Talk:Head (linguistics)

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Linguistics / Theoretical Linguistics  (Rated Start-class)
WikiProject icon This article is within the scope of WikiProject Linguistics, a collaborative effort to improve the coverage of Linguistics on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
Start-Class article Start  This article has been rated as Start-Class on the project's quality scale.
 ???  This article has not yet received a rating on the project's importance scale.
Taskforce icon
This article is supported by the Theoretical Linguistics Task Force.

Colorless green ideas is actually ungrammatical: English NPs have only one "slot" for color adjectives. Compare red blue tulips. Surely we can find a better example and not perpetuate an example (analysis) that was flawed from the beginning. --RJCraig 18:43, 24 January 2007 (UTC)

"Colorless green ideas" is a grammatical noun phrase. Reference: Colorless green ideas sleep furiously. 17:40, 14 March 2007 (UTC)

I don't think we need a reference to tell us that it is grammatical. Sprachgefuehl tells us that. However, the poster does have a good rule. It just doesn't work in this case.' —Preceding unsigned comment added by (talk) 07:04, 11 July 2008 (UTC)


While there does appear to be some information in this article (even though it's somewhat short, and it is unclear why the "parameter" article has to be separate from this one, seeing how most of the info overlaps, the contents are pretty much completely impenetrable to anyone who isn't a linguist himself.. could someone please add a few examples and more text, to clarify this article? (especially the first few lines are almost pointless, as they don't contrast the example with anything else)

"For example, in the big red dog, the word dog is the head, as it determines that the phrase is a noun phrase" < why is it meaningful to call "dog" the head, rather than just a noun? The example doesn't explain this at all, since all you give in the example is the noun phrase, and not how it relates to the rest of the sentence., or to the rest of the whole field of linguistics. What you want to point out is the difference between head first and head final phrases, not how adjectives modify nouns. —Preceding unsigned comment added by Boombaard (talkcontribs) 21:33, 5 January 2009 (UTC)


The examples given in the Head-initial vs. head-final section are illustrated using dependency grammar trees, which are rather different from the sort of trees used in mainstream syntactic theory. The difference between dependency and phrase-structure grammar is not relevant to that section (both theories make the same point), so the use of non-mainstream representations only obfuscates the point being made. This section should be redone using standard representations. rʨanaɢ (talk) 15:45, 27 December 2011 (UTC)

The benefit of dependency grammar trees (compared to their phrase structure grammar counterparts) is that they are simple. In this regard, I think they are appropriate for illustrating the distinction between head-initial and head-final structures. Consider in this regard what the phrase structure grammar trees of the example sentences would look like. They would take up more space and be less transparent, especially in view of the labels that would likely be necessary to distinguish heads from dependents. --Tjo3ya (talk) 03:40, 6 January 2012 (UTC)
First of all, dependency trees are not simpler for making this sort of introductory point to the audience that will be seeing this article. Anyone coming to this article with basic familiarity (such as from an introductory linguistics class or text) will have been exposed to standard trees but not dependency trees; likewise, in the future, they are far more likely to encounter standard trees anywhere else they go. Trying to teach a basic point using a non-standard representation just gives the readers the extra burden of having to mentally 'translate' the trees into familiar ones before they can understand the point that is being made.
Furthermore, in the particular section under discussion (head-initial vs. head-final languages), the use of dependency trees actively obfuscates the point. In a dependency tree like File:Kafka_English_tree.jpg, the head of each phrase actually appears to be wedged between two dependents, which will make a reader wonder why this is being called "head-initial" rather than "head-medial".
The bottom line is, readers will understand the concept of head-initial vs. head-final languages better if the point is illustrated using standard representations, rather than being grabbed as an opportunity to try and show off one's preferred theory. rʨanaɢ (talk) 18:05, 7 January 2012 (UTC)

Let's compromise. What I can do is produce the PSG and DG trees side by side. I'll make no judgement statements about either. Totally neutral wording. I'll choose a shorter example, so the trees do not occupy too much space. I will also state that the psg rendition is a "mainstream" analysis in theoretical syntax, but that the alternative DG trees have become the dominant analyses for parsers in computational models. These statements will all be absolutely accurate. In this manner, your concern that the DG trees will confuse readers should be avoided, and at the same time, the minority point of view (in Anglo-American theoretical syntax) will not be excluded.

Your point about head-initial vs. head-medial is going to be true for many PSGs as well (e.g. the early stages of the Transformational Grammar and most versions of HPSG), i.e. any PSG that does not assume a strict binarity of branching is going to have the same problem, i.e. there will be medial heads. What is needed in this regard, is a clarifying statement or two to reduce the source of confusion.--Tjo3ya (talk) 19:59, 7 January 2012 (UTC)

Regarding your last point, that is an issue that can't really be explained clearly to readers without talking about (in traditional syntax) the complement/specifier distinction, which is beyond the scope of this article. (i.e., in PSG, English is mostly spec-head-comp and Japanese is mostly spec-comp-head; in DG on the other hand, it looks like English is dep-head-dep and Japanese is dep-dep-head.) In other words, looking at the [over-]simple example below (the upper tree is PSG style and lower is DG style), if you tell the reader "saw" is the head of the sentence, then with either tree structure they're going to be confused because it's not in the front; the PSG response is that "she" is a specifier and I don't know what the DG response is, but either way it's not a good idea to get into specifiers in what is meant to be a simple overview article. The best way to avoid the issue entirely might be to use examples where specifiers don't come into play, like simple verb phrases (maybe something like "eat apples" and its equivalent in an SOV language). Glossing over specifiers also appears to be the strategy that was adopted by whoever edited the Head directionality parameter article. rʨanaɢ (talk) 20:25, 7 January 2012 (UTC)
  / \
She  saw'
     /  \
    saw  him

   /   \
  /     \
She     him

Both of your trees here are constituency-based (=PSG trees). To render the lower tree a DG tree, the lower occurrence of saw and the branch extending up from it must be removed. In DG trees, the number of nodes in the structure is equivalent to the number of words in the sentence, i.e. 3 words = 3 nodes.

I don't think the matter is as difficult to explain as you suggest. Your PSG tree is both head-final (because the head saw follows its spec-dependent She) and it is head-initial (because the head saw precedes its comp-dependent him). The same sort of explanation works for the DG tree; it is both head-final (because the head saw follows its dependent she) and it is head-initial (because the head saw precedes its dependent him). When a structure is both head-initial and head-final like this, the head appears in a medial position. That doesn't seem so difficult to understand.

In general, I agree with your emphasis on simplicity and understandability. But that is precisely where I see the advantage in DG trees. The DG representations are decidedly simple compared to their PSG counterparts, because they contain many fewer nodes and edges. --Tjo3ya (talk) 00:11, 8 January 2012 (UTC)

The tree I made was a DG tree, I am just not used to your formatting conventions. I used "|" where you use a dotted line in your trees; I guess you prefer if I leave the line out entirely. rʨanaɢ (talk) 00:29, 8 January 2012 (UTC)
Regarding simplicity, I will reiterate what I said before: even if you think it makes the point simply, illustrating head-initial versus head-final languages using a DG representation would be teaching readers something that is different than vast majority of representations they will see. Even if it is simpler, the unfortunate state of the field right now is that anyone learning syntax needs to understand the mainstream representations (and can then later go on to 're-learn' the idea in some other framework), and Wikipedia is not a vehicle to try to change that. I'm fine with your suggestion of presenting both types of trees (although I think presenting two representations will be more confusing than presenting one), but the standard sort of tree should not be excluded. My suggestion remains to limit the example to a small phrase. rʨanaɢ (talk) 00:35, 8 January 2012 (UTC)

I will redo the trees in a week or two, when I have access again to the tree-drawing software. I will produce the psg and dg trees side-by-side (as you suggest), using simpler examples to explain the distinction between head-initial, head-final, and head-medial. --Tjo3ya (talk) 18:42, 9 January 2012 (UTC)