I really liked what she wrote.

According to CaGEL* (Page 1073), what she wrote is not a clause but a noun phrase (NP). The reason I believe is that the head of what she wrote is not the clause she wrote but the noun what. That is, there is a clause in there but it's just that the clause is not the head of the entire thing in bold.

That said, I'd like to know whether the clause (she wrote) is a modifier or a complement of the head noun (what).

CaGEL doesn't seem to clearly say whether she wrote is a modifier or a complement of what, although it does say that the relative word part of what (i.e., prenucleus) combined with she wrote is a modifier of the antecedent part of what (i.e., head), as shown in the following tree diagram (Page 1073):

enter image description here

I don't object to this tree diagram itself, because any relative clause is a modifier (not a complement) of an antecedent, and because in a fused relative construction with what, the relative clause itself comprises the 'prenucleus' part of what as well as the following clause (e.g., she wrote).

That said, I'd like to know whether the following clause (she wrote) is a complement or a modifier of the fused relative word (what) in its entirety, which question is not answered in the above tree diagram or anywhere else in CaGEL.

I suspect that, being a fused relative, what is not just any antecedent but is a fusion of an antecedent and a relative word, and therefore that what can be construed as licensing she wrote.

Moreover, removing she wrote would definitely change the meaning of what:

?I really liked what.

All in all, is she wrote a complement or a modifier of what?


After the edit, my question does not go against CaGEL.

*The Cambridge Grammar of the English Language by Huddleston & Pullum

  • What does the subscript italic i attached to GAP mean? Is that some arcane way of acknowledging that what is the object of wrote?
    – KarlG
    Apr 26, 2018 at 9:08
  • @BillJ Please note that I'm not disputing that relative clauses can only function as modifiers of antecedents, never complements thereof. But here, I'm not asking about the role of a relative clause but that of only part of it.
    – JK2
    Apr 26, 2018 at 11:58
  • @BillJ I'm not sure why you say it isn't a constituent. In the tree diagram, it clearly is a constituent, namely, 'Clause'.
    – JK2
    Apr 26, 2018 at 15:22
  • @BillJ Do you somehow think that my original question is about whether she wrote (without GAP) is a modifier or a complement of what? If so, what makes you think that? Because I of course was asking about the role of she wrote (with GAP). For example, in I really liked this novel she wrote., you can say she wrote is a modifier of this novel without ever mentioning GAP, which is of course included in the modifier even if you don't say that.
    – JK2
    Apr 27, 2018 at 0:53
  • @KarlG It signifies that what and the gap are coindexed, that is, that they have the same referent. May 2, 2018 at 19:44

2 Answers 2


> That said, I'd like to know whether the following clause (she wrote) is a complement or a modifier of the fused relative word (what) in its entirety, which question is not answered in the above tree diagram or anywhere else in CaGEL.


This becomes clearer if we consider "that which Clause", which is a conveniently non-fused equivalent (in formal registers) to the relevant sense of "what Clause". In "that which Clause", the head is "that", and the relative word is "which". So your question is equivalent to asking whether the clause is a complement or modifier of the two-word sequence "that which"; and the answer is that it's not, because the two-word sequence "that which" is not even a syntactic constituent. "That which Clause" is constructed as "that {which Clause}"; likewise, "what Clause" is constructed as "Head {Prenucleus Clause}", with the only oddity being that the head and prenucleus are fused into a single word. (This is an example of a bracketing paradox.)

  • If "that which" is not even a syntactic constituent, how could they be fused into a single word, "what"?
    – JK2
    May 6, 2018 at 10:49
  • Are you saying that: "I liked what she wrote" is equivalent in formal register to: "I like that which she wrote"? I just don't buy that at all.
    – Lambie
    May 6, 2018 at 15:25
  • 2
    @JK2: Like I said, it's a bracketing paradox. But you shouldn't be surprised, because your tree diagram clearly shows that what isn't a constituent: it fuses a role inside the relative clause with a role outside it. (By the way, to clarify -- "what" is not specifically a fusion of the exact sequence "that which"; rather, it's a fusion of the same roles that "that" and "which" play, and is sometimes synonymous with "that which".)
    – ruakh
    May 6, 2018 at 18:05
  • 1
    @Lambie: The question is about grammar. I only brought up that which to show how the grammar of how what fuses two functions that are not a single constituent. (Did you really not understand that, or are you just trolling me?)
    – ruakh
    May 6, 2018 at 23:23
  • 2
    @JK2: I'm sorry, I thought you were specifically asking about CGEL's analysis; after all, you wrote "I don't object to this tree diagram", and you clarified that you're asking this question because it "is not answered in the above tree diagram or anywhere else in [CGEL]". If you're open to other analyses, then I think both "complement" and "modifier" are plausible (as are analyses under which those terms aren't mutually exclusive).
    – ruakh
    May 7, 2018 at 4:06

The term modifier is normally reserved for non-head constituents of a Noun Phrase, while complement is normally used for a mandatory non-NP constituent of a clause licensed by the particular verb or verb sense.

So if you prefer, it's a modifier. But I don't think the distinction is of much importance. If you are working on a Linguistics paper, just see what classification works best with your overall analysis and then try to see where the weight of evidence lies.

If you have access to Dixon's Basic Linguistic Theory, sec. 17.5.3 has a good, brief discussion on fused relative clauses and some references for further reading.

  • Huh? Are you saying that any dependent in an NP is automatically a modifier?
    – Hannah
    Jan 17, 2019 at 8:03
  • @Hannah Yes, automatically! That's just customary terminology in syntactic analysis. Everything's either a head or a modifier (though you can subdivide this into modifiers proper and complements).
    – user31341
    Jan 18, 2019 at 5:20
  • Ah, right – then I see what you mean :) Though I certainly wouldn't say that it's "customary terminology in syntactic analysis" to call complements modifiers :) In some frameworks, perhaps, but certainly not in general.
    – Hannah
    Jan 18, 2019 at 10:21

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.