I'm a student doing some basic research into LFG and I've read some of Falk 2001 Lexical Functional-Grammar: an Introduction to Parallel Constraint-Based Syntax and there's an exercise in the chapter on c-structure that's been giving me a headache.

Unlike GB, LFG has a nonprojective (non X'-category) S. Consider a sentence with no auxillary, such as Joan wrote a book. Should it be analyzed as an IP (as in (i)) or an S (as in (ii))


tree with VP under I' node


tree with VP under S node and no IP at all

Falk also states a principle of "Economy of Expression" earlier in the chapter:

All syntactic phrase structure nodes are optional and are not used unless required to license elements required to create a well-formed f-structure or to add semantic content

So now my confusion: On one hand, the I' node does not help us create f-structure or (to my mind) add semantic content. On the other hand, it seems fishy that we should need to fall back to a S node. I feel like there's some loss of information?

Can someone help me understand what's going on?

I think my intuition about what it means to "license" f-structure is off, so something even about that might help.

  • 1
    Are the trees exactly the ones that Falk produces? Most modern theories of syntax do not produce structures like that. The labeling is funny. It's a wierd mixture of endocentric and exocentric constituents. Commented Dec 3, 2014 at 5:35
  • The trees are exactly the ones in the book, I'm afraid.
    – jgriego
    Commented Dec 3, 2014 at 19:14
  • I've ordered the book. I'm now very curious to see what sort of approach to syntax is being advocated there. My exposure to LFG is limited. But if its advocates are producing tree structures like the ones you give here, I think I am going to become even more critical of LFG than I already am. Commented Dec 4, 2014 at 11:42
  • 2
    The nonprojective S is meant to be used for nonconfigurational languages. There is a chapter about it in Joan Bresnan's Lexical-Functional Syntax. (i) is the right answer here (within the LFG framework).
    – Atamiri
    Commented Jan 14, 2015 at 16:50
  • @Atamiri, (i) is a bizaar tree. It shows I' as head over VP, but there is no I. I guess I was falsely under the impression that LFG was mainly using X-bar c-structures. Note further that a DG approach doesn't need to introduce a spurious S category. It positions the verb as root, and the NPs are its dependents, the resulting structure being relatively flat: perfect for the so-called "non-configurational languages" that LFG prides itself in analyzing. Commented Feb 18, 2015 at 1:32

1 Answer 1


The principle of Economy of Expression applies to the general principle of how c-structures are built. In your example the I' node comes from I' -> (I) VP, i.e., this rule is required to create a well-formed f-structure of a sentence that may contain an auxiliary (as in Joan has written...). The same rules are applied to analyze both sentences and the f-structures are identical except for the value of the TENSE attribute.

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.