ABSTRACT
Although in general we have a reasonably good idea of constituency, we do not have as good an idea of the internal organization of constituents. A substantial amount of recent research has led to the conclusion that in many cases, a much more elaborate internal structure than previously assumed is necessary. Consider for example the internal structure of the NP the destruction of the boat. It would have been analyzed as in (1) until the mid 1980s but could reasonably be attributed the structure in (2) today (incorporating a combination of various ideas by Szabolcsi, (1987), Abney (1987), Ritter (1991), Valois (1991), Koopman (1993), and simplifying somewhat).
[NP [Det the] [N′ destruction [? of [Np the [boat]]]]]
[DP [D′ the] [Num P [Num′ [Num e] [NP [N′ [destruct]i+ion [VPe[V′ei [VP [DP the boat [V′ e1]]]]]
Every substring forming a constituent in (1) also forms a constituent in (2), but their internal organization is quite different. There are several sources of this kind of development. One undoubtedly is an ever greater scrutiny of detailed data simply ignored previously, which has led to the one (inflectional) morpheme/one head widely assumed now, as a result of the work on head movement (Koopman (1984), Travis (1984), Baker (1988), Pollock (1990)) or the developing consensus in favor of greater crosscategorial uniformity in internal syntactic organization. In this context, what makes syntactic analysis particularly challenging is the fundamental indeterminacy consequential to the existence of silent morphemes (hence, by the one morpheme/one head principle, of silent categories). Arguments (based on paradigmatic regularity) for the existence of silent morphemes (e.g., English present Tense) are as strong as linguistic arguments can get. Once the existence of silent morphemes is accepted, the general question is raised of their content and distribution. Absence of overt morphemes is no guarantee of their structural absence. To illustrate, consider the French small clause in (3).Jean considère [W Pierre malade].
John considers Peter sick.
We face the question of whether W is indeed a constituent—Stowell’s (1981) proposal. Granting that it is, we face the question of whether Pierre is generated within the AP projection of the adjective and whether it stays there. We also face the question of whether the head of W is A or not; and if not, the question of the nature of W.