Consider the following example sentence:

The ball often hit the tree and never the man.

I am trying to represent this sentence as a constituency-based parse tree, but I am having a hard time deciding where to put the adverbs often and never.

For the sentence The ball often hit the tree. it is simple: the ball often hit the tree

Even the coordination of NP can be represented easily: the ball often hit the tree and the man

Where would the constituent never be added to the tree? It is connected to the noun phrase the man, but at the same time adverbs never belong into noun phrases. What structure is generally used to represent this special case where adverbs of a verb phrase belong to a noun phrase?

  • As written, the sentence contains two coordinated noun phrases: "the tree" and "the man" each functioning as object of "hit" and linked by the coordinator "and". You can simply insert the negator' "never" immediately before the word "the", so it has scope over only the second coordinate. This is called 'non-verbal negation.
    – BillJ
    Jun 4, 2016 at 14:52
  • @BillJ I do not understand what you are trying to tell me and how it effects my question.
    – Frithjof
    Jun 4, 2016 at 17:32
  • You asked where would "never" be added to the tree, and about the structure. I answered those questions. I would also add that it would be normal to use "but" as a marker, not "and".
    – BillJ
    Jun 4, 2016 at 19:21
  • @BillJ Oh no, I did not ask that. I stated that adverbs do not belong into a noun phrase, but in a verb phrase. Sorry for that misunderstanding. I agree 'but' would be a better conjunction, but it should be the same syntactic structure with 'and'.
    – Frithjof
    Jun 4, 2016 at 19:41
  • Adverbs can of course 'belong' in noun phrases. In your second example the second coordinate is the NP and the man. Inserting "never" gives an 'expanded' second coordinate in which and/but never the man is an NP. This is clearly a constituent so what other category could it belong to? “Never” is marking the second coordinate, indicating the relation of the bare coordinate "the man" to the coordination. So your example is analysed as a coordination of two NPs. I think this is much better than saying it's a coordination of VPs (or clauses), with ellipsis within the second coordinate.
    – BillJ
    Jun 6, 2016 at 11:09

2 Answers 2


enter image description here

Here is a tree for your example showing the coordination of two NPs with "never" in the second one.

I think this is much better than saying we have a coordination of VPs (or clauses), with ellipsis within the second coordinate.

  • Thank you for that tree, I understand what you are telling me now. I am not used to these "Modifier"-nodes, but they make sense. Probably because I come frome a Latin background, where word order is not that important ..
    – Frithjof
    Jun 14, 2016 at 17:05

It looks like it could be a gapping construction. Gapping deletes repeated constituents from the second part of a coordinate structure, including the verb, and leaves behind two constituents:

[S The ball often hit the tree] and [S the ball never hit the man].
[S The ball often hit the tree] and [S never the man].  

in which case the derived structure would be as I just indicated. My first impression is that "often" and "never" are V' modifiers, but Ross has proposed that gapping always leaves behind one constituent within the main V' and one constituent outside it. That implies that "never" is an S modifier, rather than a V' modifier.

McCawley discusses the derived structure of gapping constructions in The Syntactic Phenomena of English starting on page 280.

  • Thank you for your answer. I believe Gapping is the right phenomenon, but I still have two questions: a) Why do you suggest S-Coordination, and not VP-Coordination? E.g.: The ball [VP often hit the tree] and [VP never _ the man]? b) How would a complete parse tree with Gapping look like in a constituency grammar? I only found parse trees for dependency grammars while searching on the web. Also I believe the link you provided is dead.. ;)
    – Frithjof
    Jun 5, 2016 at 10:20
  • I corrected the link -- thanks. I've never seen gapping proposed for anything other than coordinated Ss, and I don't know whether gapping ever applies to VPs. I also don't recall seeing any psg account of gapping, such as GPSG, and I don't know how one would do this. I did find a relevant reference here: ingentaconnect.com/content/jbp/avt/1991/00000008/00000001/…
    – Greg Lee
    Jun 5, 2016 at 14:13

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.