A few days ago, we had a conference on multidomination and I firstly saw an andP there. The solution was to simply abandon the conditions of well-formed phrase markers below:

  • Single Mother Condition (if I remember correctly)
  • Nontangling Condition

However, this is a brand new hypothesis in syntax, an andP merged two sentences having a structure as below (I made it):

enter image description here

So, I wanted to ask, is there any solution or description to structures as above without violating Nontangling Condition, because this tree structure seems rather odd to me?

A Brief Explanation Why I've Found This Structure Odd

I know trees are just a way of description of syntactic structure in the cognition, however I do not know the limits of this approach. Like [created] just chooses DP dominated by VP2 out of blue, or are there boundaries?

That's why I wanted to know if there are any analyzing of structures like those without abandoning Nontangling Condition.

Limitations (New)

The description I need should:

  • not abandon nontangling condition
  • use Principles and Parameters as a theory
  • The problem is, the sentence is badly formed if to copy sth. in something is. You cant just stuff that into the DP. Otherwise the sentence should stop after "file". Of course "to copy" is not obligatory bitransitive, but I read a few comments here and there alleging that *to eat for example were inherently transitive, but often eliding the direct object. I'm not sure of the rationale or the ramifications, I just remember it as e.g. Ger Essen essen, "to eat food" being highly redundant and oh so obvious, and maybe related the problem that is I speak English (verb, adverb, participle?)
    – vectory
    Commented Aug 11, 2019 at 22:46
  • The obvious answer, of course, would be to just say that there are two types of verb phrases: traditional V+O phrases, and S+V phrases. Then we could model this one the way I, at least, think about it: [[[Alan created] [and [Max copied]]] [the DAT file in the disk.]], ~like this mshang.ca/syntree/… with proper terminology. I'm sure there's a good reason why VP = V+O, but maybe things like this give us a reason to be more flexible.
    – Mr. Nichan
    Commented Mar 31, 2020 at 4:47
  • Yes, I realize that S+V fails at least the main constituency tests in English.
    – Mr. Nichan
    Commented Mar 31, 2020 at 5:00

5 Answers 5


I will go on to mention Gazdar's theory of RNR constructions in Unbounded Dependencies and Coordinate Structure, since this is the only account I know of that is in some sense a solution. Gazdar shows how to construct a phrase structure grammar which generates the RNR constructions with rules that will be required in any case for other constructions.

For your example, here is a simplified account of how it works. A finite number of rules are added to describe structures that are missing a constituent. Gazdar uses a slash notation inherited from Categorial Grammar: a sentence missing a noun phrase, for instance, is an S/NP, and a PP missing a NP is a PP/NP. These new categories behave like normal categories in coordinate structures -- you can coordinate two like categories to create a coordinate node of that same category.

For your example, we'd do something like the following. Supposing that for the full versions of "Alex created the DAT file on the disk" and the other sentence, we have these PSRs:

S -> NP VP
VP -> V NP
NP -> Alex, Max
NP -> the DAT file on the disk
V -> created, copied

and for coordination of the full sentences:

S -> S and S

then append rules

S -> S/NP NP
S/NP -> S/NP and S/NP
VP/NP -> V

There is no appeal to transformations or mysterious things going on in "PF".

  • Well, I was actually looking for description on Principles and Parameters of Chomsky. Yes, this model of Gazdar does the job quite well and there is no need for transformation or PF deletion or any other magical stuff, however I want to know how conjunction structure is handled in human mind, not to solve a computational problem. | Well, whatever, this solves the problem.
    – Eray Erdin
    Commented Mar 6, 2016 at 11:58

Your example is an RNR construction (short for "right node raising", which was Paul Postal's term). There is a good Wikipedia article on it, and a good discussion in The Syntactic Phenomena of English by James McCawley, who argues for a structure similar to yours, in that it requires unusual tree structures.

Is there another "solution"? That really depends on what it is supposed to "solve". You'd have to be more explicit. I have a theory about RNR that depends on an unusual definition of "category", but I haven't found a good way to describe RNR'd verb phrases, as in "John was, but Mary wasn't, eating caviar."

  • I was assuming more like a PF deletion or empty categories to do the conjunction, yet it seems right node raising does it well. What I'm really asking is, andP (or any other conjunction projection) is not interpretable without Nontangling Condition? (By the way, I'll edit the question right away.)
    – Eray Erdin
    Commented Mar 3, 2016 at 7:42

Well, I was just looking for information about this and I could't found any example of a tree diagrama that explain coordinated sentences. I thought something similar (the same) to what shows the OP.

SC: o1 + C' C':C + o2

So, I think this is possible because the words that regulate case take the same position than C. These words are the verbs and the preposition, which are -N. By far I havent read any classification that takes care of conjuntions, but I'm sure that conjuntions must share that characteristic: to be -N. Then, each o (o1 and o2) could be explained independently.


so I was basically looking for a similar structure to the sentence " John hacked and smashed the box". Yet to learn advanced syntax but can this be an option?enter image description here

  • Welcome to Linguistics.SE. It is not clear what did you mean by showing a diagram for another sentence. Consider expanding your answer with some details and explanation. An image alone does not qualify for an answer. Commented May 18, 2017 at 22:11
  • Your diagram has the wrong categories for "hack" and "smash". They are not verbs, but rather verb phrases missing a transitive verb. Maybe this modified example will make that clear: "... will hack into pieces and smash into tiny bits, the box."
    – Greg Lee
    Commented May 19, 2017 at 1:25
                VP_aux: 0
          /                   \
         Created               Copied
       /                     /
 Alice                  Bob
       \               /
                \  __________________  The file
  • How bad is this? Is this what's called a JOIN in SQL?
    – vectory
    Commented Aug 11, 2019 at 23:02

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.