Linguistic Contributions To The Formal Theory Of Big-Game Hunting—R. Mathiesen Lingua Pranca Contents Metalinguistic Autoreference—Benoît de Cornulier

On Revising And Extending Wh-Movement

D. Terence Nuclear
Institute for Advanced Study
in Strangeness

I assume here familiarity with Chomsky (1977), and the entire literature, published and unpublished, written and unwritten, upon which it is based. Chomsky has been criticized, unfairly in my opinion, for attempting to reduce all of grammar to wh-movement. Rather, if he is open to criticism at all, it is because he has failed to appreciate how wide the domain of wh-movement really is.

In his paper, Chomsky lists the following general characteristics of wh-movement.

1.  It leaves a gap.

2.  Where there is a bridge, there is an apparent violation of subjacency, PIC, and SSC.

3.  It observes CPNC.

4.  It observes wh-island constraints.

In fact none of these characteristics are crucial at all. Rather, what is critical is the following characteristic.

5.  It creates wh-islands.

To see the irrelevance of 1-4 and the crucial character of 5, consider participial relative clauses, like that illustrated in 6.

6.  I once uttered a sentence consisting solely of adverbs.

In such clauses, apparently, wh-movement can affect only subjects, moving them into COMP, after which they are deleted. Such movement only trivially can be said to leave a gap, and by the nature of the application of the rule in such cases, characteristics 2-4 are also trivially fulfilled. More crucially, characteristics 1-4 fail to distinguish between wh-movement, as it applies within participial relative clauses, and Equi, as applied in the derivation of 7.

7.  Someone told me only to utter adverbs.

But clearly, Equi is not wh-movement. Hence we must dispense with characteristics 1-4 and find others that will distinguish between rules like Equi and wh-movement. Such a characteristic is 5. Wh-movement creates wh-islands whereas rules like Equi do not, as the contrast in grammaticality between 8 and 9 shows.

8. *Which constituent did you utter a sentence consisting solely of?

9.  Which constituent did someone tell you only to utter?

At this point, it may be objected that characteristic 5 is itself trivial, since it would seem a matter of definition that the application of wh-movement creates wh-islands. But as Chomsky himself has repeatedly pointed out, poor formulations often lead to better ones, and in this case, we see that 5 can be generalized to express a far deeper and more penetrating insight. We first eliminate the redundancy implicit in 5, yielding 10.

10.  Wh-movement creates islands.

As formulated, 10 expresses the generalization that the domain within which wh-movement applies is an island. An even more significant result would be the demonstration that islands result solely from the application of wh-movement; cf. 11.

11.  A syntactic domain is an island if and only if wh-movement has applied within it.

On the face of it, 11 would seem to be falsified by the existence of such islands and noun complements and sentential subjects. But, as Chomsky has so brilliantly argued, a theory of conditions on rule application cannot be falsified simply by examples, but only by the demonstration that some system of rules not subject to the conditions is superior to the proposed system. Moreover, as I shall now proceed to demonstrate, the island-character of noun complements and of sentential subjects is not only consistent with 11, but is explained by it.

To effect this demonstration, I require as lemmata two principles whose truth is practically self-evident.

12.  The application of wh-movement may be vacuous.

13.  The phrase affected by wh-movement need not literally be a wh-phrase.

In support of 12, we observe that if the phrase to be moved by wh-movement is already in COMP (the target position of wh-movement), then it simply remains there, as 14 illustrates.1

14.  I doubt whether massive doses of vitamin E increase one’s linguistic competence.

In support of 13, we need simply cite the range of phenomena that Chomsky subsumes under wh-movement, none of which involves morphemes spelled with wh, and the fact that wh-movement has counterparts in other languages which lack such morphemes.

Now consider noun complements and sentential subjects, as illustrated in 15 and 16.

15.  I was unaware of the fact that chimp languages lack wh-movement.

16.  That English has nondenumerably many sentences is explained by my theory.

Notice that in both these cases, and unlike cases in which complement clauses are not islands, the appearance of an explicit complementizer is obligatory. This can be explained on the grounds that the complementizer in these cases has undergone vacuous application of wh-movement, and hence must be present in the phrase-marker at the point of application of wh-movement; whereas the complementizer that optionally appears in clauses that are not islands is optionally inserted later, possibly not even in complementizer “position”.2 Then the fact that noun complements and sentential subjects are islands is a consequence of 11;3 and our proposed revision and extension of the characteristics of wh-movement provides us with a principled way to unify the island phenomena in universal grammar (UG). Syntactic islands are simply the consequence of the application of wh-movement, itself a rule that is fully specified in UG.4

We conclude this paper with yet one more demonstration of the beauty and elegance of our results, as if any were needed. Imperative sentences, as is well known, are syntactic islands; cf. 17.

17. *Which constituents speak in?

It is a general characteristic of imperative sentences that they lack surface subjects.5 Can we not say that such subjects, instead of being simply deleted, are first moved into COMP by wh-movement, and then deleted? Indeed we can, and indeed we must. Notice that deletion of the subject of an imperative sentence is not in fact obligatory, but if it is not deleted, it is spoken with an intonation contour that indicates that it is separate from the rest of the sentence, exactly as if it had been moved into COMP; cf. 18.

18.  You guys speak in adverbs.

Only in negative imperative sentences is a non-deleted subject not moved into COMP by wh-movement, but in that case note that the resulting sentence is not an island, exactly as predicted by 11.

19.  Don’t you be nice to speakers of dead languages.

20.  Why don’t you be nice to speakers of dead languages?

Hence even the analysis of imperative sentences falls under the domain of wh-movement. Are there any worlds left for it to conquer?


1I assume there to be no COMP-to-COMP movement of wh-phrases; cf. fn. 2.

2In other words, island domains are analyzed as instances of S-bar, and non-island domains as S. If S-bar, and not S, is a cyclic node, then no subjacency arises in cases like (i), even though COMP-to-COMP movement is ruled out in principle.

i. Who do you think Tom believes speaks only in adverbs?

For discussion of the important and far-reaching ramifications of this analysis, see Nuclear (forthcoming).

3This observation generalizes to other islands as well, for example complements of verbs like snap, and clauses that function as predicates.

4All that a language learner has to learn about wh-movement is whether it applies in the grammar of the language he is learning. To determine whether it applies, all he has to determine is whether his language has syntactic islands.

5On negative imperative exceptions to this pattern, see below.


Chomsky, Noam (1977) “On Wh-Movement,” in Formal Syntax, ed. by P. Culicover, T. Wasow and A. Akmajian, Academic Press, New York.

Nuclear, D. Terence (forthcoming) “What Gear to Use When Cycling through the Grammar,” in Occasional Papers of the I.A.S.S.

Linguistic Contributions To The Formal Theory Of Big-Game Hunting—R. Mathiesen
Metalinguistic Autoreference—Benoît de Cornulier
Lingua Pranca Contents