I'm only beginning to review phrase structure rules, so let's take a very basic example:

"A sentence consists of a noun phrase + a verb phrase." S --> NP + VP

Now the NP can consist of "NP + NP," right? Or NP --> NP + NP + NP and so on ad infinitum.

How is this noted in the most-used phrase structure grammar notation?

  • Not "+". And. This is natural language; there are words and things don't go on ad infinitum, really. As for how it's handled -- are you familiar with regular expressions?
    – jlawler
    Jul 3 '14 at 4:30
  • I'm afraid not. Could you describe these? Jul 3 '14 at 5:19
  • So cold I write NP --> (AP) N ((and N)+) (PP) ? or NP --> (AP) N (and N)* (PP) ? Jul 3 '14 at 17:43
  • It depends on what programming language -- or pseudocode -- you're using, and what you want to accomplish with the code. There are no PS grammars of natural languages, but it's easy enough to combine short strings into longer strings that resemble natural language. As the Chomskybot demonstrates.
    – jlawler
    Jul 3 '14 at 17:54

I am not directly aware of a source that produces phrase structure rules for coordinate structures, but I can imagine a notation like the following:

 NP --> NP, ... and NP

 VP --> VP, ... and VP

In these rule, the three dots "..." represent an unspecified number of coordinate NPs or VPs, zero to very many. Note that the "and" is positioned immediately in front of the final NP or VP in the coordinate structure.

But Thomas Gross' answer and the comments above point to something more important about the question. If one is trying to capture the essence of coordination in natural language using traditional phrase structure rules, the attempt is going to fail miserably. Phrase structure rules as the are commonly understood do not begin to provide the theoretical apparatus necessary to shed light on the nature of coordination.

Perhaps above all, phrase structure rules alone cannot handle non-constituent conjuncts, e.g.

[Did he] or [did she] solve the problem?

Fred called [you yesterday] and [me today]. 

Does [he write] and [she read] the poems? 

The coordinated strings marked by the brackets in these cases do not qualify as any recognizable unit of syntax in most theories, i.e. they are not VPs or NPs or ... To deal with such data, many theories have to introduce additional machinery, for instance they assume ellipsis in terms of conjunction reduction or right node raising).

Another problem for phrase structure rules is associated with conjuncts that are distinct in syntactic category as pointed out by Thomas Gross, e.g.

 Sam is [conservative] and [a closet Republican].

 Larry is [unhappy] and [trying to find a new job]. 

 Jim called [last night] and [during class today]. 

The coordinated strings are distinct in syntactic category in these cases (Adj+NP, Adj+VP, NP+VP), hence an account of such cases in terms of phrase structure rules is quite at a loss; there are hardly any conceivable rules that could capture such coordinate structures. The notion of syntactic function is a more promising way to approach such data, i.e. the coordinated strings must be alike in syntactic function, as pointed out by Thomas Gross.

The greater point is that one is not going to get far with phrase structure rules as they are commonly understood. They are quite incapable of shedding light on the nature of coordination and many other areas of syntax. When it comes to coordination, I have worked on them extensively and would be happy to point to relevant literature that provides another means of analysis (and one that I think is much more promising than phrase structure rules), if anyone is interested.

  • Hi, Tim. I would like to see the relevant literature that you mentioned. Jul 9 '14 at 0:36
  • Hi James. Thanks for your interest. This paper here coli.uni-saarland.de/~tania/CMGD/… illustrates the challenges of coordination and lays out a dependency grammar (DG) account. I have other peer-reviewed publications that develop my DG approach to coordination in various directions. If you'd like to see more, send me an email (tjo3ya@yahoo.com). I'll send you much more. Jul 9 '14 at 7:02

Answer: mismatches in syntactic category are acknowledged.

  • This is useful info, but I don't think it answers my question. Jul 3 '14 at 17:44
  • @JamesGrossmann The point is - and this is mentioned in the link I provided - that conjuncts must have equivalent syntactic function, rather than equivalent syntactic category. Jul 3 '14 at 18:52
  • I'm asking my question to get information about which notation I can use. Jul 4 '14 at 2:27
  • @JamesGrossmann Check out section 3.2 here. Jul 4 '14 at 6:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.