Abstract
The overreliance on introspection and the use of acceptability judgements performed just by a single linguist have proved as a "serious methodological weakness affecting much research in syntax and semantics" (Gibson & Fedorenko, 2010: 233). I demonstrate that this methodological weakness has hampered the research on noun valency (at least) in English.
It is usually claimed that with valent nouns the omission of a complement occurs much more often than with verbs, that with nouns "this complement is never obligatory" (van Hout & Muysken, 1994: 55), or that "for verbs, syntactic realization of semantic participants is obligatory, while for nominalizations it is optional" (Meinschaefer, 2003: 234; cf. Quirk et al., 1985: 62; Panevová et al., 2014: 90; Karlík & Nübler, 1998: 111).
This is typically taken as uncontroversial and possibly universal, which often leads to claims that (English) nouns lack valency, or that only complex-event nominals have argument structure (a claim put forward influentially by Grimshaw, 1990; cf. Alexiadou, 2017).
However, as I illustrate using data from the British National Corpus (BNC), these claims are hard to square with empirical evidence. It is hypothesized that when a noun is not contextually bound, i.e. when it is first introduced to discourse (which may be marked by the indefinite article in English), its complement is in fact obligatorily expressed.
Based on the analysis of several potentially valent English nouns, this is confirmed. It can be concluded that the common view that complements of valent nouns are never obligatory cannot be squared with empirical evidence, that the common view that only complex-event nominals have valency properties is somewhat misguided (ability is hardly one), and that we need more studies relying on empirical data rather than introspection.