How nominal compounds are modified by two adjectives

Abstract

Basing itself on a corpus of one thousand complex NPs, this study investigates the relationships that two attributive adjectives contract with the constituents of nominal compounds of varying size in English (e.g. new basic safety standards). Essentially, there are four logically possible relationships: (i) both adjectives modify the nominal head, (ii) both adjectives modify the nominal modifier, (iii) the first adjective modifies the head and the second adjective the modifier and (iv) the first adjective modifies the modifier and the second adjective the head (crossed modification). While options (i) and (iii) are strongly represented in the data, crossed modification is not at all present. Across all compound sizes, at least three factors shape the empirical patterns: a functional factor whereby major heads are more easily singled out than minor heads, which in turn are more available than modifiers; a structural factor whereby more deeply embedded constituents are less available than more independent constituents; and a proximity effect which encourages the modification of the first noun by the second adjective. There may be an additional saturation effect which discourages the modification of one noun by two adjectives. On the face of it, the non-occurrence of crossed modification may be connected to the well-known ban on crossing association lines. However, despite its descriptive adequacy, this principle is unconvincing. Instead, a functional explanation is proposed which centres on the possibility of working out modification relationships. Initial steps are taken towards developing a model of when (and why) the no-crossing constraint is inviolable, violable or non-existent.