Order and Structure in Noun Phrases: the mystery of Kîîtharaka
One of the typological universals that has come to the fore in theoretical discussions over the past decade or so is Greenberg’s Universal 20, which makes a claim about the possible orders of modifiers of nouns in the world’s languages. The vast majority of orders found can be understood as deriving from a universal constituency structure looking as follows:
(1) [ Demonstrative [ Numeral [ Adjective Noun ] ] ]
If we allow both possible orders in each constituent, we derive 8 possible orderings (out of 24 logically possible orderings of 4 categories), and those 8 orderings are by far the most frequent in the worlds languages. Experimental work using artificial language learning has suggested that something like (1) is a universal representation that is accessed during learning (e.g. Culbertson and Adger 2014, Martin et al, 2019).
This raises the question of what to do with the other orderings. One possibility is that they are derived from (1) but involve more radical restructuring than simply reordering sisters (a pretty standard approach in generative grammar, following on from Cinque 2005). An alternative, more common in the functionalist literature, is that they involve a radically different grammatical organization, unrelated to (1), so that there is no universal structure for noun phrases, making the experimental findings somewhat mysterious.
In this talk, I’ll look at this issue through the lens of the Bantu language Kîîtharaka, which has a surface order that looks as follows
(2) Noun Demonstrative Numeral Adjective
I’ll contrast two perspectives. In one, representative of the functionalist position and due to Van de Velde 2019, the modifiers are essentially added in apposition to the head noun, giving a flat structure. In the other, following Cinque and (1), they have a hierarchical organization. I’ll argue that only a structure like (1) can account for several subtle scopal effects, constraints on the free ordering of modifiers, the interpretation of elided elements and the behaviour of different classes of adjectives. This means that, the surface order of Kîîtharaka notwithstanding, an abstract hierarchy like (1) is at play. I’ll discuss some of the consequences of this, and some of the theoretical problems raised by accounts of the patterns in the generative literature.