The Typology of Tri-Nominal Compounds in Romance: The Case of Italian

This paper presents a corpus-based typology of Italian trinominal compounds (NNN), focusing on their internal structure, distribution, and productivity. Drawing on data extracted from the ItWac corpus, the study demonstrates that Italian formations consisting of three adjacent nouns are not structurally uniform, but fall into two distinct types: recursive binominal compounds, and relational compounds. While both share the same surface configuration, they differ fundamentally in their internal organisation and interpretive mechanisms. Recursive binominal compounds result from the embedding of one N+N compound within another, yielding either right-branching or, more marginally, left-branching structures. Their distribution is shown to be strongly constrained by lexical “trigger” elements, typically head nouns, which license embedding in the non-trigger position. The data further reveal that recursive compounding is not restricted to verbal-nexus constructions, as previously assumed, but also productively involves subordinate grounding compounds. Relational compounds, by contrast, constitute inherently trinominal structures in which a simple head combines with a non-autonomous binominal modifier. The modifier typically encodes a coordinate relation (often interpreted as ‘between x and y’), whose interpretation is fully licensed by the head noun. Two key properties characterize this class: the lack of independent lexical status of the modifier and the polyvalency of the head. The analysis shows that both types are sufficiently frequent in contemporary Italian to be considered productive word-formation patterns rather than marginal or nonce formations. More importantly, the findings demonstrate that Italian allows the embedding of complex material within NN compounds.