Grammatical alignments and the gradience of lexical categories

The notion of part of speech dates back at least to 5th century BC Sanskrit grammar (Matilal, 1990, ch.3) and is adopted in one form or another by most current linguistic theories. In spite of this, no objective definition for part of speech labels has ever been put forth (Croft, 1984, 2001; Langacker, 1987; Aarts, 2007). The present work explores the idea that parts of speech and their syntactic distribution may be the result of language-specific alignments of several linguistic aspects: meaning (in terms of actions, individuals, or properties), semantic function (i.e. predicates, arguments, and modifiers), and morphosyntactic properties. Since Sign-Based Construction Grammar (Sag, 2010, 2012) is in an ideal position to capture such alignments in terms of fine-grained featural and constructional specifications, a grammar fragment of English is formulated along these lines, with the goal of obtaining a linguistic model where the various phrasal classes of English emerge without appeal to part of speech labels of any kind. Of special interest are the various environments which tolerate a wide range of syntactic categories, such as predicative structures and ‘coordination of unlikes’ phenomena. It has long been recognized that one of the remarkable traits of predicative phrases is that they can be instantiated by a wide range of different categories. For example, verbs like be and remain allow for NP, AP, PP, and VP complements as shown in (1).