Chomsky (1995: 69) says (115) that "(115b) is ruled out for independent reasons of control theory." What reasons?

(115) a. the book that you filed [without PRO reading e]

b. *the book that was filed [without PRO reading e]

The difference between (115a) and (115b) lies on the appearance of overt subject NP in the embedded clause. It seems that the reason might be that only overt NPs could be controllers of licensing parasitic gaps here.

"According to the context given, the argument provided by Chomsky is merely that (115) does not suffice to show that A'-chains (like that in (115a)) license parasitic gaps while A-chains (like that in (115b)) don't—because the judgment contrast in (115) can (and should) be explained by something else (i.e., control theory). "

Is it correct?

  • 2
    115a is definitely grammatically valid, even if it’s quite a strange sentence semantically that requires a very specific context to make much sense. You can use a file to scrape matter off a book, which I guess counts as ‘filing it’ (who has ever done that?); and though we usually only file documents by archiving them in filing cabinets, I suppose you could archive whole books that way too (who has ever done that?). But quite apart from Chomsky’s reasoning, I don’t see why 115b should be ruled out at all. To the extent that 115a is valid, so is 115b. Commented Nov 18, 2023 at 16:05
  • 2
    You need to unpack the sentence for readers here. They won't understand what e is or that the bit in brackets is meant to be part of the sentence. Commented Nov 19, 2023 at 12:14
  • 1
    @Araucaria Oh I see – I thought “[without PRO reading e]” was meant as a reference to a ‘PRO’ reading of something designated e in the surrounding text. The weirdness of filing books muddies the waters, but if we substitute a more reasonable verb like return, I would say that, “The book that you returned without reading” is definitely possible, and “The book that was returned without reading” is less felicitous, but still possible. “The book that was returned without being [or having been] read” is better. You definitely find examples on the Internet of cases like, “The [something] → Commented Nov 19, 2023 at 14:11
  • 1
    → was returned without opening” to refer to people who’ve sent purchased items back to the seller without opening the box, which seems to me to be entirely parallel. It’s not infelicitous enough that I would notice or react to it in regular speech. Commented Nov 19, 2023 at 14:12
  • 1
    @JanusBahsJacquet I agree with you mostly, it's marginal for me too. Anyhow, you might be interested in the answer that I just posted. (haven't proofread it yet, might be full of typos but going to take a break before I check it.) Commented Nov 19, 2023 at 16:37

1 Answer 1


Parasitic Gaps

  1. The book(i) that you filed __(i) without reading __(i)
  2. The book that [you filed the book [without reading the book]]

In the example noun phrase, modelled in (1) and (2) above, we see that there is a gap in the relative clause which represents the object of the verb filed. We interpret this gap as referring to the book in question, in other words as being coreferential with the noun book. The relative clause has its own adjunct (read 'adverbial' or 'modifier'), which itself contains a clause with an unexpressed object. It has a gap representing the object of the verb reading. Our interpretation of this gap is reliant upon our interpretation of the gap higher up in the tree. In other words we understand the gap after the verb reading to be coreferential with the gap functioning as the object of the verb filed. Hence it is a 'parasitic' gap (strictly speaking, parasitic because this gap can only exist on account of the gap higher up).

Now if we passivise the relative clause, the following adjunct no longer seems to work and the larger noun phrase is ungrammatical:

  1. *The book that was filed [without reading] ungrammatical

Now it could be assumed that the reason that (3) above is ungrammatical has to do with the way that parasitic gaps are licensed, and one might invoke fancy theories of A-chains and A'-chains to do so. However, there is, Chomsky implies, a much simpler explanation for the ungrammaticality of (3).


We might have thought that the noun phrase in (1) was already full of gaps, but in fact we missed one out:

  1. The book that you(j) filed [without __(j) reading the book].

The subject of the clause inside the without-PP is missing. And our interpretation of this gap is obligatorily controlled. It is interpreted as being coreferential with the subject of its matrix clause. In other words, the subject of reading is coreferential with the subject of filed; it is understood to refer to the listener, you.

However, if we passivise the clause, the subject of filed is now understood to be the aforementioned book, and thus so is the subject of reading, and as this is almost definitely not what the speaker intended, the example breaks down:

  1. *The book that __(i) was filed [without __(i) reading].

A problem with Chomsky's argument

However, notice that I said that the obligatory control of the interpretation of the subject of reading caused semantic problems. I disagree with Chomsky that the ungrammaticality of (3) can be properly explained using theories of control. Even if it would be very odd for a book to be reading anything, that madcap interpretation should be available from (3). We can see that there is no grammatical reason why a passive in the relative clause here should cause a problem for our interpretation of the without-PP. The following is fine:

  1. The book that was filed without being properly catalogued
  2. The book that [the book was filed [without the book being properly catalogued]

The problem seems to be entirely with the fact that reading in the subordinate clause does indeed require an object, and there is no way of our successfully interpreting it in (5).

To substantiate this, let's change the noun involved:

  1. The professor that was promoted without having read a single novel

That NP seems fine despite the passive in the relative clause. However, if we remove the object of the verb read it is seriously deprecated, if not ungrammatical:

  1. *The professor that was promoted without having read.

It seems to me—not that that counts for much—that a theory of control is not, in fact, enough to explain away the ungrammaticality of Chomsky's (115b).

  • 1
    You have missed the point of [without Pro reading e]. Please refer to the text in my comment under the question. "without reading e" is not part of the utterance...the stuff in brackets comes from the Chomsky text, somehow.
    – Lambie
    Commented Nov 19, 2023 at 17:34
  • 1
    @Lambie PRO is the phonetically null pronoun which is subject to control. e is the parasitic gap. That’s exactly the point. The bit in brackets is part of the example!!! Commented Nov 19, 2023 at 17:37
  • PRO (called "big PRO", distinct from pro, "small pro" or "little pro") is a pronominal determiner phrase (DP) without phonological content. en.wikipedia.org/wiki/….
    – Lambie
    Commented Nov 19, 2023 at 18:23
  • 2
    @Lambie That is exactly why Araucaria is correct that it is part of the example. With all gaps, including PRO gaps, indexed (to i and x) are: The bookᵢ that youₓ filed [_ᵢ] without [PROₓ] reading [_ᵢ] and The bookᵢ that [_ᵢ] was filed without [?PROₓ] reading [_ᵢ]. Both contain gaps corresponding to index i (= the book) and to index x, but only the first example has an antecedent that PRO can refer to; this is why the second example is predicted to be ungrammatical. The brackets are used to group units in the sentence here, not add editorial notes. Commented Nov 19, 2023 at 19:00
  • 1
    Thanks very much for your comments. The answer that @Araucaria-him has provided is very clear to me.
    – Yili Xia
    Commented Nov 20, 2023 at 1:06

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.