Cylicity and Cyclic Nodes
OWM led to further refinement of some of the universal conditions that restrict the expressive power of the general transformational rules. One refinement we would like to mention in particular concerns the class of cyclic/ bounding nodes that are relevant for Subjacency. In OWM, Chomsky proposes that, besides NP and S0, the category S should be added to the inventory of cyclic nodes. This has the beneficial effect that the impossibility of extracting from a subject NP—so far captured by the ad hoc Subject Condition—can be explained in terms of a general condition, Subjacency (see OWM, 111). In (1), for example, whom crosses two boundaries, NP and S.
(1) *[Whomi did [S[NP your interest in ti] seem to me rather strange]]?
As Chomsky notes (OWM, 112), wh-movement cannot extract a phrase from a nonsubject NP either. In certain cases, this prediction turns out to be correct.
(2) *Whoi did [S John destroy [NP a book [about ti]]]?
In others, extraction from NP appears to be possible.
(3) a. Who did John write a book about?
b. Who did you see a picture of?
For (3a), Chomsky adopts Bach and Horn’s (1976) proposal that the PP about who is in fact not part of the NP at D-Structure but rather a con stituent within VP (see (4a)). Evidence for this is the possibility of pronominalizing the sequence a book, as in (4b).
(4) a. John wrote [a book] [about who]?
b. John wrote it [about Nixon]
As Pronominalization cannot be applied to (3b)—*John saw it of Nixon—Chomsky concludes that (3b) is derived by first applying PP Extraposition to of who and then applying Wh-Movement to the (NP external) PP. Schematically:
(5) Whok did you see [a picture ti] [PP of tk]i?
Although we cannot do full justice to the many interesting facets of OWM in this brief introduction, we believe that the issues we have touched upon sufficiently indicate that OWM forms one of the landmarks in the development of generative syntax, and especially in the develop ment of a general theory of movement. Perhaps most important, it set the stage for a radical shift from a construction-oriented and language specific approach to language to a more principle-oriented approach. Furthermore, it showed that this shift did not have to be at the expense of empirical coverage. Quite the contrary, we would say. This integration of principle-based syntax and empirical coverage also characterized the next stage in the history of generative grammar, the Principles-and Parameters (P&P) approach (Chomsky 1980, 1981, 1986b), which unifies a great many seemingly disparate grammatical phenomena in a conceptually simple and elegant overall framework of principles.
