On page 1037 of the Cambridge Grammar:

enter image description here

In [2iii] and [2iv] above is shown a gap "in object position".

In fact, this is where all the grammars that I know of would put the gap. If the gap corresponds to an object of a verb -- as here -- or of a preposition, all the grammars that I know of put it after the verb or the preposition.

But I wonder why it has to be.

If you look at [2ii], which is used instead of the gap in [2iii] or [2iv], and that which is also acting as an object there. But it isn't placed after gave me but is placed at the start of the relative clause.

Since the gap you can't see anyway, I wonder why the gap cannot be placed right before my, as follows:

I accepted the advice [that ____ my neighbour gave me].

I accepted the advice [____ my neighbour gave me].

This way, I think the structure of the relative clause might be more coherently explained and understood.

Also, does anyone know of a grammar that places the gap like this?

  • I think the point of putting in the gap is to show the original position of the NP that is missing in the relative clause. Since it's not there any more, it has to be imagined. It's not a great device, and it's not part of a formal theory, so don't worry about it. – John Lawler Dec 21 '18 at 3:19
  • @JohnLawler So there's no reason not to put it at the start of the relative clause, then. – JK2 Dec 21 '18 at 4:50
  • 1
    There is a gap after "me". "I accepted the advice which my neighbour gave me ___ ", (see P1038 [4]). Btw, I take it you realise that "that" is not a relative pronoun, but a subordinator. – BillJ Dec 21 '18 at 9:30
  • @BillJ Of course, it's a suboridnator! I also know The Cambridge Grammar sometimes puts a gap even in a wh-relative clause --as in your comment-- and sometimes doesn't, as shown in the OP. But when they do, I think the gap merely shows the original place of the relativized element, which isn't necessarily the case with the gap in a that-relative or a bare relative. – JK2 Dec 21 '18 at 9:59
  • 1
    Gap appears in wh relatives where the relativised element is not subject, irrespective of whether it's a wh relative or otherwise. The full analysis of [2ii] is discussed on p1038. I don't see what your problem is. – BillJ Dec 21 '18 at 10:08

The location of the gap is not negotiable. It is placed where the object would be placed in a 'canonical' (i.e. non-relativized in this case) version of the clause.

A few lines below the section you quoted (pp. 1037-1038), CGEL says that, formally speaking, [2ii] also has a gap in the same place.

Formally, there is a gap after gave me in this construction as well as in the non-wh relatives. The difference is that while in [2iii-iv] the gap is related directly to the antecedent advice, in [2ii] it is related indirectly, via which. Example [2ii] can thus be represented as in [3] …

[3] I accepted the advicei [whichi my neighbour gave me ____i].

The principle in question generalizes to cases in which the missing constituent is something other than an object. The gap always occurs where that missing constituent would be in a (more) canonical clause; see Sec. 7.2 'Gaps and antecedents' on p. 1082 of CGEL.

  • Then, how would you place the gap in these examples? (1) He was wearing a tall black sheepskin hat [from the top of which dangled a little red bag ornamented by a chain of worsted lace and tassels]. (2) I may be late, [in which case I suggest you start without me]. I mean, does it really matter what a canonical version is? – JK2 Jan 20 at 16:28
  • @JK2 Your first sentence serves as a good illustration of why the canonical form matters. Here is a simpler version of that sentence that will do just as well: It was a hat [from the top of which dangled a bag]. The non-relativized version is From the top of the hat dangled a bag. This is still not canonical, but a subject-dependent inversion of a canonical clause. What is canonical is A bag dangled from the top of the hat. Note that CGEL doesn't attempt to diagram such inversions (which are discussed in Ch. 16, 'Information packing'); that's for more advanced texts, I'd imagine. – linguisticturn Jan 20 at 17:42
  • @JK2 For now, let's leave the inversion as is. In that case, there would be no gap in this sentence. From the top of which is already in the correct place, and which could be simply replaced by its antecedent, (the) hat. Now you may begin to see why referring to the canonical form is helpful: if the inverted version is allowed to be the referent one, then we seem to have a contradiction of the principle that integrated relative constructions are characterized by an anaphoric gap. But if we insist on canonical versions, there is no problem. – linguisticturn Jan 20 at 17:42
  • @JK2 Here is your sentence without the inversion: It was a hat [from the top of which a bag dangled]. In this sentence, there is indeed a gap after dangled. So the full analysis of your sentence involves two phenomena (some might say two 'processes'): relativization and inversion. It's perhaps too much to expect that simple in-sentence notation would be able to describe both at the same time. You'd need more sophisticated diagramming to do that, I would think. – linguisticturn Jan 20 at 17:42
  • @JK2 Your second sentence is an example of supplementation, where things are a lot 'looser'. In particular, instead of syntactic compatibility of antecedents, all you need is semantic compatibility of anchors (p. 1351). Moreover, CGEL says that the whole in which case is a relative phrase (p. 756). So the non-relativized version would be something like if I'm late I suggest you start without me. It is not that surprising that anaphoric gaps do not really feature here: gaps are syntactical; they really require antecedents. But in supplements all you have is semantic anchors. – linguisticturn Jan 20 at 17:43

Unfortunately, I'm not a syntactician, so I'm not really familiar with the nuances of these theories and I'm not really sure what "gaps" in relative clauses are supposed to be, in a technical sense.

But BillJ's comments and linguisticturn's answer indicate that according to CGEL, it wouldn't be technically correct to say that "[in] [2ii], which is used instead of the gap in [2iii] or [2iv]". For some reason, it seems [2] presents a simplified analysis of the grammar: [2ii] actually does have a gap (after me) as well as the relative pronoun which.

I have found another source that indicates that there would be a "gap" in the wh-relative clause just as much as in the other kinds of relative clauses:

In terms of terminology, some other sources say that in wh-relatives, the relative pronoun "moves" from the original place, leaving behind a "trace". I don't know if that's different in any significant way from a "gap" resulting from a "deleted" constituent. I have the impression that the concept of "movement" has been debated a lot (Wikipedia describes it as "controversial").

Chapter 11 "Wh- movement in English" of The syntax of natural language: An online introduction using the Trees program (Santorini, Beatrice, and Anthony Kroch. 2007-) seems to say that all English relative clauses start with a relative pronoun slot followed by a relativizer slot; the relativizer is covert in wh-relatives and the relative pronoun is covert in that-relatives. There is additionally a trace later on in the clause for both kinds of relative clauses.

enter image description here

enter image description here

  • Gap is the standard device for marking a missing constituent. Also, "that" is not a relative pronoun, but a clause subordinator, so it cannot function as a subject or object in a relative clause. Instead, these functions are marked by gap. – BillJ Dec 21 '18 at 8:36
  • @BillJ: What do "device" and "missing" mean exactly? I don't think I called "that" a relative pronoun anywhere in this answer; I said "the relativizer is covert in wh-relatives and the relative pronoun is covert in that-relatives", which seems to be the situation that is shown in the diagrams given by Santorini and Kroch. – sumelic Dec 21 '18 at 8:53
  • I don't see the relevance of Santorini and Kroch' s approach to the better (and more logical) one taken by H&P in their CGEL. I only mentioned the category of "that" because it is, I think, the core of the OP's problem. You simply must look at CGEL to see what we're talking about here. (see my last comment to the OP for the answer) – BillJ Dec 21 '18 at 9:00
  • @BillJ: I think you should write an answer. I don't own CGEL so I can't check it right now. If, as you say in your comment below the question, "The full version of [2ii] is "I accepted the advice which my neighbour gave me ___ "", then why isn't the "__" included in the quotation? Is it misquoted? – sumelic Dec 21 '18 at 9:06
  • 1
    No: it's because the book takes things in stages. Initially (as seen in in [2ii]) it gives a simple explanation about relativisation and the contrast between that relatives and wh relatives. As the grammar progresses, it goes into more detail about gaps, and we find an explanation with examples showing gap in wh relatives. I would post an answer but I think the answer to the OP's question is a simple one, i.e. that they need to look further into the chapter on r/c's. Do please try to get a copy of CGEL - does your local library have it? – BillJ Dec 21 '18 at 9:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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