Talk:X-bar theory

Untitled

Though this is an excellent explanation for x-bar syntax, I believe it does not encompass all of x-bar theory. I think there needs to be a seperate section for x-bar theory, x-bar syntax, and x-bar morphosyntax, as well as the different syntactic theories (i.e. minimalism, LFG, and so on).—Preceding unsigned comment added by Linguizic (talk • contribs) 01:37, 29 January 2004 (UTC)

I must disagree with the example sentence tree. It states that the "He studies linguistics at the university" is a verb phrase. This is incorrect - it is a complete sentence (or inflectional phrase, if you prefer). The noun phrase is not the specifier of the verb phrase, it is the specifier of the entire sentence. There is no specifier for the verb phrase in this case. However, it's always possible that I was simply raised with a different philosophy of X-bar Theory, and that saying a sentence is equivalent to a verb phrase is perfectly fine in other schools of thought. If someone wants to defend the usage here, I'm willing to discuss it. 129.2.211.72 04:54, 7 Oct 2004 (UTC)

I agree with User:129.2.211.72. Though I think that the NP is initially generated in the VP (as it is shown in the picture), but the NP raises to the SpecIP position. So, i think that we would need another graphic to show both the Deep and the Surface structure. --Javier Carro 11:07, 1 Jan 2005 (UTC)

By the way, I just began a stub in Wikibooks in order to write a module about Transformational grammar (wikibooks:User:Javier_Carro/Transformational_Grammar). Everybody is welcome there :) --Javier Carro 11:07, 1 Jan 2005 (UTC)

The tree can't be incorrect, because X-bar theory is just a phrase structure schemata, not a theory of syntax in general. Most theories in transformational grammar either have the subject base-generated in Spec,IP, or base-generated in Spec,VP and then raised to Spec,IP (or the specifier of a functional projection in a split INFL). However, in Head-driven phrase structure grammar and probably some other theories, the subject remains in Spec,VP. I think the diagram is perfectly OK as it is, since we're not trying to describe any particular theory of syntax on this page. Cadr 09:19, 19 Apr 2005 (UTC)

To clarify, I'm not saying that the current diagram is any more or less correct than the alternatives which have been proposed above. I just see no reason to change it given that it's already there. Cadr 09:21, 19 Apr 2005 (UTC)

Oh...I guess all this really needs is a good clarification...I'll try to do that now. -- Beland 06:30, 24 Apr 2005 (UTC)

NP → ?

     NP
    /  \
DetP    N'
 |      |
Det'    N
 |      |
Det    cat
 | 
the

What evidence is postulated to demonstrate that there is a constituent Det' and DetP if you posit DetP as a specifier of NP?

I suggest:

   DP
  /  \
     D'
    /  \
  the   NP
       / \
          N'
         / \
       cat

where the is the head of DP and the NP is the complement of D; the specifier of DP is used for constructions such as

      DP
   /     \
  DP      D'
 / \     / \
    D'  's  NP
   /\      / \
  0  NP       N'
     / \     / \
        N' dad
       / \
     John

Although this has the weakness of an unused NP Spec, this is at least prefered to having two vacuously branching nodes (or five, if you consider the way that this NP is drawn, with (superfluous) category labels branching to their heads).

(See: Abney, Steven (1987) The English Noun Phrase in its Sentential Aspect. Ph.D. dissertation, MIT.)—Preceding unsigned comment added by Ashdurbat (talk • contribs) 17:40, 7 November 2005 (UTC)

I strongly concur. While it is true that it's a moot point as many syntactitions feel X' is wrong, we can at least give the most advanced version (and the version the few who still follow it believe). Clay 07:44, 21 September 2006 (UTC)

Add reference to Jackendoff

I haven't learned to edit wikipedia entries yet, but for this one I think it should definitely be mentioned that X-bar theory was invented by Ray Jackendoff. The principal reference is: Jackendoff, Ray. 1977. X-bar-Syntax: A Study of Phrase Structure: Linguistic Inquiry Monograph 2. Cambridge: MIT Press. Barbara Partee 15:20, 14 March 2007 (UTC)

Taken care of. Thanks for the reference, and feel free to contribute more to Wikipedia as you see fit. –jonsafari 21:29, 15 March 2007 (UTC)

Actually, it was Chomsky's (1970) paper "Remarks on Nominalizations" where X-Bar Theory first started. —The preceding unsigned comment was added by Mike barrie (talk • contribs) 19:30, 17 March 2007 (UTC).

DetP as a specifier of N??

Which framework or theory do you use in the article? I have never seen DetP as a specifier of N.Russky1802 22:00, 25 May 2007 (UTC)

Analysing determiners as specifiers of NP was standard in transformational grammar up to the late 1980s. If the determiner is a specifier, it has to be included in a maximal projection, otherwise X' theory is violated.

Anyway, I've been opposed to changing the NPs in this article to DPs because I think it makes things unnecessarily complex for an introductory article, but the DP hypothesis seems to be so widely assumed these days that maybe it's a bad idea to use NPs. Cadr 22:10, 25 May 2007 (UTC)

Actually, looking at some GB theory textbooks, it seems that back in the day people just put the determiner directly under "Spec", and didn't have a DetP or any equivalent. I'll change the article back to your version. Cadr 00:04, 26 May 2007 (UTC)

The head is the determiner (D) which projects into a determiner phrase (DP or DetP). The word cat is the noun phrase (NP) which acts as the complement of the determiner phrase. More recently, it has been suggested that D is the head of the noun phrase. => This is not clear at all, very confusing. VincentKREBS 21:04, 18 September 2021 (UTC)

NP vs. DP

Well, I use both NP and DP. If there is a null determiner I see no reason why I should make it more complicated by having something like

  DP
  D'
D   NP
    N'
    N

for [John].

On the other hand, if we have NPs with genetives we'd better use the DP hypothesis, [DP John's criticism of the theory]

The question was why should we use DetP for a simple NP theory? I usually have Det as a specifier, though there is no consistency with the X-bar theory. But it is a traditional analysis.

I think we should have a separate section on DP hypothesis (only for complicated cases). In simple ones we may still use the olde NP.Russky1802 01:57, 26 May 2007 (UTC)

In "John's criticism of the theory", "John's" was originally assumed to be in [Spec,NP], so these sentences don't require a separate DP projection. Anyway, I'm not sure whether it would be better to stick to 70s/80s X-bar theory and not use DPs at all, or always use DPs, but I think we should be consistent and do one or the other. Cadr 02:10, 26 May 2007 (UTC)

Ok. But many people have contracted structures, something like

   DP
  / \
  ___
the book

because it is so obvious. Moreover, Chomsky claims that X-bar theory should be eliminated at all. Russky1802 18:21, 26 May 2007 (UTC)

Sorry, I'm not quite sure what you're getting at here. It's true that Chomsky doesn't believe in X' theory these days (although phases serve a similar purpose to the original X' theory), but that doesn't have anything to do with the DP/NP question.

As I said, I think we should used either DPs or NPs, but choose one or the other and be consistent. Cadr 18:37, 26 May 2007 (UTC)

I agree with you. What I meant is something like (a la Chris Collins):

  / \
 /   \
the book

Russky1802 19:57, 26 May 2007 (UTC)

OK, but we can't use that in the article because it's not an X' structure. Cadr 20:06, 26 May 2007 (UTC)

This article is as clear as mud

I read the article on X bar and couldn't make head or tail of it. This is the closest that we find to a definition:

The letter X is used to signify an arbitrary lexical category; when analyzing a specific utterance, specific categories are assigned. Thus, the X may become an N for noun, a V for verb, an A for adjective, or a P for preposition.

Huh?

It's fine for linguists who know what an X-bar is, but for ordinary people, this makes zero sense. Can't some linguist put this article into a form that people who haven't been through the linguistics mill can understand?

In fact, I went back and had a look at the very first incarnation of the article, and it actually made sense! Who said that collaboration among many people makes a better Wikipedia article? Bathrobe (talk) 15:11, 17 January 2008 (UTC)

I was about to post the exact same thing

I agree with Bathrobe -- I came to this discussion page to say the exact same thing. I read the entire article and have no idea at all how X bar works. And it's not because I'm an idiot, I got a 149 on my last IQ test. This article is as clear as mud. I gather that people who are already experts in the field think that it's informative, but we're not. Sometimes you have to click on further links to fill in the missing pieces, but I feel that the only thing that would fill in the missing pieces here is a complete rewrite of the article. Inhahe (talk) 17:30, 2 July 2008 (UTC)

Reply to Inhahe above

On the surface, X bar theory seems obvious from a computer science perspective. Okay, so a bunch of linguists discovered that:

1. there is a commonality between different kinds of phrase structures. Gee, a VP is a lot like an NP, so why don't we merge them into one concept and use the letter X? Now we can use XP to discuss any kind of that sort of thing.

2. N-ary trees can be replaced by binary trees, while retaining the same information. A linear list in which the items are regarded at one level can actually be a linked list of frames which are themselves anonymous, constituting a lob-sided binary tree. The X' correspond to these anonymous nodes in a linked list. They had to be invented to serve as the backbone that glues the "X-bar" trees together.

Obvious so far, right?

But what is important is that there is some mental reality to these binary trees. I.e. the theory doesn't just exist for some structural convenience, but reflects the theoretical belief that there is a corresponding structure in the mind of the speaker who is uttering these phrases in some given human language.

For instance, consider the English noun phrase "law school entrance requirement test administration procedure". The thing described by this phrase is, first and foremost, a kind of procedure. It is not a kind of law. So according to English syntax, this would be NP -> N N' -> procedure N'. Further generation would expand the N' to administration and so on. The abstract syntax would be:

  NP ->    N'
         /    \
procedure      \ 
               N'
              /   \    
  administration  [ etc ]

The word order is backward, but this would be taken to be a deep structure, requiring a transformation to generate the English word order, where the main stem appears last.

The important thing is that this models the hierarchical nature of phrases. The theory asserts that all languages have phrases in which there is a main stem which heads off the phrase, and and then the additional words represent a successive refinement of that stem, making the meaning more and more specific.

In our example, the thing is a procedure. What kind? An administration procedure. An examination administration procedure. An entrance examination adminstration procedure. Et cetera.

What does the "X bar" mean? that just refers to these glue nodes that hold the binary tree together. We used "N bar", but since the theory covers all kinds of phrases, not just noun phrases, X is used, and the characteristic glue nodes are "X bars".

Inside a computer, we would use some unamed memory objects instead of "X bars". For instance in the programming language Lisp we could represent the noun phrase as a list of symbols like this:

 (procedure administration examination entrance school law)

When the machine reads this list, which is just a printed representation ultimately denoting an object in the machine's memory, it finds memory for six binary cells, and initializes their contents to make a tree. Part of the tree can be shown in this diagram where the cells are depicted as [ | ], showing that they are made of two information fields stuck together.

         [ | ]
         /   \
procedure     [ | ]
              /   \
 administration    .. etc

The left half of the cell holds a pointer to the symbol, like procedure, and the right half of the cell holds a pointer to the next cell. The cells are just the backbone; the symbolic software doesn't concern itself with cells as such. They don't convey any meaning themselves, but they arrange the symbols into a particular tree shape for us.

Lisp lets us write the list another way, where we make it more explicit that there are cells. This is called the dot notation, and it looks like this:

(procedure . (administration . (examination . (entrance . (school . (law . nil))))))

The nil is a terminator symbol that ends the list: it shows that the last cell has no next cell. Likewise in the tree notations of X-bar theory you see nodes that only have one child; there is no further "X-bar" at the bottom! In Lisp we don't even want to see the cells, so the dot notation is only used in special situations. Even if you enter a list as (a . (b . (c . nil))), the machine will hide the dots and print it back at you as (a b c).

So these X bars are like binary cells. Don't try to read anything deep into X bars. Sometimes scientific theories are named due to some feature of the notations that they use, that is all! For instance, in category theory there are "comma categories". Why are they called comma categories? Because once upon a time, some notation using a comma was used in the notation used to manipulate these things on paper. Even though the notation is no longer used, the name has stuck. But that doesn't mean that the comma is important!!!—Preceding unsigned comment added by 192.139.122.66 (talk) 21:25, 17 December 2008 (UTC)

I agree! From a compsci and a mathematical perspective, the theory is trying to make up a "fantastical discovery" from a quite trivial and pretty nonsential "obvious" observation such as "everything can be structured pairwise in any tree structure". Or so the article seems to allege. The article content confuses semantics matters with purely data representational issues. Humans aren't Turing machines or anything nearly similar, so compsci data representation will have no impact on psychological and semantic issues whatsoever. ... said: Rursus (bork²) 13:48, 9 April 2009 (UTC)

If you look at the Substitution Test section of the article, you will see the psychological justification behind X bar theory. In fact, it was substitutions like the example that prompted linguists to develop the theory. Hadrian89 (talk) 13:48, 21 February 2010 (UTC)

Maximal projection not defined

The m-command entry refers here for a definition of maximal projections, but this entry doesn't actually define them. There's a definition of it at http://www.unlweb.net/wiki/X-bar_theory#Projections but I couldn't find what the license for copying is. (Some components of the UNL project are available under CC-BY-SA, but I couldn't find anything about the wiki.) The Crab Who Played With The Sea (talk) 10:54, 22 January 2013 (UTC)

Carnie 2013

There appears to be a concerted effort going on to add information to Wikipedia articles on syntax. One source is cited time and again, Carnie 2013. I fear that those who are conducting this effort are not yet knowledgeable enough to be adding information. Would one of the authors of these efforts please respond here. I may begin a more aggressive strategy to guard against these efforts. Any additions that either lack a citations (with a page number) or any addition that cites just Carnie (2013) may be removed without discussion. --Tjo3ya (talk) 23:40, 3 October 2013 (UTC)

Some changes

I'm making some changes to this article to try to reflect what is essential to X-bar theory per se (as a theory of projection and headedness) and what other claims have gotten lumped in with it, but are strictly speaking independent of it: that all languages have X-bar schemata, that all phrases follow the X-bar schemata, that there are no exocentric structures, that all phrases are binary-branching. These last four claims aren't claims of X-bar stx per se, and GPSG and even Chomsky from 1970-1986 would be surprised to find them. Chomsky's 1981 LGB used X-bar schemata, but didn't claim that S or S' had a head (IP and CP came in Barriers, 1986). GPSG never adopted those. Nor was binary branching a common assumption until much later, and not codified until the MP (1995) and Kayne's Antisymmetry book (1994). So I want to just have an explication of what X-bar is, and reduce the extraneous claims and evaluative prose. Mundart (talk) 11:18, 23 August 2015 (UTC)

I disagree with your changes. In my view, you are cutting out any sort of commentary that allows for critical reflection on value of X-bar theory. The article is now poorer because of the removals you have conducted. I am aware that you are now going to claim that whatever appears has to be sourced. That is of course a fair point. Let me see if we can come to some sort of compromise, though. --Tjo3ya (talk) 11:48, 23 August 2015 (UTC)--Tjo3ya (talk) 11:48, 23 August 2015 (UTC)

You seem to be appealing to old versions of X-bar theory that are no longer commonly assumed. For instance, do you disagree with the stance that X-bar theory as it is commonly understood in our modern times assumes strict binarity of branching? I'm pretty sure that most modern textbooks that present X-bar theory present a version that assumes all branching to be strictly binary. I do not have my sources with me at present (because I'm at a conference), but when I get back to my library, I will easily be able to back up the strictly binary branching approach. I'll be able to demonstrate beyond a doubt that most modern approaches to X-bar theory view it as strictly binary branching. I will also be able to back up the stance that most modern versions of X-bar theory assume all structures to be endocentric.--Tjo3ya (talk) 12:51, 23 August 2015 (UTC)

confusing inflectional abbreviation

In the x-bar theory graph for "He studies linguistics at the university", the inflection (I) is marked "—Pst" (presumably short for "past") even though "studies" is clearly in the present tense. I'm still a complete amateur at syntax so I very well may be overlooking something, but wouldn't "Prs" or "Pres" be a more appropriate abbreviation to use? Someone who actually knows what they're doing with this sort of thing, please help me out here. Elykzyr (talk) 17:22, 14 May 2020 (UTC)

@Elykzyr:: Me too being an amateur in linguistics, me too I am struggling with understanding the contents of the article (in its current form, which is most probably different from the one you read in 2020). But I may be able to supply an answer to your question (sixteen months after you posed it): "—Pst" is most probably intended to indicate "minus Past" for non-past, i.e. everything other than past, so either present or future (if time is progressing linearly, but that is a wholly different matter). From other things I have read on linguistics and languages, I have come to understand that writing "—Something1" may be much easier to use than writing "+ Something2 ∨ Something3 ∨ Something4 (etc.)" (where "∨" is mathematical notation for English inclusive "or"; I do not know about any linguistics convention for writing "or") if there are many possible Somethings while only one Something needs to be excluded.Redav (talk) 15:16, 25 September 2021 (UTC)

Regarding ambiguity and PSR in "Hierarchical Structures"

The following example: "I saw a man with binoculars" does NOT prove that PSR is incapable of capturing sentence ambiguities. Source [37] is a blog from a linguistics professor on a homework assignment addressing structural ambiguity with a primitive working PSR model. It does not help to support the argument made in this section, with the author writing "[w]hen we start drawing trees to describe the structure of the sentences, this results in two different trees". The argument made has no sources beyond that; it is pure conjecture on the editor's part. Creating a simple NP rule that includes an optional PP ( NP -> (D) (AP) N (PP)* ), would allow the PSR model to capture the NP modification reading.

TL;DR Either a better example to support the argument that PSR is "...incapable of capturing sentence ambiguities" needs to be found or this entire section needs to be rewritten. Azirars (talk) 18:09, 17 April 2024 (UTC)

Incorrect trees

The trees in the "directionality of branching" section about John and a book of linguistics have the constituent "of linguistics" as a sibling of the head, i.e. in complement position. I'm happy to change this, so long as others don't object - we do want an additional N' level to show "of linguistics" as an adjunct, right? AviationFreak💬 06:01, 13 October 2024 (UTC)

You're right about this being in the complement position. The question is whether it is indeed a complement. I'm with you in thinking it's an adjunct, but others may feel differently. Perhaps the solution is a different example. Changing book to discussion might make the of PP more clearly a complement.--Brett (talk) 11:02, 13 October 2024 (UTC)