Explaining language structure

Typically, certain grammatical features are associated with one particular lexical category rather than some other category. Nouns can be modified by numerals or adjectives, can take determiners like demonstratives or possessive attributes, can be inflected for number, case, etc. Verbs, by contrast, take markers of tense, aspect, modality and can be negated, etc. But cross-linguistic observations show that one and the same linguistic expression can also be associated with more than one grammatical category. For example, in many languages there are forms that serve the expression of verbal tense or aspect in some of their uses but behave like lexical verbs in other uses; adpositions may be homophonous with nouns, or relative clause markers with demonstratives, etc. In many of these cases, grammaticalization theory has been used to account for such situations.The present paper argues that this framework is also able to explain connections between linguistic structures that appear to be entirely unrelated to one another. Such a case of “categorial misbehavior” is reported from the Walman language of Papua New Guinea, where two ‘and’-conjunctions that have the function of conjoining noun phrases have the morphological structure of transitive verbs. Drawing on typological evidence from a number of genetically and areally unrelated languages, the paper proposes a reconstruction of the situation in Walman based on regularities of grammatical change. The main goal of the paper is to argue that grammaticalization theory can provide explanations that appear to be beyond the potential of other linguistic frameworks. Such explanations are external rather than internal, and they are restricted to the question of why languages are structured the way they are, that is, they concern neither the question of how people use their language nor what knowledge they have about their language.