I'm really struggling with drawing a tree representing wh-movement in "How was the plot discovered by the authorities?"

In the main VP, we have "by the authorities" as an adjunct, as well as "how," which moves all way from its adjunct position to the spec of CP.

[Howi was the plot discovered by the authories ti] (only showing the relevant movement).

Is this correct? That is, is "how" rightly an adjunct and/or is there meant to be an intermediate movement in here somewhere? Since there is no embedded clause and there is already a complement ("the plot"), this seems like the only way.

  • 2
    Welcome to Linguistics SE! Your analysis looks right. Commented Nov 25, 2015 at 0:31

1 Answer 1


Your analysis is not correct. It is well known that wh-extraction out of adjuncts is impossible. See Huang 1982. Thankfully, the derivation of your sentence involves neither wh-extraction nor adjunction. Notice that it involves the passive construction The plot was discovered by the authorities, in which case the plot and the authorities are both assigned theta roles by discovered. Ignoring the technical details of passive constructions, the derivation involves building up

[TP the plot was discovered by the authorities]

And in the usual manner, we merge in an interrogative C bearing a wh-feature that needs checking

[C [TP the plot was discovered by the authorities]]

The subject inverts with the auxiliary, and then how gets merged into the tree to yield

[how [[C+was] [TP the plot was discovered by the authorities]]]

  • 1
    But that's not extraction out of, that's whole adjunct moving. Do you suggest that when as in When did it start?" is also base-generated in SpecCP rather than within TP? Also there is such thing as *thematic adjuncts, which extraction is allowed from. I mean, please correct me if I'm wrong. Does Huang base-generate adjuncty wh-words? Commented Nov 25, 2015 at 8:28
  • @IvanKapitonov You are correct that adjunct islands don't account for whole adjuncts moving. But it would seem that derivational economy principles would prevent such movement, so that in When did it start, when is generated in Spec, CP rather than within TP. The reasoning is simple. Assuming the copy theory of movement, adjunction of when and movement to Spec, CP requires the operations Merge+Copy+Merge, which is computationally less efficient than a single instance of Merge.
    – Ink
    Commented Nov 26, 2015 at 1:15
  • 1
    Surely it would be most economic to base-generate everything :) But how does the interpretation proceed? Imagine you had When did she say it started? -- if we had a trace (lower copy or whatever), we'd know where when is to be interpreted. And we'd account for ungrammaticality of When did she say that it started t? (that-trace effect). Commented Nov 26, 2015 at 1:45
  • @IvanKapitonov I think my approach can account for the dual interpretations of When did she say it started. Either when merges into matrix Spec, CP or it merges initially into embedded Spec, CP and moves to the higher one. Interpretations at LF can proceed accordingly. Crucially, notice my approach involves fewer computations, as your approach will produce more than one trace. The ungrammaticality of When_i did she say that it started t_i does not enter into the discussion.
    – Ink
    Commented Nov 26, 2015 at 3:04
  • Hm. I need to sit down and think, maybe I'll get persuaded :-) Commented Nov 26, 2015 at 3:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.