I am reading about binding theory and I have a little problem. The book (1) says that:

A binds B iff:

(i)  A c-commands B;
(ii) A and B are co-indexed.

and c-commanding is defined as the following:

Node A c-commands node B if and only if:

(i)   A does not dominate B,
(ii)  B does not dominate A, and
(iii) The first branching node that dominates A also dominates B.

So far so good. But then there are a few examples. One of them is this:

Poirot's brother invited himself. 

Here, according to the author, "brother" is c-commanding "himself" and they are obviously co-indexed.

Tree diagram of the sentence

But as far as the definition goes, the first branching node for "brother" here is NP and it clearly doesn't dominate "himself". So how can the c-commanding be valid here?

(1) Haegeman, 1994. Introduction to Government and Binding Theory, 2nd edition, page 211

  • 1
    The first branching node that dominates your NP [Poirot's brother] is IP, not NP. A and B stand for XP, not X.
    – Alex B.
    Commented Dec 15, 2013 at 5:09

1 Answer 1


The binding theory as presented in Haegeman's book has long since been abandoned by most people who study binding phenomena in a serious way. That book presents the traditional binding theory according to Chomsky (1981, 1986). While it may be good to study and learn that stuff to gain the big picture of the development of syntactic theory, be aware that the account of binding you are learning there is just a reference point nowadays. It's validity is defended by almost no one anymore.

I don't have Haegeman's book here with me, so I cannot verify the example in her book. Taking what you give at face value, I can see why it is confusing. But as Alex B. points out, the way to interpret the example is that the relevant NP is Poirot's brother, not just brother. The first branching node to dominate this entire NP is IP, so the definition sort of works. The confusion stems from the fact that you've got two nouns, Poirot and brother, and it seems like one of these should be the relevant node for the definition, whereas what is really going on is that the two together form the relevant NP.

Compare your example with a similar example like

(1) *[His1 sister] invited himself1.

The definition of c-command and condition A of the theory make the correct prediction in this case, because the reflexive pronoun himself should be bound in its governing category. Since its binder his is embedded inside the NP his sister, it does not c-command out of the NP and therefore cannot bind himself.

When you learn about the DP-hypothesis (NPs are really determiner phrases, not noun phrases), you will begin to see why the traditional binding theory that is presented by Haegeman (and many others) doesn't really work anymore. On a DP analysis, the possessive determiner his heads the subject DP in (1), which means it should be the relevant unit that c-commands out of the subject DP. In other words, if one assumes a DP analysis of noun phrases, the traditional binding theory does not work anymore.

So I think that the greater message is that you should be prepared to experience further confusion and that as you progress to further stages of the development of syntactic theory, you will see that the field has been changing too fast.

  • Or, since Chomsky's theories all get abandoned eventually by people who study language seriously, you could abandon them now and go study a theory that doesn't go to all those silly extremes.
    – jlawler
    Commented Dec 15, 2013 at 19:28
  • Agreed! But unfortunately, we have to know about the silly extremes. If we don't, we isolate ourselves and become irrelevant. Commented Dec 15, 2013 at 20:43
  • 1
    I have to admit, i don't really understand how the DP hypothesis is incompatible with G&B-era binding theory. We just say that DPs (and only DPs) are assigned indices and the binding conditions apply to DPs. You say: "On a DP analysis, the possessive determiner his heads the subject DP in (1), which means it should be the relevant unit that c-commands out of the subject DP" - It doesn't 'c-command out', rather, it's the entire DP that c-commands the bound reflexive. No problem here.
    – P Elliott
    Commented Dec 16, 2013 at 10:22
  • 1
    It is of course a great deal of fun to argue about angels and pinheads. But none of these are actually phenomena; they mark presuppositions made at one point or another in the recursive history of generative grammar. This is precisely why I prefer McCawley -- he avoids Occam's razor by sticking with the data and detailing every assumption as it's made, noting but ignoring arguments based on other assumptions, which saves everybody a lot of work.
    – jlawler
    Commented Dec 16, 2013 at 18:12
  • 1
    @P. Elliot, "man" would be assigned an index like other non-referring expressions are assigned an index, e.g. "men", "no man", "every man". I do not have Haegeman's textbook available. The question above implies, however, that that is what Haegeman is doing. Let me nevertheless concede the point. My concession does not remove the problem, though, because the DP-hypothesis necessitates that one accept constellations in which a DP can bear an index that is distinct from the index of its head word D. That seems problematic to me, mainly because the NP-analysis is not faced with this necessity. Commented Dec 16, 2013 at 23:19

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.