| Description |
In this paper, we argue for a tripartite typology of expressive nouns (ENs) which can occur in Italian evaluative binominal structures (QBNPs), where ENs signal the emotional judgment of the speaker on the (situation featuring the) the referential noun (RN). We show that (i) C-ENs – as in un cazzo di professore ‘a fucking professor’ (lit. ‘a dick of a professor’) – are pure expressive items that convey an intensifier semantics, whereas (ii) M-ENs – e.g. una merda di professore ‘a shithead professor’ (lit. ‘a shit of professor’) – and S-ENs – e.g. uno sttronzo di professore ‘an asshole of a professor’ (lit. ‘a turd of professor’) – are mixed expressive items, which feature both an expressive and at-issue component, and a degree semantics. We also show that M- and S-ENs differ from each other in that (iii) S-ENs require an animate RN, whereas M-ENs do not – una merda di motorino (lit. ‘a shit of scooter’) vs *uno stronzo di motorino (lit. ‘a turd of scooter’) –, and because (iv) only S-ENs have an adjectival core, which allows them to occur in predicative position without the preposition di – un professore stronzo vs *un professore merda, *un professore cazzo. We furthermore illustrate that these three types of ENs show distinct morphosyntactic properties with respect to distributional and concord patterns. The key distributional pattern is that, prenominally, C-ENs can embed M-Ens and S-ENs, while all other stacking patterns are ruled out. The main diNerences in concord patterns is that (i) C-ENs do not show any concord with the RN, nor are they visible to higher functional heads (e.g. D) probing down for phi-feature valuation, (ii) M-ENs show agreement for number with the RN, but keep their own gender, which is agreed with by higher heads, and (iii) S-ENs show full concord with the RN. In our analysis, we argue for a diNerence in degree of grammaticalization of the Italian ENs. C-ENs are fully functional, non-decomposable elements, which lack a lexical core. We argue that they consist of an intensifier feature, possibly the highest of a set of degree projections (Vanden Wyngaerd et al. 2025). For M-ENs and C-ENs, we argue instead that they are semi-lexical elements, which do contain a root, which gets nominalized and which can therefore project its own phi-features. Importantly, though, M-ENs are defective, in that the only valued phi-feature is gender, while the number feature enters the derivation unvalued. In S-ENs, because of their adjectival core, both phi-features enter the derivation unvalued. With respect to the merge position, we argue that C-ENs are merged in the Deg-head of an empty adjective sitting in the Spec,SubComP position of the extended projection of the RN, while M/S-ENs sit in the Spec,ValP position of the extended projection of the RN. Using sample derivations of QBNPs with the three types of ENs, we show that our theoretical assumptions allow us to derive both the observed concord stacking patterns, while ruling out the impossible ones.
|
