Adjectival versus Nominal categorization processes: The Rule versus Similarity hypothesis

This paper presents the thesis that adjectives and nouns trigger processing by two different cognitive systems, those reported in the literature to be involved in the processing of rulebased versus similarity-based artificially construed categories, respectively. To support this thesis, the paper illuminates a number of links between findings reported in the literature concerning ruleversus similarity-based categories and a variety of corresponding structural, distributional, neural and developmental findings reported in the literature pertaining to adjectives and nouns. These links suggest that the rule versus similarity (RS) hypothesis for the adjective-noun word class distinction should be studied more directly in the future, by empirical means such as those available in the field of psycholinguistic research of rules and similarity. As part of the rule versus similarity hypothesis, the paper defends the thesis that nouns are conceptually gradable, but, unlike adjectives, their ordering dimensions are processed implicitly. Furthermore, nouns differ from multidimensional adjectives in that the nominal dimensions are incorporated through weighted mean operations (averaging), while adjectival dimensions are integrated through Boolean operations (such as and and or). Finally, the implications of the hypothesis are examined against several non-paradigmatic sub-classes of nouns and adjectives, including nominalizations, +human nouns, relational adjectives and Determiner-Adjective constructions. 1 The noun-adjective puzzle Scholars generally agree that natural languages provide evidence for a typology of predicates consisting of word classes such as nouns, adjectives and verbs. 1 While nouns tend to occur in argument position, where their main function is to refer to objects, adjectives typically predicate of or modify arguments, thus occurring in predicateor modifier-positions. Nonetheless, unlike verbs, adjectives rarely inflect for tense and aspect and in many languages for adjectives to occur in predicate position they have to combine with a copula or an affix. Formal semanticists typically classify these different word classes uniformly as ̳predicates‘, as in classical logic or other logical systems, disregarding the typologically dominant morphological, syntactic, functional and intuitive differences. While some semantic theories distinguish between verbs and other predicate types by taking verbs to denote actions or other event types, what distinguishes between adjectives and nouns? People intuitively suppose that nouns (like, for example, bird) denote ̳object categories‘, while adjectives (like, for example, red) denote ̳properties‘. However, the difference between categories and properties is not well defined. After all, one cannot deny that red refers to the set of red objects and bird to the property of being a bird. Addressing this issue, I propose that nouns and adjectives trigger processing by different cognitive systems. Empirical results demonstrate that grammatical categories have a neuroanatomical basis; for example, Miozzo et al 2010 review reports about bilingual aphasics involving disruption selectively affecting the production of verbs versus nouns and regular versus irregular verbs. Critically, these selective deficits were manifested in a strikingly similar manner across the two languages spoken by each of the individuals. To take another example, Cappelletti, Fregni, Shapiro, Pascual-Leone