Jump to content

Talk:Constituent (linguistics)

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Initial post

[edit]

I cleaned up the headings above the first paragraph, but I think it needs more work. I'll look at it again later. Dalbury 21:01, 13 October 2005 (UTC)[reply]

I made some relatively minor changes and removed the "wikify" tag, as I believe the article now substantially conforms to standards. - Dalbury 16:37, 16 October 2005 (UTC)[reply]


Ad coordination test: how does non-constituent coordination fit into the picture? --Jirka6 02:56, 21 April 2007 (UTC)[reply]

The introduction needs more work. The second paragraph should make its example clearer, should probably state explicitly that "what he said" is a constituent. Thomas Tvileren (talk) 17:11, 8 November 2011 (UTC)[reply]

consitituent tests

[edit]

Just thought I'd point at this page Talk:Polysynthetic_language#syntactic_constituent_tests - the "who kicked the carrot" examples on that page cleared things up very nicely for me, so I wondered whether they might be a good import for this page (if all the aspects those examples describe are not already covered in this article) EdwardLane (talk) 09:39, 5 December 2011 (UTC)[reply]

Phrase structure, dependency structure

[edit]

Before an edit war breaks out, let's have some discussion here. I think the latest edit, at least in the last paragraph of the section, is pushing a POV. Moreover, the section is poorly sourced. On the other hand, I think it would be a mistake to simply remove the whole section. So, suggestions for improvements, please. -- Donald Albury 10:56, 27 December 2011 (UTC)[reply]

You beat me here. I agree this section is unwarranted, for the following reasons:
  • Like Donald pointed out, it's pushing for the author's preferred linguistic theory (see User:Tjo3ya), which is not an appropriate use of Wikipedia. The purpose of Wikipedia is to present factual information, not to make arguments in an unfolding debate. It's one thing to summarize other people's widely documented arguments (done in articles like, e.g., Debate on traditional and simplified Chinese characters and Plate tectonics), but it's entirely another to tell readers "here's evidence that the theory I subscribe to is right, let me walk you through it".
  • The section is not even about the topic of constituency; rather, it's using the topic of constituency to make an argument about whether phrase-structure grammar or dependency grammar is "right". Ignoring for now the logical issue (i.e., the author's conclusion doesn't necessarily follow from the data he presents in this section--but that's not relevant to the editorial issue here so it would be a waste of time to discuss it here), this argument is just not relevant to the topic at hand.
To retain what little information is relevant and do so in an NPOV way, the whole section could be replaced (and possibly moved to elsewhere in the article) with a single sentence along the lines of "in [[dependency grammar]], [[head (linguistics)|]]s are not constituents". (I assume this is a slight oversimplification, but this appears to be the general idea: in dependency grammar, nodes dominating other nodes cannot be constituents, and for the most part those nodes correspond to the heads in a traditional tree structure.) rʨanaɢ (talk) 15:33, 27 December 2011 (UTC)[reply]
Rjanag, you write "The purpose of Wikipedia is to present factual information, not to make arguments in an unfolding debate". I think the section presents factual information, and I think by doing so, this information will help the debate unfold. If there is anything inaccurate about the claims in the section, I encourage you (or anyone else) to please point them out. I am certainly not interested in misrepresenting the facts. Quite to the contrary, I want the facts to be known and for a debate to be unleashed.
I encourage you and Donald to examine the issue closely. Take an example sentence and apply the constituency tests to each and every phrase structure constituent in that sentence in a neutral fashion. You will find that most (coordination is the one major exception here!) of the constituency tests fail to identify a majority (I repeat, a majority) of phrase structure constituents as constituents. At the same time, they succeed at identifying a majority (but not all!) of the dependency grammar constituents as constituents.
Wikipedia is widely used by novice learners. The novice learner is going to be frustrated by the fact that widely employed constituency tests do not deliver results that match the constituency-based structures that they are learning. In this regard, there needs to be general awareness that there is a major alternative to the phrase structure grammar understanding of sentence structure. Without the key section that draws attention to the alternative, the debate cannot unfold and we are all poorer as a result.
Concerning sourcing the section, I can add more citations to peer-reviewed publications. The difficulty, however, is that the sources that address the issue directly are those of me and my co-authors.
Finally, I would like to avoid an editing war. I am not comfortable, however, with omitting the section entirely or to reducing it to a single sentence. I can rewrite it in such a manner that seems more neutral, but in the scheme of things, the facts are difficult to interpret in a manner that leads one to a neutral conclusion.
I encourage you guys to comment further. - Tjo3ya.

The reason that Rjanag deleted this section is apparently that it says that one kind of structure claims that single words are constituents whereas the other kind of structure does not make this claim. This section does not say that one kind of structure is right and the other is wrong. But perhaps Rjanag believes that this section encourages this inference.

To settle this issue, there would have to be discussion in places other than Wikipedia as to the correctness of this claim. Are single words constituents or are they not? Dnmaxwell (talk) 02:57, 28 December 2011 (UTC)[reply]

"Constituent" is not a term unique to PSGs. D[ependency]G[rammar]s can acknowledge constituents, too. Clearly, the theories dubbed "main-stream" in this discussion don't have the prerogative over the term. Rather the term is open to all approaches in syntax. Hence I see no compelling argument why "constituent" shouldn't be discussed also from the DG angle. The argument from "main-stream" also seems dubious. I didn't know it was Wikipedia policy to feature issues as seen solely from the "main-stream". If so, further clarification is required what qualifies as "main-stream". One should also be aware that some of the issues raised by Tjo3ya feature in several publications in grade A journals (Syntax, The Linguistic Review, Cognitive Linguistics). Rikaido88 (talk) 04:06, 28 December 2011 (UTC)[reply]

Debate is necessary

[edit]

Rjanag, I have revised and restored the section. I have tried to use as neutral wording as possible, but again, it is difficult to arrive at a neutral conclusion when one examines the facts. You write that the section should not be restored until consensus has been reached. I think the opposite holds. The section should not be removed until there is a consensus to that effect. No consensus will be reached, however, if the section does not appear, for there will be no debate at all. Please do not delete the section, since this debate is really necessary. -- Tjo3ya.

Edit warring

[edit]

Despite my request to avoid an edit war and instead discuss changes here on the talk page, both of you are edit warring. Neither of you are helping your case by doing so. I'm not going to get involved in the technical details, as my linguistics is way too rusty (more than 35 years since I left the field). Cool it, talk here, seek third party opinions, but stop reverting each other. -- Donald Albury 12:58, 28 December 2011 (UTC)[reply]

I'm afraid Tjo3ya is not familiar with WP:BRD, which says that once his addition is removed he should refrain from re-adding it until consensus is reached; he also seems to be unfamiliar with Wikipedia's guidelines on canvassing, given that he appears to have canvassed his colleagues off-wiki to come support him in this discussion.
Regarding the content issues, my main points above have not really been responded to. Even in the most recent version, this section is still not about the topic of constituents, it's still framed as an argument about which type of tree better represents constituency (and, even though Tjo3ya and his colleagues are denying it, it clearly presents a slanted version of the argument that comes out in favor of his preferred framework). If Tjo3ya can't manage to write a section that is actually about the topic of this article, then there is no reason to include it. If he wants to include a section like this, it should be a section that is actually about some argument over whether single words are constituents (citing sources that indicate there is a debate over this), and that can be done without even mentioning PSGs and DGs.
I'm not a theoretical syntactician and it's been a long time since I've read an intro syntax book or any syntax articles, but I am somewhat doubtful that "can single words be constituents" even is a major debate. As I hinted at above (but wanted to avoid getting into), Tjo3ya's argument is a bit shaky. First of all, as he himself acknowledges in this section, one of the words in question does pass at least one common constituency test (it also passes a replacement test, which Tjo3ya failed to mention), and the reason it fails topicalization and clefting/pseudoclefting tests is because English word order is not as free with regard to verbs as it is for some other things. Secondly, it is widely known that constituency tests are merely an aid and they don't necessarily dictate what is and is not a constituent; they are to be used judiciously, and the results taken with a grain of salt.
Long story short: this section is written in such a way that it's not relevant to the article's topic, it's POV even though the author doesn't seem to see it, and its claims are logically shaky. It needs to be removed until such time as Tjo3ya can draft an acceptable replacement (preferably in his own userspace, for instance in User:Tjo3ya/Sandbox, where it can be reviewed by other editors before being moved to the article). rʨanaɢ (talk) 14:47, 28 December 2011 (UTC)[reply]
Rjanag, I do not agree with your interpretation of BRD. You acted, Tjo3ya reverted. At that point, you both should have stopped editing the section and discussed it here. Your reversion of Tjo3ya was a step down the slope to an edit war. -- Donald Albury 21:25, 28 December 2011 (UTC)[reply]
Since I am one of the contributors of the revised page, I need not be canvassed. Note that Rjanag seems to see the argument from canvassing as an excuse not to address the content of what I wrote. PSGs as well as DGs make use of the term “constituent”. What qualifies as a constituent is different, though. This difference needs to be shown and explained on the respective page. If there is a difference, then the question arises which approach is the more adequate one. A comparison of PSG candidates and DG candidates for constituency will show that DG candidates pass constituency tests, while many PSG candidates do not. The caveat on constituency tests exists for exactly the reason that many PSG units, many of them again central to the system, fail to conform to these tests. The discussion of these issues can surely appear in different sections on the page, but stating the fact that PSGs do worse in constituency test than DGs is not really “pushing a POV”.Rikaido88 (talk) 07:15, 29 December 2011 (UTC)[reply]
Rikaido88, who are you? You claim to be a contributor to the page, but your only edits are to this talk page. -- Donald Albury 14:51, 29 December 2011 (UTC)[reply]
Donald, I created picture files of the tree structures Tjo3ya uploaded. In this sense, I am a "silent" contributor, and an interested party. My professional field is theoretical linguistics.Rikaido88 (talk) 00:22, 30 December 2011 (UTC)[reply]

Constructive criticism

[edit]

Rjanag,

Edit warring is bad for both of us. We clearly profoundly disagree about the core issue. I want to reach a compromise. Here's what I can do:

1. I could email to you the exact pages from A-level peer-reviewed articles where the issue is addressed directly. I could also cite these articles directly in the section (although doing so would fill the literature list with my name, which I want to avoid).

2. I could expand the section, including more data to further illustrate the point. The drawback here, though, is that the general point would become stronger and would therefore likely be perceived as an even greater challenge to phrase structure grammar.

3. I could produce a much lengthier illustration and discussion of the issues here on this discussion page. This approach might be beneficial since it would provide more background information to help others add their views. Be aware, however, that for me to do this, the original section needs to remain more or less intact on the main page. If it is relegated to my Userpage or appears only on this discussion page, the issue will die because nobody will see it. The debate will not occur. That result is unacceptable for me.

4. I could rewrite the section in a manner that directly addresses suggestions from you. To do this however, I need concrete constructive suggestions. I need to see more clearly how I can revise the section in a manner that can take your views into account.

5. I could produce a separate article that addresses the controversy in depth. I am quite open to this possibility. But to do this, the section needs to remain more or less intact where it is. I would link to the additional article. If the section does not remain more or less intact where it is, no one would proceed to the additional article and the message would be lost; the debate would again be silenced.

In sum, what can I do to ease your concerns?

--Tjo3ya (talk) 18:51, 28 December 2011 (UTC)[reply]

You still have not responded to the issues I raised: the section is irrelevant to the topic of the article and is pushing a POV. Also, please be aware that Wikipedia is not a soapbox for spreading awareness to the masses. Arguments like "the issue will die if it's relegated to my userpage" and "the debate will be silenced" are not appropriate or helpful; if your goal is to 'inform the masses', you may wish to consider a related project such as Wikinfo. (Also, if you read my message you will see that I never said the content should be relegated to your userpage; I said you should write a draft there and allow it to be reviewed by editors before posting it in the article.) But I am tired of repeating myself so I have asked some uninvolved editors to comment. rʨanaɢ (talk) 19:27, 28 December 2011 (UTC)[reply]


Rjanag,

If I write a draft on my user page, how do I get editors to read it and who are these editors? How do I get people to look at the page if it is not a main page? You, for instance, would likely not visit my user page to provide feedback on the draft of a section for the main page of this article. In fact my guess is that I am now debating this issue with you precisely because you originally read the section on the main page here and disagreed with it.

Your stance that the section does not belong in the article is mistaken, in my view. Many aspiring or established linguists who are examining constituent structure and considering constituency tests (something which most syntax and linguistic textbooks do) is going to question why a majority (I repeat, a majority) of the units that are posited as constituents (in a phrase structure grammar) are not identified as constituents by most of the standard constituency tests. Let's take another example and examine the point. Consider the word "wrote" in the following sentence with respect to the eight constituency tests enumerated in the main article (note that most of the constituency tests were already in the article before I revised it):

(1) Fred wrote a draft.

*Wrote Fred a draft. - Topicalization
*It was wrote that Fred (did) a draft. -Clefting
*What Fred did a draft was wrote. - Pseudoclefting
What did Fred do about a draft? *Wrote. - Answer fragment
?Fred did a draft. - Pro-form substitution.
A draft was written by Fred. - Passivization (test inconclusive, since "wrote" does not appear in the passivized sentence)
*Fred a draft. - Ommission
Fred [pondered] and [wrote] a draft. - Coordination

Given the standard definition of the "constituent", phrase structure grammars unanimously view "wrote" as a constituent, whereas dependency grammars unanimously view it as a non-constituent. Of the eight tests, five agree that "wrote" is not a constituent. One of them (coordination) disagrees, and two of them are inconclusive. Passivization is inconclusive because the test word "wrote" no longer appears in the passivized sentence, and pro-form substitution is inconclusive because it is not clear that we can view "did" as a pro-form standing in for "wrote" -- it may have full lexical content (especially in view of the requirement that it occur with an agent subject and can therefore not stand in for non-agentive verbs, e.g. "contain, happen, take place, etc.").

The aspiring or established linguist who ponders data like these might be led to the conclusion that constituency tests are really not very useful at all, since they (with the exception of coordination) suggest that a unit that all phrase structure grammars take to be a constituent is not really a constituent. At this point, awareness of the dependency grammar alternative needs to come into play. The aspiring and established linguist needs to be made aware of the fact that there is another way of understanding constituent structure that may be more compatible with the results of the constituency tests. The article on constituents is the primary place where this alternative needs to appear.

Let me state again that I am very willing to work with you, Rjanag. How can the issue just presented be developed in as neutral a manner as possible? I could change the wording significantly. I could draw attention to the fact that phrase structure grammars reach to the distinction between minimal, intermediate, and maximal projections to address the discrepancy. Perhaps that might ease your concerns. Let me know.

--Tjo3ya (talk) 21:27, 28 December 2011 (UTC)[reply]

Tjo3ya, what are your sources for your argument? Please provide citations to reliable sources for all of your claims. I hope that you are not presenting your own original research here. -- Donald Albury 14:56, 29 December 2011 (UTC)[reply]
@Tjo3ya: I waited until I got home from vacation to respond to your message, so I could look at what some popular introductory syntax textbooks say about constituency (since you seem to be concerned about students who might be reading this). I noticed that both Adger (2003) and Carnie (2002) [the former is a Minimalist perspective and the latter is GBT] define a constituent as a group of words (Adger 2003:63; Carnie 2002:31), which makes your concern about verbs and content words moot. I have nothing more to say about the theoretical issue because, as I have said repeatedly above, I am not interested in getting into a theoretical debate with you—the issue here is not which theory is better, the issue is an editorial issue of how you are editing the article. I will repeat (for the last time) the issues that I have already pointed out and you have failed to address:
  1. The section is POV, with you cherry-picking examples to try and show that your theoretical standpoint is the right now.
  2. The section is poorly framed, such that what you write doesn't even come across as being about the topic of constituency, but rather about what theoretical model is a better way of representing it, which is not really relevant here.
But I am tired of repeating myself so I have asked some uninvolved editors to comment. rʨanaɢ (talk) 07:00, 3 January 2012 (UTC)[reply]

Definitions

[edit]

Rjanag,

Please dig a little more. For definitions of the "constituent", see the following sources: Napoli 1993:167; Jacobson 1996:55; Haegeman and Guéron 1999:51; Carnie 2008:37.

Carnie, A., 2008. Constituent Structure. Oxford University Press, Oxford.

Haegeman, L., Guéron, J., 1999. English Grammar: A Generative Perspective. Basil Blackwell, Oxford.

Jacobson, P., 1996. Constituent structure. In: A Concise Encyclopedia of Syntactic Theory. Pergamon, Cambridge.

Napoli, D., 1993. Syntax: Theory and Problems. Oxford University Press, New York.

By attempting to argue that phrase structure grammars do not necessarily view individual words as constituents, you are weakening your credibility in this debate and increasing my willingness to resist at all costs. You need to concede the issue. Phrase structure grammars do necessarily view each individual word as a constituent (and in this regard, the results of the constituency tests do not support the phrase structure grammar approach to syntactic structure; the tests deliver results that are much more (but not entirely!) congruent with the dependency grammar understanding of sentence structure).

Take a simple verb phrase as an example, e.g.

(1) acknowledge fact

You are attempting to argue that this verb phrase is a constituent (in the greater sentence that contains it), but that it cannot be broken down into the smaller constituents acknowledge and fact. In other words, acknowledge and fact cannot be viewed as the constituent parts of acknowledge fact. I challenge you to find any established phrase structure grammarian who would accept that claim. Let me know who they are (so I can contact them).

Where you and Taivo below are correct concerns the majority vs. minority approach. Phrase structure grammars are dominant in Anglo-American linguistics. What you have backing your point of view is therefore the argumentum ad populum, i.e. the phrase structure grammar approach must be right because it is the dominant view. The argumentum ad populum is, however, considered a fallacious argument in rhetoric.

Observe that Taivo below seems to be willing to compromise in allowing a reduced and revised version of the section to appear. Hopefully you too will compromise and allow the revised section to survive. --Tjo3ya (talk) 14:59, 4 January 2012 (UTC)[reply]

Sources

[edit]

Donald,

The relevant sources that discuss the issue directly are listed here (exact pages included):

Osborne, Timothy 2005. Beyond the constituent: A dependency grammar analysis of chains. Folia Linguistica 39/3-4, 251-297. See pages 254ff..

Osborne, Timothy 2006. Shared material and grammar: A dependency grammar theory of non-gapping coordination. Zeitschrift für Sprachwissenschaft 25, 39-93. See pages 53ff.

Osborne, Timothy 2008. Major constituents: And two dependency grammar constraints on sharing in coordination. Linguistics 46/6, 1109-1165. See pages 1126ff.

Osborne, Timothy, Michael Putnam, and Thomas Groß 2011. Bare phrase structure, label-less trees, and specifier-less syntax: Is Minimalism becoming a dependency grammar? The Linguistic Review 28: 315-364. See pages 322-324.

These journals are peer-reviewed A- and B-level. Linguistics, Zeitschrift fuer Sprachwissenschaft, and The Linguistic Review are A-level (I think), and I think Folia Linguistica is B-level. I would happily email to you (or anyone else) these papers so that you can look at them directly.

I list one of these sources in the section, and I could add the other three. I chose not to do this because I did not want my name to dominate the literature list (a point which I mentioned above).

While I am the primary author of these sources, the issue should not be understood as something particular to one narrow research agenda of a particular linguist. The constituency vs. dependency distinction is far-reaching. Both constituency grammars (= phrase structure grammars) and dependency grammars are going to be interested in the broadest sense and will have something to say about the issue. If you remain skeptical about this point, I can solicit a number of professors in linguistics to come to this page to back me up on the point, e.g. Thomas Gross, Sylvain Kahane, Kim Gerdes, Igor Mel'cuk, Mikel Forcada, Mike Putnam (google these names!). But please note that these linguists may or may not be willing to come here to deliver a statement on the matter.

Finally, the dependency grammar approach to syntax and grammar is a minority point of view in theoretical syntax in Anglo-American linguistics. It's representation in Europe is much stronger, however. My understanding of the Wikipedia mission is that a plurality of voices should be accommodated. The minority point of view should not be excluded. --Tjo3ya (talk) 17:24, 29 December 2011 (UTC)[reply]

Another View

[edit]

I have reverted the whole "dependency theory" material because it represents a minority view. Wikipedia should presents alternate points of view, however, the section that was written suffered from two major flaws. One, it was too long relative to its importance in the field. A single paragraph or less is quite sufficient based on the length of the article and the relative obscurity of the theory. Two, it was written in an argumentative style to attempt to prove its adequacy over standard theory. That is an unacceptable writing style in Wikipedia. This article is about what linguistics has to say about what constitutes a "constituent", not a debate about whether a constituent is X or Y. In referencing about a dozen introductory textbooks and several dictionaries and encyclopedias of linguistics, there is not a single one that defines "constituent" as a single word. All of them define "constituent" as "a word or group of words" and include the fact that individual constituents can be combined into larger units that are themselves constituents of higher level units, etc. This is a virtually universal definition of "constituent" in linguistics. While the minority viewpoint of dependency grammar should be mentioned, it should only be described briefly and not argued for. --Taivo (talk) 13:21, 3 January 2012 (UTC)[reply]


Taivo,

Who are you? You know who I am from my Userpage. I'd like to know who you are. I'd like to get a sense of who I am dealing with here. Note that you can send me an email (my email address is on my Userpage). I will not divulge your identity here.

The constituent is typically defined as follows:

Constituent
A node/word plus all the nodes/words that that node/word dominates.

This definition identifies any complete tree or subtree of a tree as a constituent. It is neutral with respect to the constituency vs. dependency distinction. Hence in a phrase structure grammar (e.g. GB, MP, HPSG, LFG, etc.), every individual word qualifies as a constituent by definition (contrary to what you claim) as well as as any group of words (=phrase) that constitutes a complete subtree. In a dependency grammar (e.g. Word Grammar, Meaning-Text Theory, Lexicase, Functional Generative Description) in contrast, only those individual words qualify as constituents that do not dominate other words (the trees in the section you cut illustrate this point). Consider the following example:

Fred has been trying to say something.

In a phrase structure grammar, every individual word is a constituent (Fred, has, been, trying, to , say, something) and every phrasal group of words is a constituent (say something, to say something, trying to say something, been trying to say something, and has been trying to say something). In a dependency grammar in contrast, only those individual words are constituents that do not dominate other words (Fred and something), and the following groups of words are also constituents (say something, to say something, trying to say something, and been trying to say something).

Examining what the constituency tests tell us about sentence structure, they by and large (again, coordination is the one major exception here) support the DG stance. The tests fail to identify has, been, trying, to, and say as constituents. I encourage you (and anyone else reading about this debate) to probe these claims. Perform the constituency tests on each of the phrase structure constituents.

You acknowledge that the minority stance should be accommodated. I appreciate that. It is important to me in this regard that the dependency vs. constituency distinction be illustrated with the trees. I request further input from you in this regard. I can envisage a way to shorten the section a bit. The actual test sentences (topicalization, clefting, and pseudoclefting) can be cut. But again, it is important to me that the minority stance be accommodated with the appearance of the trees.

Finally, you characterize the DG understanding of syntax and grammar with the word "unimportance". You of course realize that you have just insulted a significant tradition in the study of syntax and grammar. Google Richard Hudson and Igor Mel'cuk in this regard. Once you have acquired a sense of the work of these linguists, maybe you will not be so quick discount the DG tradition. The unfortunate choice of words reveals a lack of breadth in the field of theoretical syntax.

--Tjo3ya (talk) 16:42, 3 January 2012 (UTC) (Timothy John Osborne, tjo3ya@yahoo.com, Email me!)[reply]

First, it doesn't matter who I am. I am an informed Wikipedia editor. This enterprise is not about measuring how large my testicles are compared to yours. It's entirely about reliable sources and relative weight. Second, I have edited my comment above because "unimportance" was a typo and should have been "importance". Dependency grammar is of much less relative importance to the field than phrase-structure or universal grammar. Third, it's very important that you compose a neutrally-worded sentence or two about the role of constituents in dependency grammar and clearly (and simply) state how they do or do not differ from the more mainstream view of what a constituent is. It's not the place for arguing your point, but the place for a very simple distinction. Remember that this article's intended audience is not professional linguists or even linguistics students (I don't allow my students to cite Wikipedia, for example), but a place for more generalized readers to understand the broad strokes of a topic. It's not the place for theoretical argumentation and persuasion, but a place for a neutral overview with proper references if the reader desires to investigate the issue further. Since this has proven to be controversial, my suggestion to you is to compose the paragraph you would like to see included and present it here on the Talk Page first. I assure you that if it's of appropriate length, neutrally worded, and not overly technical or filled with jargon, we can probably come to a consensus on it or tweak it only slightly. --Taivo (talk) 18:06, 3 January 2012 (UTC)[reply]

Taivo,

I have revised and shortened the section as follows. I hope the draft is closer to what you could approve.

--Tjo3ya (talk) 20:38, 3 January 2012 (UTC)[reply]

Competing theories

[edit]

Alternate theoretical approaches to syntax make different assumptions regarding what is considered a constituent. In mainstream phrase structure grammar (and its derivatives), individual words are constituents in of themselves as well as being parts of other constituents, whereas in dependency grammar[1] certain core words in each phrase are not a constituent by themselves, but only members of a phrasal constituent. The following trees show the same sentence in two different theoretical representations, with a phrase structure representation on the left and a dependency grammar representation on the right. In both trees, a constituent is understood to be the entire tree or any complete labelled subtree (a node plus all the nodes dominated by that node)—note that words like killed and with form subtrees (and are considered constituents) in the phrase structure grammar representation but not in the dependency structure representation.

Illustrating constituency and dependency

Notes

[edit]
  1. ^ Agel et al. (eds.) 2003/2006 full reference required

Comments on revised section

[edit]

This sounds way too much like a discussion in a textbook and not an encyclopedic entry. Don't start a section with "Examine the following two trees". All the text below the diagrams should be above them and replace what is there. The last sentence is also kind of "squishy". It would be better: "Constituency tests will yield different constituent results on the two trees." "More accurate" is a subjective and unencyclopedic judgement call on two competing theories. The two trees should be labeled "Phrase-Structure Grammar" and "Dependency Grammar". The current labels aren't really meaningful except in a Dependency Grammar context. It needs an introductory statement like, "Dependency Grammar treats constituency differently than in Phrase Structure Grammar. The following tree on the left is..." The way it reads now makes it sound like this is part of mainstream theory (and it's not). --Taivo (talk) 20:48, 3 January 2012 (UTC)[reply]

Taivo,
I have revised the section in accordance with the your suggestions. To relabel the trees, however, I need to use the special software that produces pictures for Wikipedia. I do not have this software with me at this time (but I can have a helper get it done if necessary). But is that really necessary? I've added the clarification "(= constituency grammar)", which should prevent misunderstanding. I think it is beneficial for the reader to know that phrase structure grammars are constituency grammars. Please let me know if you can accept this version. --Tjo3ya (talk) 16:59, 4 January 2012 (UTC)[reply]
Remove the statement "(perhaps outdated)". That's a judgment call and not neutral. We need to see what Rjanag thinks about this as well. It's better than it was, but I need to spend a bit more time with it than I have today. But we definitely need Rjanag's input as well. --Taivo (talk) 17:06, 4 January 2012 (UTC)[reply]
Taivo says that Dependency Grammar is of much less relative importance to the field than Phrase-Structure Grammar. Whether that is true or not depends on your interpretation. PSG is certainly much more widely taught at universities and is much better represented in journals. On the other hand, DG is dominant within computational linguistics. The organizer of DepLing, the first international conference on DG, said at the beginning that 24 of the 25 parsers he was able to find out about are using DG. I think we all hope that work on linguistics has applications to the real world. The most widely developed kinds of applications are coming from computational linguistics. And DG is way ahead of PSG in this area. This gives evidence of a big gap between theory and practice in linguistics as in other areas. Dnmaxwell (talk) 17:31, 4 January 2012 (UTC)[reply]
Taivo,
My helper has provided the tree that you requested, which I've inserted it into the draft above. If Rjanag remains silent concerning the draft, I plan to go ahead and post it to the page tomorrow. I hope you are OK with that. --Tjo3ya (talk) 16:29, 5 January 2012 (UTC)[reply]
I find it disturbing that 1) you continue to try to "out" editors and 2) you still try to throw around credentials here. I have edited your comment above to remove the personal identification. Please respect the Wikipedia process--anonymity and equality. We edit here based on reliable sources you bring to the table, not on the degrees you have framed on the wall or who signs your paycheck. --Taivo (talk) 17:52, 5 January 2012 (UTC)[reply]
Taivo,
Hmmm. If I click on your signature, I am one second away from the following information:
  • Warning: This editor is a specialist in a scientific field, so he often runs afoul of the "Randy in Boise" trap.
  • I am an associate professor of Linguistics at a major U.S. university. I have taught Linguistics in the U.S. and English Linguistics in the Ukraine.
Now who is touting whose qualifications? Please refrain from editing my comments, and if you were honest about this process, you would not have removed your characterization of the DG tradition using the word "unimportance".--Tjo3ya (talk) 18:45, 5 January 2012 (UTC)[reply]
I told you already that "unimportance" was a typo, so get over it. If you look at my comments here, you'll notice that I am not using my academic position as an arguing point for this discussion. Stating in broad terms on my user page who I am is not the same as using my qualifications as an arguing point on this page. You'll notice that the basis for my points is looking at the introductory texts and linguistics dictionaries and encyclopedias--reliable sources, not my qualifications. You will notice that I don't have a list of publications or even my specializations on my user page, just a general comment about what I work on. But my comment wasn't about you or spreading your qualifications out for all to see, it was about you "outing" another editor with his/her personal information and using their position as "full professor" as some sort of endorsement for the issue. I don't care if you shout your own name from the rafters, just as long as you know it doesn't matter here, but it is inappropriate to ask for identification from other editors and especially to "out" other editors. You are obviously sensitive about who you are and your qualifications (and apparently the qualifications of other editors), but you need to understand Wikipedia's position concerning them and work within that framework or find another outlet for your linguistic endeavors. This is all I'm going to say on the matter. You're a new editor to Wikipedia, so I was trying to offer you some collegial advice on how best to work here, but you seem to be uncomfortable with it. I'm not going to say anything else on the issue and will let you flounder on your own if that is your desire. --Taivo (talk) 20:10, 5 January 2012 (UTC)[reply]
Just so it's not lost, your new paragraph looks much better than the original one. I'd still like to hear from Rjanag to see whether or not it meets his concerns before you post it. Give it a couple of days--sometimes real life gets in the way of responding on Wikipedia. --Taivo (talk) 22:18, 5 January 2012 (UTC)[reply]
And concerning the editing of your comment, you don't have the right to post any personal or identifying information about another editor or "assistant". Period. This is not subject to debate and any other editor or administrator would have done exactly the same. Read WP:OUTING. --Taivo (talk) 00:32, 6 January 2012 (UTC)[reply]
The version I'm looking at is so vague as to be useless to just about any reader. If I weren't reading it without the background I have (by which I mean awareness of this very discussion), I wouldn't understand information the section is trying to convey. As I already explained above, an appropriate way of framing this section would be to explain that some alternative theories of grammar have different views about what constitutes a constituent. Going into detail about which tests various words fail and pass is unhelpful (both because that level of detail is unnecessary for a summary and would just look like argumentation, and because, as I said above, they are inconclusive anyway--some tests are passed and some are failed, you appear to have cherry-picked which tests to present, and in many cases traditional grammar has alternative explanations [such as island phenomena] for why some words might fail certain movement-based constituency tests... which is the whole reason for the caveat that consituency tests need to be taken with a grain of salt). As I explained above, little more needs to be said other than one sentence pointing out that there are disagreements about what counts as a constituent, and providing references for further reading. Your original version of this section was inappropriately written as an argument supporting one particular view (without even making the main disagreement clear), and the latest version doesn't even get the point about constituency across. I have already explained (here and in earlier messages) what would be an appropriate and encyclopedic way to write this, but my suggestions have been repeatedly ignored because you seem to be interested only in plugging your preferred grammar rather than actually improving the article. rʨanaɢ (talk) 17:56, 7 January 2012 (UTC)[reply]
Well I thought that the section was clear and easy to understand. Of the two citations, "Osborne 2008" clearly meant: "Major constituents: And two dependency grammar constraints on sharing in coordination". I have no idea what "Agel et al. (eds.) 2003/2006" refers to. Please can this be made clear.--Toddy1 (talk) 19:33, 7 January 2012 (UTC)[reply]
Agel et al. is a massive two volume collection of essays on dependency and valence grammar. The complete citation is as follows: Ágel, V., Ludwig Eichinger, Hans-Werner Eroms, Peter Hellwig, Hans Heringer, and Hennig Lobin (eds.) 2003/6. Dependency and Valency: An International Handbook of Contemporary Research. Berlin: Walter de Gruyter.
Like I said in my last message: what the section says is clear, but it says almost nothing. In particular, it doesn't express the one idea that is important, the idea that different theories supposedly make different predictions about what qualifies as a constituent (the reader has to try to figure that out based on the sentence "In both trees, a constituent is understood to be the entire tree or any complete subtree within the tree" and the fact that the trees have different subtrees). It is also confusing: the sentence "Constituency tests will yield different constituent results on the two trees" is both confusing and inaccurate (because constituency tests are performed on sentences, not trees, and the same sentence is shown twice so the same constituency test results will be obtained each time; I assume what Tjo3ya wants to be saying is that one tree better reflects the constituency tests than the other, although as we have explained ad nauseum above that would be POV argumentation).
That whole section actually says less (at least, less that is apropos to this article) than my single-sentence suggestion above. rʨanaɢ (talk) 19:42, 7 January 2012 (UTC)[reply]

A full illustration

[edit]

Rjanag,

Since you continue to dispute the original message of the section with your "cherrypicking" accusation, I will try to illustrate the issue now more fully here with another example:

(1) Sam has been trying to understand the distinction.

PSGs assume 15 distinct constituents for this sentence: Sam, has, been, trying, to, understand, the, distinction, the distinction, understand the distinction, to understand the distinction, trying to understand the distinction, been trying to understand the distinction, has been trying to understand the distinction, and Sam has been trying to understand the distinction.

DGs, in contrast, assume just 8 constituents: Sam, the, the distinction, understand the distinction, to understand the distinction, trying to understand the distinction, been trying to understand the distinction, and Sam has been trying to understand the distinction.

The main difference lies with the individual words, although DGs also reject the existence of a finite VP constituent. Concerning the individual words, the majority of the eight constituency tests enumerated in the article will identify only Sam as a constituent. Coordination is the one major exception; I discuss it below. Since the determiner the is a constituent in both PSGs and DGs, its failure to pass the constituency tests sheds no light on the constituency vs. dependency debate. The reason it fails is in part Ross' (1967) Left Branch Condition, i.e. extraction cannot occur from a left branch under a noun (phrase). Your point about islands is correct in this regard. Since the DG analysis sees the majority of individual words as non-constituents, it is much more accurate.

Consider next the finite VP string has been trying to understand the distinction. Modern PSGs unanimously assume a finite VP constituent, whereas DGs unanimously reject the status of finite VP as a constituent (DGs do, however, readily acknowledge the status of non-finite VPs as constituents). The constituency tests again strongly support the DG stance:

*Has been trying to understand the distinction, Sam. -Topicalization
*It is has been trying to understand the distinction, that Sam (has done). -Clefting
*What Sam has done is has been trying to understand the distinction. -Pseudoclefting
*What has Sam been doing?  ??Has been trying to understand the distinction. -Answer fragment
*Sam did so. (did so = has been trying to understand the distinction) -Pro-form substitution
??? -Passivization (Inapplicable)
*Sam. -Omission

Based on these results, we have no reason to view has been trying to understand the distinction as a constituent.

While I picked an example that would illustrate well the issue under debate, I have considered all the constituents in the example sentence. The DG view of sentence structure is again much more congruent with the results of the constituency tests. I examine coordination and the "cherrypicking" accusation in some detail in the next section.

Like I said several times, I'm not interested in having an extended discussion about the merits of the different theories; the editorial issues can be settled without getting into all this. You are welcome to write as much as you want here but I won't be responding to these details; I would prefer if you spent your effort responding to 19:42, 7 January 2012 (UTC) and 17:56, 7 January 2012 (UTC) messages in the above section (specifically, discussing the Wikipedia-related editorial issues rather than the linguistic theory issues). rʨanaɢ (talk) 22:15, 7 January 2012 (UTC)[reply]
I agree with Rjanag in this regard. This article is not a journal article to put forward your favorite theory and argue for it. It is simply a descriptive encyclopedia, so the only things that we concern ourselves with are descriptions not persuasion. I do agree with Rjanag's specific comment that this needs to be presented as two competing theories. He doesn't think that's been emphasized enough, that the prevailing theory is PSG and that DG is only a minority view. The first sentence still sounds too much like a textbook. --Taivo (talk) 22:25, 7 January 2012 (UTC)[reply]
I've taken a shot at rewriting the paragraph to 1) clarify that these are competing theories and 2) spell out what the difference between the theories is as far as constituency goes. --Taivo (talk) 22:39, 7 January 2012 (UTC)[reply]
Taivo, I fully support your draft; it's better than what I produced. How about it, Rjanag? --Tjo3ya (talk) 23:00, 7 January 2012 (UTC)[reply]
I made revisions [1]. I didn't like "There are two competing theories on what constitutes a constituent"—both because there are more approaches to syntax than just phrase structure (whether it be GBT, Minimalist, or what have you) and dependency grammars, and because I would venture to guess these are not 'competing theories about what constituency is' but rather competing frameworks with more fundamental differences, and the different accounts of constituency fall out of those more fundamental differences. So most of my changes involved reframing that. The other change was removing the sentence "constituency tests will have different results". As I explained a few comments above, the sentence as it was worded was already incorrect, and anyway I don't feel that it's necessary to mention. If you guys feel it's necessary to say something about constituency tests (other than just flat-out saying "constituency tests are more consistent with the DG notion of constituency than the PSG notion"), we will have to discuss it more. rʨanaɢ (talk) 23:38, 7 January 2012 (UTC)[reply]
Good changes. --Taivo (talk) 00:29, 8 January 2012 (UTC)[reply]
I'm good with this. Whew! Can I post it now? --Tjo3ya (talk) 00:37, 8 January 2012 (UTC)[reply]

Extensive revision and expansion

[edit]

I am preparing an extensive revision and expansion of this article based on research that is being published very soon (tomorrow). I am working in my sandbox at my user page (User:Tjo3ya). I will periodically transfer my work to the main article. Comments are of course very welcome.--Tjo3ya (talk) 10:53, 10 April 2018 (UTC)[reply]

Constituent test: proform substitution (replacement)

[edit]

From the article: Proform substitution, or replacement, involves replacing the test string with the appropriate proform (e.g. pronoun, pro-verb, pro-adjective, etc.).... If such a change yields a grammatical sentence where the general structure has not been altered, then the test string is likely a constituent

This seems incomplete. Doesn't this test also require that the change preserve the basic meaning/essence of the original sentence? If such a change yields a grammatical sentence where the general structure and the general sense [or: the basic meaning/essence] has not been altered, then the test string is likely a constituent

Omc (talk) 23:19, 9 September 2018 (UTC)[reply]

"Clause element" and does not deserve "B"

[edit]

Is "constituent" the same as "clause element"? In any case, the article should reveal how "constituent" relates to "clause element". I disagree with the "B" ratig. The article is long but lacks a comprehensible introduction. It is only about "testing" and predominantly about EN. IMHO it needs substantial work. Article "Clause element" redirects to "English clause element". This is not great either. Taylor 49 (talk) 13:57, 7 July 2019 (UTC)[reply]

I agree. The article is good for what it is, but it is not a general introduction to constituents; it's an article about how to test for them in English. I'll downgrade it to C class on the grounds that it fails 2. of the B requirements.--Megaman en m (talk) 17:12, 7 July 2019 (UTC)[reply]
Thanks ... better info about "constituent" vs "clause element" is needed. Taylor 49 (talk) 06:22, 10 July 2019 (UTC)[reply]

Class project?

[edit]

Hello fellow editors: User:Dalbury, User:Jirka6, User:Rjanag, User:Thomas Tvileren, User:EdwardLane, User:Rikaido88, User:Tjo3ya, User:Taivo, User:Toddy1, User:Dnmaxwell! I am one of three editors (along with User:Syntactitian2 and User:Syntactician1) that will be working on this page for a final project for the next month. Our plans for editing will be posted soon. Syntactician3 (talk) 21:12, 28 March 2020 (UTC)[reply]

Hi! I'm also one of the editors that will be working on this page for a final group project! We are planning on reorganizing the order in which the constituency tests are listed, according to the three main categories: substitution, displacement, and deletion. If time allows, we would like to add trees to support each of the tests. Syntactician2 (talk) 21:39, 28 March 2020 (UTC)[reply]

Hi, I am one of the other editors of this page for the next little while. Syntactician1 (talk) 21:15, 28 March 2020 (UTC)[reply]

To the three Syntactiicians, is this part of a class project? If so, please ask your instructor to look at Wikipedia:Education program/Educators for instructions on how to coordinate class projects with Wikipedia. If this is some other project, could you please explain further what kind of project it is. - Donald Albury 23:05, 28 March 2020 (UTC)[reply]
Hi User:Dalbury. Yes, this is a final class project in our upper level Syntax course. Syntactician1 (talk) 04:29, 29 March 2020 (UTC)[reply]
It would be helpful for all involved if your instructor would set a course page, as described in Wikipedia:Education program/Educators. - Donald Albury 13:39, 29 March 2020 (UTC)[reply]
Thanks for the feedback, our instructor is following up on this now. Syntactician3 (talk) 23:26, 2 April 2020 (UTC)[reply]

Planned edits

[edit]

I am opposed to most of the planned edits for this article. From what I can see, the proposed additions are going to be associated narrowly with one particular approach to syntax. And the article will become too long and opaque. I have spent many hours polishing the organization and content of the article. I strongly suggest that the students working on this page abandon their project and return the content to what it was before they started. They should choose a topic that is better suited to their backgrounds. I will likely be stepping in soon myself to undo the initiated edits.--Tjo3ya (talk) 14:50, 7 April 2020 (UTC)[reply]

There are again signs of a desire to expand this article. I am opposed to a new large section presenting the tests again according to a different organization, since that would be redundant. A single brief section that simply lists and groups the tests according to the criteria mentioned (e.g. permutation, proform substitution, focusing) might be a helpful addition. I also think expanding the article to consider tests for constituents in other languages can be good. In any case, I would appreciate some discussion here first about the intended changes before they are made. If my request is ignored, I will remove the additions.--Tjo3ya (talk) 02:44, 15 April 2020 (UTC)[reply]
Our current plans are to develop the new section of the page that lists the tests by how they test for constituency. We plan on providing a brief description of each test, as well as a novel example to show how they apply. We are not aiming to be redundant with this list, but rather we would like to demonstrate how the tests can be classified into groups. We would appreciate your patience during this editing process. Syntactician3 (talk) 19:44, 15 April 2020 (UTC)[reply]
I will remove any redundancy concerning the discussion and illustration of the tests. I think the students should not continue with their planned additions. They are not in a position to contribute in an insightful and helpful way. They are parroting what they read in their textbook and hear from their instructor. --Tjo3ya (talk) 01:52, 16 April 2020 (UTC)[reply]

Edit war beginning

[edit]

There are a number of problems with the planned additions. I have removed these additions, yet they were immediately reinserted. To avoid an edit war, please respond here to my concerns. My main concern is that the planned editions will contain much redundant information. For instance, examples of the *do-so* test are already present above. More such examples are not needed. There are also lesser problems that demonstrate a lack of sophistication by the new editors, for instances, incorrect use of italics and the inclusion of statements about thematic roles, which belong in the article on thematic roles. If there is no response to my concerns here, I will again remove the additions. We will be in a full blown edit war. --Tjo3ya (talk) 06:23, 17 April 2020 (UTC)[reply]

Our goal is not to create conflict. We do not want an edit war and are not interested in disparaging your work. As our original edits were opposed, we now ask for your respect as we reorganize, develop, and edit the new section to this page. If there are indeed "lesser problems" of your concern (such as an improper use of italics), then please, by all means, let us know and we will fix these issues. As editors, we want to work together to improve the overall accessibility and quality of the page. Regarding thematic roles, a brief mention on this page should not be an issue. Providing readers with this information is necessary to ensure thoroughness in our explanations. Perhaps a link to Thematic relation would be beneficial. We will soon return to further develop this page and finish with the planned additions. If at that time you still believe the quality of this page has been degraded, you can go ahead and edit as you please. We just ask that you wait until our project is completed, until early May. Syntactician3 (talk) 21:23, 18 April 2020 (UTC)[reply]
Thank you for responding here. The main problem with your work is that you are repeating what is further above in the article. Such redundancy reduces the quality of the article, since it becomes too long and the organization is confused. Please do not attempt to re-add redundant information. I will remove it. I am in favor, though, of expanding the article in the direction of other languages. Some discussion, reflection, and illustrations of the value of the tests for other languages would be a helpful addition. Just a brief section that merely lists the tests in groups as to whether they are permutation, proform substitution, or ellipsis diagnostics would also be a positive addition.
What you are doing in general is problematic, though, so regardless of your efforts, I am going to be highly skeptical of the value or your work. You are students of linguistics, perhaps with only a few months, or perhaps just a year or two, of linguistics studies under your belt. You are anonymous; you are not self-identifying by establishing user pages, but rather you use aliases (i.e. Syntactician1, Syntactician2, etc.). Your edits therefore are easy to discount. The biggest problem, though, is that you are under pressure because your grade depends on your work in Wikipedia. The result of that pressure is poor quality. I have some empathy for your position insofar as the main problem here lies with your instructor. That person should not be assigning/allowing Wikipedia work for a grade! Please pass this message on to him or her. --Tjo3ya (talk) 02:43, 19 April 2020 (UTC)[reply]
I have a follow-up suggestion. Perhaps you could request of your instructor that you be allowed to write a term paper on the constituent unit instead of this Wikipedia effort. You could then freely produce the narrow account of constituency that is likely to please your instructor. --Tjo3ya (talk) 03:56, 19 April 2020 (UTC)[reply]

another aspect

[edit]

Could we not argue that the phrase could also potentially mean "Drunks [rather than, say, 'bums']..." ? Stjohn1970 (talk) 15:31, 18 October 2023 (UTC)[reply]