This question may be a bit basic - I'm reading Rowlett's "The Syntax of French" (great text), without much background in syntax.

He states,

This approach to theta-role assignment also means that thematic dependants only ever merge as specifiers; they can't be complements. Thus, direct objects and dependent clauses are merged as specifiers rather than complements. The complement position is restricted to a grammatical (rather than a thematic) dependency.

With respect to his last sentence, I am looking for some examples which illustrate the difference. Thanks.

  • I have that book. What page is the passage from? Jul 8, 2014 at 23:14
  • Page 11, just above the last paragraph. For those who don't have the book I can also give context.
    – user2194
    Jul 8, 2014 at 23:21
  • It's only one way of using the terms; there are others. It also assumes that all sentences have a "complement position", whether they actually do or not. Not to mention "specifier" and other positions that may be assumed.
    – jlawler
    Jul 9, 2014 at 0:57

2 Answers 2


Within the book itself, you have a perfect example on page 126 (maybe your 125 because the paragraph you quote is on my page 12).

Jean, il m'aime bien.

The NP Jean has no thematic relation to aime: it obviously does not fit anywhere in the thematic hierarchy outlined by Rowlett (and you can omit it and still get a perfectly well-formed sentence). Examples of thematic dependency abound in the sections 2.1.3 and 2.2.2.

Note that within this perhaps seemingly innocuous paragraph is hidden (in plain sight) a very strong and very controversial theory of universal rigid mapping from syntax to semantics.

  • Like left dislocation in English: Counterexamples, I can't stand'em.
    – jlawler
    Jul 9, 2014 at 15:40

I've now read the relevant section. I indeed find the quote on my page 11. The quote appears in Section 1.5, which presents the "Theoretical Framework" employed in the book. This framework is a highly idealized version of syntactic structure that builds on some very specific presuppositions about the nature of syntax.

John Lawler's comment is on the money in my view. One has to swallow a lot to go with this sort of understanding of syntax. In particular, the following presuppositions are present:

  1. All syntactic structure is binary branching.

  2. All syntactic structure is right branching.

  3. Numerous empty elements are present in the structure, i.e. theta heads.

  4. All arguments merge as specifiers, whereby the predicates to which they belong appear originally to their right as complements of the empty theta heads.

  5. Leftward movement (actually copying) occurs on a massive scale in order to achieve surface word order. For instance, verbal predicates must move leftward in order to precede their object arguments, e.g. one wants eat meat instead of meat eat, one wants believe that it works instead of that it works believe, one wants say it instead of it say. etc.

In my view, this is junk science. It sounds impressive, but it's nonsense. In order to buy into this view of syntactic structure, one has to take the leap of syntactic faith that I associate with much of Chomskyan syntax. I even think many Chomskyans would reject the understanding of syntax that is put forward in the book. The approach is, for example, contrary to basic notions of X-bar theory, in which complements (as well as many specifiers) are arguments. On the approach in the book, all complements are actually predicates.

But to answer the actual question, my understanding of the final sentence in the passage is that the head-complement dependency associated with standard X-bar-type structures is actually present, but its presence is manifest as the dependency between the theta head, which is always an empty element!, and the predicate, which is represented by X in the complement position. The empty theta head θ takes a predicate X as its complement; a dependency exists between the two. But since the dependency is unlike the dependencies that are traditionally assumed, Rowlett calls it a "grammatical dependency", a nebulous term.

Hence Rowlett appears to be stipulating two basic types of dependencies:

  1. Grammatical dependencies between empty theta heads θ and their predicate Xs (head-complement dependencies) and

  2. Thematic dependencies between empty theta heads θ and their specificiers (specifier-head dependencies), these specifiers being what one traditionally understands to be the arguments of the predicate Xs.

Rowlett is thus pursuing an approach to syntax that takes specifiers to be the arguments of complements in a sense, as opposed to the arguments of head predicates, which is much more standard. I think such an approach is going to be difficult for many a syntactician to swallow.

  • Oh, don't be too sure. Swallowing camels and straining at gnats is a required course these days.
    – jlawler
    Jul 9, 2014 at 15:42
  • Thanks for the information, and for both of your responses. I would love to ask more about the points you brought up about the theory but I am a bit strapped for time now, so maybe in a future question.
    – user2194
    Jul 9, 2014 at 17:09
  • Though I do wonder, despite my lack of background, whether the kind of theory we see here that involves so many convert elements and moves is part of a movement in syntactic theory away from representing the 'surface structure' and toward deeper approaches that aren't quite as salient empirically or formally but which comprise a theoretical apparatus which can generalize more neatly over a larger variety of phenomena. Did that sentence make sense?
    – user2194
    Jul 9, 2014 at 17:16
  • 1
    @JohnDoe, My message to aspiring linguists is to question the claims. Demand emperical evidence. Demand evidence that all syntactic structure is binary branching. Demand evidence that all branching is to the right. Demand evidence for the existence of the numerous covert elements. Demand evidence for the countless movements. I can produce evidence that much of syntactic structure is n-ary branching and that left-branching is common, and I know of no empiricism that supports the existence of all those covert elements. Jul 9, 2014 at 21:08
  • 1
    @JohnDoe, and be wary of institutional authority (i.e. what sounds like impressive science citing numerous sources). Just because some people at MIT or Harvard are delivering such claims does not mean that those claims are backed up by empericism. The junk science perpetuates itself because there are too many sheeple in the linguistics community. Jul 9, 2014 at 21:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.