simply treated as monomorphic (in terms of the category-theoretic notion of monomorphism). Hence, it leads to a split bimorphic representation of language. Then it is shown how divergent psycholinguistic findings on the conflicts between the derivational type of grammar and the representational type can be accommodated by appealing to the present model. Overall, this essentially shows that choices of representation of linguistic structure are partly determined by cognitive constraints/principles and any uncertainty between such choices can be accommodated in the current model that can admit of both entanglement and flipping between choices of representation of linguistic structure.