1.One-noun NPs, e.g., John, students,
2.Determinative + N, e.g., that book, Alison’s divorce,
3.Determinative + modifier + N, e.g., the unpleasant boy,
4.Determinative + N + modifier, e.g., the dog on the sofa.
Is there any general pattern here? We can easily formulate a general principle for cases 3 and 4 if we say that dependents other than determinatives combine to form nominals, whether those dependents appear before or after the head noun, and determinatives combine with nominals to form NPs.
Case 2 can be unified with this same formulation if we assume that book in that book or divorce in Alison’s divorce also constitute one-word nominals. We have already seen one-word phrases, so this assumption is not a stretch. In that case, our diagram would look like this:
There is support other than theoretical symmetry for this analysis. One, which as we have seen substitutes for a nominal, can replace book alone. For example:
(18) This book has water damage, but that one is in perfect condition.
For this reason, we will assume that there is always a nominal level in every NP. As a practical matter, however, diagrams that show every single nominal become unwieldy and harder to read. We add a label that, because it is completely predictable, doesn’t add much useful information. So in our diagrams, we will only show a nominal node if it branches.
What about case 1? Our assumption about nominals will apply here too. In other words, if we were diagramming non-branching nominals, a diagram of a one-word NP would look like this:
The only thing that distinguishes this case from the others is the lack of a determinative. We will call such NPs bare because of this absence. One way to make our analysis consistent for all cases would be to represent the determinative slot as present but not filled by any audible word. In other words, we assume that every NP is formed by combining a determinative with a nominal.
The question then becomes, what are we to make of this empty slot, which I have represented with the character ?. Much of the recent technical literature on syntax assumes that there is actually something in the slot, a silent determiner, often called a zero determiner. According to this view, the zero determiner behaves like other determiners in the sense that it helps specify the interpretation of the nominal. Notice, for example, that the meaning of the bare NP cats is not the same as the determined NP the cats. Of course that change in meaning is no proof that there is actually a silent determiner present in the bare NP. We could also simply say that the determiner slot in such cases is truly empty and attribute the difference in meaning to the absence of a determiner rather than the presence of a silent one. The theories that posit zero determiners typically have theory-internal reasons for doing so. But with the scheme that we are developing here, there is no particular reason to prefer one hypothesis over the other, and so we will apply Occam’s razor and assume that there is, in fact, no determiner. Further, in keeping with our attempt to keep our diagrams free of unnecessary clutter, we will not diagram these empty determinative slots. In other words, we will diagram one-word NPs as shown earlier in this chapter, showing neither the nominative level nor the empty determinative.
Bare NPs do not always consist of one word. How should we represent phrases like stray cats? If we drew a diagram showing all our levels, it would look like this:
If we remove the empty determinative slot from our diagram, we get the following:
Although there’s nothing wrong with this representation, we can simplfy our diagrams still further and omit the nominal level in this case too. This leaves us with a relatively simple diagram:
To summarize, we assume that nominals are present in all noun phrases, but the diagrams in this course will only show them if there is a branch both above and below the nominal. If you find it more helpful to show these hidden levels, then by all means put them in your own diagrams, but do so consistently.
 The principle that entities should not be multiplied beyond necessity. In other words, prefer the hypothesis that creates the fewest complications. Postulating a silent entity is more complex than postulating simple absence.