Determiner phrase

In linguistics, a determiner phrase is a type of phrase headed by a determiner such as many. Controversially, many approaches take a phrase like not very many apples to be a DP headed, in this case, by the determiner many. This is called the DP analysis or the DP hypothesis. Others reject this analysis in favor of the more traditional NP analysis where apples would be the head of the phrase in which the DP not very many is merely a dependent. Thus, there are competing analyses concerning heads and dependents in nominal groups. The DP analysis developed in the late 1970s and early 1980s, and it is the majority view in generative grammar today.
In the example determiner phrases below, the determiners are in boldface:

a little dog, the little dogs
my little dog, your little dogs
this little dog, those little dogs
every little dog, each little dog, no dog
The competing analyses

Although the DP analysis is the dominant view in generative grammar, most other grammar theories reject the idea. For instance, representational phrase structure grammars follow the NP analysis, e.g. Head-driven Phrase Structure Grammar, and most dependency grammars such as Meaning-Text Theory, Functional Generative Description, and Lexicase Grammar also assume the traditional NP analysis of noun phrases, Word Grammar being the one exception. Construction Grammar and Role and Reference Grammar also assume NP instead of DP. Noam Chomsky, on whose framework most generative grammar has been built, said in a 2020 lecture,

I’m going to assume here that nominal phrases are actually NPs. The DP hypothesis, which is widely accepted, was very fruitful, leading to a lot of interesting work; but I’ve never really been convinced by it. I think these structures are fundamentally nominal phrases. As far as determiners are concerned, like say that, I suspect that they are adjuncts. So I’ll be assuming that the core system is basically nominal.

The point at issue concerns the hierarchical status of determiners. Various types of determiners in English are summarized in the following table.

Article	Quantifier	Demonstrative	Possessive
a/an, the	all, every, many, each, etc.	this, that, those, etc.	my, your, her, its, their, etc.

Should the determiner in phrases such as the car and those ideas be construed as the head of or as a dependent in the phrase? The following trees illustrate the competing analyses, DP vs. NP. The two possibilities are illustrated first using dependency-based structures :
The a-examples show the determiners dominating the nouns, and the b-examples reverse the relationship, since the nouns dominate the determiners. The same distinction is illustrated next using constituency-based trees, which are equivalent to the above:
The convention used here employs the words themselves as the labels on the nodes in the structure. Whether a dependency-based or constituency-based approach to syntax is employed, the issue is which word is the head over the other.

Arguments for DP over NP

The DP-hypothesis is held for four main reasons: 1) facilitates viewing phrases and clauses as structurally parallel, 2) accounts for determiners often introducing phrases and their fixed position within phrases, 3) accounts for possessive -s constructions, and 4) accounts for the behaviour of definite pronouns given their complementary distribution with determiners.

Parallel structures

The original motivation for the DP-analysis came in the form of parallelism across phrase and clause. The DP-analysis provides a basis for viewing clauses and phrases as structurally parallel. The basic insight runs along the following lines: since clauses have functional categories above lexical categories, noun phrases should do the same. The traditional NP-analysis has the drawback that it positions the determiner, which is often a pure function word, below the lexical noun, which is usually a full content word. The traditional NP-analysis is therefore unlike the analysis of clauses, which positions the functional categories as heads over the lexical categories. The point is illustrated by drawing a parallel to the analysis of auxiliary verbs. Given a combination such as will understand, one views the modal auxiliary verb will, a function word, as head over the main verb understand, a content word. Extending this type of analysis to a phrase like the car, the determiner the, a function word, should be head over car, a content word. In so doing, the NP the car becomes a DP. The point is illustrated with simple dependency-based hierarchies:
Only the DP-analysis shown in c establishes the parallelism with the verb chain. It enables one to assume that the architecture of syntactic structure is principled; functional categories consistently appear above lexical categories in phrases and clauses. This unity of the architecture of syntactic structure is perhaps the strongest argument in favor of the DP-analysis.

Position

The fact that determiners typically introduce the phrases in which they appear is also viewed as support for the DP-analysis. One points to the fact that when more than one attributive adjective appears, their order is somewhat flexible, e.g. an old friendly dog vs. a friendly old dog. The position of the determiner, in contrast, is fixed; it has to introduce the phrase, e.g. *friendly an old dog, *old friendly a dog, etc. The fact that the determiner's position at the left-most periphery of the phrase is set is taken as an indication that it is the head of the phrase. The reasoning assumes that the architecture of phrases is robust if the position of the head is fixed. The flexibility of order for attributive adjectives is taken as evidence that they are indeed dependents of the noun.

Possessive ''-s'' in English

Possessive -s constructions in English are often produced as evidence in favor of the DP-analysis. The key trait of the possessive -s construction is that the -s can attach to the right periphery of a phrase. This fact means that -s is not a suffix. Further, the possessive -s construction has the same distribution as determiners, which means that it has determiner status. The assumption is therefore that possessive -s heads the entire DP, e.g.

's dog
's scarf

The phrasal nature of the possessive -s constructions like these is easy to accommodate on a DP-analysis. The possessive -s heads the possessive phrase; the phrase that immediately precedes the -s is in specifier position, and the noun that follows the -s is the complement. The claim is that the NP-analysis is challenged by this construction because it does not make a syntactic category available for the analysis of -s, that is, the NP-analysis does not have a clear means at its disposal to grant -s the status of determiner. This claim is debatable, however, since nothing prevents the NP-analysis from also granting -s the status of determiner. The NP-analysis is however forced to acknowledge that DPs do in fact exist, since possessive -s constructions have to be acknowledged as phrases headed by the determiner -s. A certain type of DP definitely exists, namely one that has -s as its head.

Definite pronouns

The fact that definite pronouns are in complementary distribution with determiners is taken as evidence in favor of DP. The important observation in this area is that definite pronouns cannot appear together with a determiner like the or a in one and the same DP, e.g.

they
*the they
him
*a him

On a DP-analysis, this trait of definite pronouns is relatively easy to account for. If definite pronouns are actually determiners, then it makes sense that they should not be able to appear together with another determiner since the two would be competing for the same syntactic position in the hierarchy of structure. On an NP-analysis in contrast, there is no obvious reason why a combination of the two would not be possible. In other words, the NP-analysis has to reach to additional stipulations to account for the fact that combinations like *the them are impossible. A difficulty with this reasoning, however, is posed by indefinite pronouns, which can easily appear together with a determiner, e.g. the old one. The DP-analysis must therefore draw a distinction between definite and indefinite pronouns, whereby definite pronouns are classified as determiners, but indefinite pronouns as nouns.

Arguments for NP over DP

While the DP-hypothesis has largely replaced the traditional NP analysis in generative grammar, it is generally not held among advocates of other frameworks, for six reasons: 1) absent determiners, 2) morphological dependencies, 3) semantic and syntactic parallelism, 4) idiomatic expressions, 5) left-branch phenomena, and 6) genitives.

Absent determiners

Many languages lack the equivalents of the English definite and indefinite articles, e.g. the Slavic languages. Thus in these languages, determiners appear much less often than in English, where the definite article the and the indefinite article a are frequent. What this means for the DP-analysis is that null determiners are a common occurrence in these languages. In other words, the DP-analysis must posit the frequent occurrence of null determiners in order to remain consistent about its analysis of DPs. DPs that lack an overt determiner actually involve a covert determiner in some sense. The problem is evident in English as well, where mass nouns can appear with or without a determiner, e.g. milk vs. the milk, water vs. the water. Plural nouns can also appear with or without a determiner, e.g. books vs. the books, ideas vs. the ideas, etc. Since nouns that lack an overt determiner have the same basic distribution as nouns with a determiner, the DP-analysis should, if it wants to be consistent, posit the existence of a null determiner every time an overt determiner is absent. The traditional NP analysis is not confronted with this necessity, since for it, the noun is the head of the noun phrase regardless of whether a determiner is or is not present. Thus the traditional NP analysis requires less of the theoretical apparatus, since it does not need all those null determiners, the existence of which is non-falsifiable. Other things being equal, less is better according to Occam's Razor.