Inflection and derivation are central concepts in morphology. Inflection characterizes the relationship between
different forms of the same lexeme (bright-brighter-brightest), derivation the relation between lexemes (bright-brightness). In standard English grammars (Biber et al. 1999, Huddleston & Pullum 2002), the -ly in brightly is derivational. But in most theoretical accounts, the same -ly is argued to be inflectional (Giegerich 2012, Pittner 2014), or simply not categorizable (Bauer et al. 2013). This thread, drawing on Schäfer (to appear), presents the core problem that makes the classification controversial, and argues, based on distributional semantics analyses, that the heterogenuity of the adjective classes involved is a further point to consider.

There is agreement that the central feature distinguishing -ly and non-ly forms is that -ly forms can only be used as adverbials, while non-ly forms can’t. But this does not rule out an inflectional analysis. While the positive and superlative forms of English adjectives add independent meaning and are clear examples of inherent inflection, the -ly forms can be analysed as contextual inflection, inflection that is required by the syntactic context alone. Proponents of derivational and inflectional analyses both consider the lexcial meanings of the two forms to be the same. While this speaks for an inflectional approach, I show that actually the lexical meanings require different kinds of adaptions across different adjective classes. Human propensity adjectives (“intelligent”, “stupid”) naturally go together with animate objects, which are prototypical noun referents. In contrast, speed adjectives (“quick”, “swift”),
naturally go together with events, prototypical verb referents. The class differences show traces in distributional semantic comparisons. I argue that this heterogenuity is plausibly behind the preferences of some authors for an inflectional and others for a derivational analysis.