The position of adjectives and other phrasal modifiers conclusion

In conclusion, Cinque’s general idea that the fine syntactic structure of the DP can be put to use to constrain adjectival ordering is superior to any alternative thus far available. However, it does not seem necessary to go so far as to introduce dedicated functional heads for each adjective encountered.

The idea advanced here is that Universal Grammar (UG) dictates a structure for argumental noun phrases that necessarily involves a fair amount of functional structure, following Zamparelli (2000), Baker (2003), Borer (2005a), and others. The layers of functional structure represent stages of building referential argumental DPs from the abstract concepts associated with roots. Languages may then (and generally do) invent ways to modify the different layers of structure. A typical modifier for the DP level is a demonstrative, though other kinds of modifiers are possible; Zhang (2004) argues, in effect, that Chinese has a way of modifying the D layer with a relative clause. A demonstrative may be grammaticized as a D head when a reanalysis takes place from the demonstrative being a phrasal adjunct to DP in one generation to being a head of D in another.

Similarly, a language may over time invent a way to modify the unitP level, with various quantifiers or numeral phrases. Again, some of these may become grammaticized as heads of unitP.

And so on down the line. UG provides the basic ingredients for the category Adjective, and all languages appear to avail themselves of it in one form or another. The adjective turns out to make a particularly suitable modifier for some of the lower levels of the DP, but exactly how this is done varies substantially from language to language. What is most common, judging from Dixon’s (2004a) typological survey, is that adjectives expressing dimension, age, value, and color are developed and combined with some functional structure to create nP modifiers or sortP modifiers.

There seem to be at least two ways in which this might occur.¹ If the functional structure takes the adjective as complement, changing its type into that of a modifier of, say, sortP, then the adjective may be able to apply iteratively within a projection. This is the case, for example, in Jarawara, according to Dixon (2004b), which has only fourteen adjectives but may use more than one in a single DP, with apparently free ordering. Another possibility is that a language might use the nominal structure itself to introduce the adjectival modifier, e.g. by allowing n or sort to take an adjectival specifier. In such cases, a single adjective (of any given type) would be the norm. This seems to be the situation in Wolof, as described by McLaughlin (2004), where an adjective may cooccur with a relative clause, but two adjectives cooccur only if coordinated.²

Eventually, a language might innovate ways of modifying each layer of the

 DP, either with specialty inventories of modificational elements for each layer, or, as in English, with a large class of adjectives being compatible with more than one functional option. The impression of a large class of adjectives comes then from the fact that there are many roots that can be used as adjectives, and from the fact that the functional heads introducing NP, nP, and sortP modifiers are not morphologically distinct. The impression of strict ordering comes mainly from the fact that sortP modifiers are strictly ordered before nP modifiers (and both are strictly ordered before NP modifiers), and that iteration within a layer is avoided when possible. Thus, pairs of adjectives will normally be arranged, in English, so that one is an nP modifier and the other is a sortP modifier. The one which is the most sensitive to shape, or the one which is the most robustly gradable, will then appear as the sortP modifier.

1 See discussion in Truswell (2004), including discussion there of work by Albert Ortmann, which I have not read.

2 Interesting in this regard is the characterization of Chinese presented in Sproat and Shih (1991), where it is claimed that direct (i.e. without de) modification of a noun is possible for exactly one adjective expressing size or quality, and one expressing color or shape, and if the two cooccur it is in strict order: quality/size>shape/color. This would suggest that Chinese provides a position for a single quality or size adjective above sort, and a position for a single shape or color adjective below sort. The element de makes iteration possible.