The 1997 paper "English Relative Clause Constructions" by Ivan A. Sag has these diagrams: enter image description here

(53) shows a diagram of to go to the UK, and (54) of for them to go to the UK. In (54), complementizer (C) for is shown to take two complements (COMPS): an NP (them) and a CP (to go to the UK).

In generative grammar, is it possible for a comlementizer to have two complements as above? Or is it something that is allowed only in HPSG?

  • 1
    I would say them goes in the specifier of the infinitival clause, an analysis without two complements.
    – Keelan
    Mar 18 at 8:21
  • @Keelan You mean them as the subject of the infinitival clause? But that's not Sag's approach in the paper, where he specifically says that them to go to the UK isn't a constituent.
    – JK2
    Mar 18 at 10:07
  • I haven't read the paper and don't know HPSG, but indeed, I would differ from Sag on this from the perspective of more traditional generative grammar. The possibility of conjunction (I want for [them to go] and [him to come]) suggests that them to go ... is a constituent.
    – Keelan
    Mar 18 at 12:35
  • @Keelan The paper shows examples (57)-(60) on pages 28-29 to present arguments against NP+CP forming a constituent. Specifically, (57) rebuts your 'possibility of conjunction' argument. That said, how to analyze this construction is not an issue here. How a complementizer can have an NP and a CP as complements is.
    – JK2
    Mar 18 at 13:10
  • Yes, such tests are only indications and you should always use more than one.
    – Keelan
    Mar 18 at 14:04

1 Answer 1


Later generative grammar usually insists on binary branching, so, as Keelan already said in a comment, you get C:for + IP:[them to go to the UK]. Then, you need a rule of exceptional case marking, in which C can assign case to the specifier of its complement phrase ("them" has objective case, which can only depend on "for"). Maybe the selection of a clausal complement plus case government of a constituent different from that is what prompts this analysis with two complements. You might then state a predication relation between the two complements separately, in order to emulate the existence of a full clause.

I don't knwo what exactly HPSG says here. GG would not use this structure, it treats "for" as a complementiser of one single clause. I don't know what sense of "C" is assumed above such that a "C" element could have an NP argument. The above analysis already treats infinitival "to" as a C, and that's also problematic. So I would conclude that "C" already has a different meaning in HPSG

  • The paper says C stands for complementizer, so I don't think C has a different meaning in this paper or HPSG in general. I gather you're not aware of C taking two complements like this. So I'm not the only one who thinks this is strange.
    – JK2
    Mar 18 at 10:12
  • Yes, but I meant that the notion of "complementiser" itself seems to be different...
    – Alazon
    Mar 18 at 14:36
  • That's possible, but since HPSG is a type of phrase structure grammar, I don't think they would use the same terminology differently in a confusing way.
    – JK2
    Mar 18 at 23:35

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.