This   level-5 vital article is rated Start-class on Wikipedia's content assessment scale. It is of interest to the following WikiProjects:

Linguistics: Theoretical Linguistics High‑importance 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.LinguisticsWikipedia:WikiProject LinguisticsTemplate:WikiProject LinguisticsLinguistics articles High This article has been rated as High-importance on the project's importance scale. This article is supported by Theoretical Linguistics Task Force. Cognitive science ( inactive) This article is within the scope of WikiProject Cognitive science, a project which is currently considered to be inactive.Cognitive scienceWikipedia:WikiProject Cognitive scienceTemplate:WikiProject Cognitive scienceCognitive science articles

This article is or was the subject of a Wiki Education Foundation-supported course assignment. Further details are available on the course page. Student editor(s): Tuf80688.

Above undated message substituted from Template:Dashboard.wikiedu.org assignment by PrimeBOT ( talk) 22:11, 16 January 2022 (UTC) reply

• Should it not be mentioned that this is an obsolete theory?

• i don't know about other fields but this is still relevant in computer science

• This is still a relevant theory in the basic study of linguistics.

• it is still a releant theory and not a dead one although popularity has declined. it shold not be merged with gen. ling.

It is far from "obsolete", though there is much argument about its meaning in linguistics. Rainwarrior 02:29, 8 April 2006 (UTC) reply

Not at ALL obsolete. We're still actively working and researching in this paradigm!!! AndrewCarnie

It's not a paradigm, it's just structuralism that claims not to be structuralism. In fact it's a structuralism taken to the extreme. Chomsky's rebellion against his predecessors was mainly with behaviourism. The fact that the structuralist skeletons have never been cleaned out of the closets accounts for the tendency of generativists to claim attested diachronic changes (that is, the results of diachronic processes) as synchronic processes. I agree however that generativism is not obsolete, it is sadly receiving still a staggering amount of funding.-- AkselGerner ( talk) 20:53, 13 March 2008 (UTC) reply

I'm currently an MA Ling student at Case Western. Two specific problems with GG are 1) Poverty of stimulus argument breaks down when you examine usage-based corpora in depth and 2) the You Can't Get There From Here Problem, that is, GG provides no explanation linking child language use to adult language use to the supposed universal grammar. Ref, Tomasello- "Constructing a Language". It should be mentioned that GG is relevant in comp. sci and formal models, but for balance it must be stated that this theory has been largely discredited in terms of its ability to account for actual language production. —Preceding unsigned comment added by 75.185.187.54 ( talk) 04:07, 1 February 2010 (UTC) reply

Please ignore the last paragraph. The Cognitive linguistics ideologues must have filled his head with nonsense. —Preceding unsigned comment added by 93.139.20.191 ( talk) 17:42, 1 November 2010 (UTC) reply

Are you kidding? You are one of those chomsky haters for sure. This is 2017 and we are still actively working on this unique theory throughout the world with over 2000 researchers. Linguist91 ( talk) 07:24, 7 January 2017 (UTC) reply

• Looks like two article about the same subject. Merge Kerowyn 00:36, 19 January 2006 (UTC) reply

• JA: Keep separate. These are two distinct concepts. A generative grammar is a technical concept that originated within the field of study called generative linguistics, but it is often used in other fields, such as computer science, and it is also used to some perhaps modified extent in branches of or perspectives on linguistics that would not be described as generative linguistics by their participants. Also, its full development as a subtopic within any other article would eventually need to be broken out as a technical subsidiary, anyway. Jon Awbrey 13:50, 3 February 2006 (UTC) reply

• Don't merge. Generative grammar is used widely outside of linguistics. Rainwarrior 02:23, 8 April 2006 (UTC) reply

It's been a few months and we seem to have more against than for. Mostly it seems people are uninterested in comment. Anyhow, I've decided to remove the merge suggestion. Rainwarrior 07:53, 22 April 2006 (UTC) reply

Don't merge. -- IceHunter 04:57, 11 November 2007 (UTC) reply

• To merge or not to merge, it's preposterous to have a microarticle for the general theory and then a vast amount of text for the specific subtheory. Some material could be moved to the generative linguistics article, or generative grammar could be moved in it's entirety to that article as generative grammar is about 95% of generative linguistics anyway. Generative grammar in computer science would have to have it's own article anyway. Anyway, something should be done to flesh out the "generative linguistics" article, even if they are not merged.-- AkselGerner ( talk) 21:03, 13 March 2008 (UTC) reply

I agree. The Generative Linguistics article seems to be little more than a stub. If there is more to say about it apart from a technical description of various flavors of generative grammar, then it needs a lot of expansion. But if, as I believe to be the case, there is not much to add there, then the Generative Grammar page should be incorporated there. I also agree that there should be a different page for a CS audience. Ailun ( talk) 16:18, 11 August 2008 (UTC) reply

• Don't merge, the linguistic theory is broader than just the grammar. Given that all but the first vote was against merging, I'll remove the merge tag. linas ( talk) 16:57, 20 August 2008 (UTC) reply

On July 9th, 2006, 62.192.140.219 made changes to this page which, even if new information was added (it is difficult to tell), removed several pieces of useful information, pictures, wiki links, and formatting; furthermore, it was written in a very poor style (it reads like an undergraduate essay). I have reverted this edit. - Rainwarrior 17:59, 9 July 2006 (UTC) reply

It's not that I don't think that there may be some candles of truth in this article's description of the theory presented, BUT for the most part the page is written in such a fashion that it sounds like a lot of BS masquerading as intelligent discourse -- sort of what you would present to a college professor in an oral exam if you wanted him to think you knew what you were talking about -- but really didn't cause instead of studying you had stayed up too late partying the evening before. Fungible 09:01, 7 March 2007 (UTC) reply

Really? To me it sounds technical, if a little overdone for an encyclopedia. Most BSing I've done on college papers, at least, hasn't included appropriate references to very technical and specific facets of the ideas being discussed. Although I do think this article needs improvement and better organization, in particular w.r.t. its subject matter's applications in other areas. Space Dracula 05:22, 18 June 2007 (UTC) reply

It IS too technical and waffley - the whole idea of an encyclopedia is to educate - not to reconfirm what someone already knows. Don;t forget that although you can have technical explanations, the first paragraph should sum up first - then you can have technicalities after. I came onto this article out of interest (from Chomsky entry) and I have no idea of what Generative grammar is after reading this article as I am not a linguistics student Gruffster 20:30, 11 July 2007 (UTC)

Same. I am non the wiser as to what generative grammar actually is. —Preceding unsigned comment added by 88.106.30.140 ( talk) 23:01, 22 October 2007 (UTC) reply

Some linguists go so far as to claim that the set of grammatical sentences of any natural language is indeed infinite.

Isn't this obvious, though? You can make an infinitely long sentence. All you have to do is stack clause upon clause upon clause. I doubt there is a language where such a thing isn't possible, because this sort of flexibility with clauses is an essential feature of language. It logically follows that if you can generate an infinitely long sentence, then there is an infinite number of possible sentences. - furrykef ( Talk at me) 21:53, 19 July 2007 (UTC) reply

Sorry, this statement is not true. If you look at formal definitions of rewrite grammars, all generated sentences are finite. Of course you can generate infinitely many of them, groing longer and longer, but the sentences itself are finite. See my grammar theory text book. Some footnote in the intro chapter and also in the discussion part of the book. Chomsky never claimed that one can have infinitely long sentences. The question is whether one should assume infintely many. This is also disputed. See GT textbook Section 13.1.8. StefanMülller ( talk) 07:59, 12 April 2022 (UTC) reply

IINM, the consensus is that while sentences have no bounds on their lengths, an infinite string is not a sentence. (For one, its grammaticality would be indeterminate.) However, the set of all sentences of unbounded length could be put into a one-to-one correspondence with, say, the infinite set of natural numbers. jackaroodave 69.207.251.137 21:39, 8 August 2007 (UTC) reply

• Seems to me that this page needs some work to make it more balanced. It seems to give little or no indication that there have ever been legitimate challenges to Chomskyan generative grammar. And although it's nice that it mentions other generative frameworks, it proceeds to provide an article about only one. At least LFG etc. should be included in the history section. On the other hand, if it is considered appropriate to only provide links to articles on LFG, etc, then this should be made a disambiguation page and almost all content here should become a page with a more specific title. Also, it seems to me very difficult to evaluate or even grasp Chomskyan generative grammar (at least) without a fairly thorough treatment of its fundamental assumptions and arguments (POS, etc.) along with, whenever possible, links to alternative arguments or approaches. I would try to do some of this myself but I'm pretty certain there are people here much better qualified than I. PS: I don't hate generative grammar. :) Ailun ( talk) 16:36, 11 August 2008 (UTC) reply

  I agree. The article starts with innateness claims. Most non-Chomskyan frameworks do not assume this anymore. A generative grammar in one definition of this term is a grammar generating a set. This is quite formal and applies to many grammars. Pullum & Scholz call this generative-enumerative. No innatenes-claims implied. HPSG and LFG and even some variants of CxG consider themselves generative grammars but in the sense of Chomsky 1965: A grammar of a language purports to be a description of the ideal speaker-hearer's intrinsic competence. If the grammar is, furthermore, perfectly explicit - in other words, if it does not rely on the intelligence of the understanding reader but rather provides an explicit analysis of his contribution - we may call it (somewhat redundantly) a generative grammar.

  One could say that many researchers equate the term with Chomskyan mainstream linguistics. This is connected with innateness. StefanMülller ( talk) 08:06, 12 April 2022 (UTC) reply

In my opinion the topics of transformational grammar and generative grammar would best be handled in one article. There's content in the TG article that should really be covered here (e.g. e-language, i-language). Perhaps a Chomskyan could help us out here. ---- Action potential ^{t c} 12:17, 8 September 2008 (UTC) reply

This is a mistake. Generative grammar and transformational grammar are different things. Transformational grammar is a kind of generative grammar. But so is relational grammar, Lexical-Functional Grammar, Minimalism-linguistics, Government and Binding theory, and Head-driven Phrase Structure Grammar. These are not transformational grammars, but they are generative grammars. (A generative grammar is one in which the set of grammatical sentences are generated (in the mathematical sense of the term, not the psychological sense) via a set of rules and constraints. A transformational grammar is one that uses a particular kind of rule, known as a transformation to do this. Please do not merge. AndrewCarnie ( talk) 17:29, 8 September 2008 (UTC) reply

(PS, I do agree that the TG article needs some work, if I have time later this semester I'll do my best) AndrewCarnie ( talk) 17:31, 8 September 2008 (UTC) reply

Thanks. I've removed the merge proposal. ---- Action potential ^{t c} 00:26, 9 September 2008 (UTC) reply

In http://en.wikipedia.org/wiki/Generative_grammar#Frameworks the article says "The term generative grammar has been associated with at least the following schools of linguistics: " and then lists, inter alia, Categorical Grammar as an instance. However, in the article about syntax, they are described as something completely different ( http://en.wikipedia.osyrg/wiki/Syntax#Generative_grammar and http://en.wikipedia.org/wiki/Syntax#Categorial_grammar). This seems to be incoherent, at least to the naive reader. An explanation might rectify this, otherwise Categorical Grammar should be removed from the list.

92.233.157.159 ( talk) 23:20, 18 October 2008 (UTC) reply

Please keep Categorial Grammar in this list. It is true that there to readings: Generative grammar (with a lower-case grammar) is any grammar with recursive rules which enable it to generate an infinite number of sentences from a finite set of symbols (say, words). This is also true for TAG, HPSG, Categorial Grammar and other grammars listed under monostratal grammars in the article. On the other hand, Generative Grammar (with a capital second G) is a collective term for all the Generative grammar conceptions Chomsky proposed. In the past, these were usually transformational grammars, but recently, Chomsky seems to start listening to psycholinguists and perhaps computational linguists who have never believed in transformations. In summary, transformational grammar used to be the better term for Chomsky's Generative Grammar, but since the Minimalist Program it probably is no longer. Perhaps Principles-and-Parameters approach is the better term for Chomskyan Generative Grammar. -- Thüringer ☼ ( talk) 01:04, 6 February 2009 (UTC) reply

The first paragraph ends with the phase "the morphology of a sentence." Looking at the linked article on morphology (linguistics), I find that morpohology is all about the forms of words, not sentences. Therefore, the phrase "the morphology of a sentence" seems defective, unless the article about morphology is defective instead. Looking at the history of this article, I see that this phrase has been standing for over a year and a half, but nonetheless, it is a little confusing. (As it stands right now, I still don't know what "morphology of a sentence" exactly means, if anything.) Could someone who knows the subject well clear this up?

It may be of interest to note that the editor that originally composed this sentence wrote it as:

In most approaches to generative grammar, the rules will also predict the semantics and morphology of a sentence.

This edit on 21:28, 21 January 2008 was made by G.broadwell, who on that day apparently rewrote (and overall substantially improved) the introductory section. The sentence was modified about two months later to remove "semantics and"; here is the edit history record:

13 March 2008 AkselGerner (talk | contribs) m (8,915 bytes) (Chomsky himself said that Semantics has no place in linguistics. Sentence semantics cannot be calculated.) (undo)

Since one claim of the original sentence (it appears) was incorrect, the rest of the sentence can be questioned. From what I knew about generative grammar before reading this article, I can see why a generative grammar might indicate a lot about the morphology of individual words (if "of words" is not redundant here) in that grammar, but that is not clearly what the sentence says, and certainly a reader with no prior exposure to generative grammars could not be expected to see that. A generative grammar necessarily defines the structure of sentences, so it is possible that what "morphology of sentences" means to say is redundant and underly definite, like saying, "A cat is a carnivorous four-legged furry mammal. Most cats are quadrupedal and warm-blooded."

71.242.6.231 ( talk) 12:49, 1 October 2009 (UTC) reply

Hey, how about we explain how the deep structure of a sentence can be represented in a variety of ways using contrasting tree diagrams? Example (do tree diagrams for the following sentences): "The dog ate the bone." Vs. "The bone was eaten by the dog." Is this a good idea for this page? Xetxo ( talk) 18:19, 7 December 2009 (UTC) reply

Could someone go through the article of Ad Neeleman? I came across it referencing unreferenced BLPs. the article states that Neeleman is part 'of the tradition of generative grammar' (but I couldn't find him here). Secondary sources establishing notability and extra information however would be welcome. Should he not pass WP:notability (academics), a deletion proposal might be in order; but I am not capable of adding or further judging this... L.tak ( talk) 00:49, 15 December 2010 (UTC) reply

I think we should study to have a historical sense. — Preceding unsigned comment added by 119.154.125.226 ( talk) 12:48, 31 August 2011 (UTC) reply

I noticed that when I create a pdf version of this article, the subsubsection 1.1.6 is not listed. I guess it is because it only contains a link to the main article on Minimalist program. Now, this whole section acts as a list of subsequent stages of Chomskian linguistics, so the title of this subsubsection should definitely be included in the offline version. How do I achieve that? Anša ( talk) 23:36, 16 January 2012 (UTC) reply

In view of the fact that the generative school of thought is highly debated in linguistics, it is a bit surprising there is no section devoted to the criticism of the theory. I am no specialist in the history of the critique, but it would be useful if someone added such information. It is also somewhat misleading to include some theories such as Dependency Grammars as part of Generative Grammar, when from the outset some of them were designed as a critique of the generative view. בוקי סריקי ( talk) 16:06, 26 February 2014 (UTC) reply

Absolutely agree with the above. Why should this article be immune from the criticism which would routinely be a part of any other article discussing a current, and heavily contested, theory. As well, consider that any theory that changes itself so often – six major revisions/adaptations in thirty years or so – is probably misconceived in the first place. For a detailed critique of the inability of any of Chomsky's versions, or those of his acolytes to deal adequately with semantics, see John Ellis, Language, Thought, and Logic (Northwestern UP, 1993), who argues that the whole attempt to divorce syntax from semantics, stemming from Chomsky's initial Standard Theory, is a basic error which no amount of reconfiguration of the theory can fix (Chapter 8). Looked at this way, the swerve into politics might be interpreted an attempt to avoid dealing with the problems of the theory. Note also that the debate in the section above, "A small point of detail," circles around exactly this issue, and fails to come to conclusion, mostly because it can't, except to repeat Chomsky' obiter dicta, that synatax and semantics can have no connection. Theonemacduff ( talk) 06:18, 2 June 2014 (UTC) reply

This has to be changed. Chomsky states that's it not a theory but a program, hence the name. "Chomsky presents MP as a program, not as a theory, following Imre Lakatos's distinction." - Wikipedia. Why was it written on this page that it's a theory? When jumping to MP page it says that it's not a theory. — Preceding unsigned comment added by Senethys ( talk • contribs) 10:34, 2 December 2014 (UTC) reply

In the subsection "Standard Theory (1957–1965)" we have the following sentence and no other publication cited for this interval: "The so-called Standard Theory corresponds to the original model of generative grammar laid out in Chomsky (1965)." To me it is not clear why the time interval should start at 1957, if there hasn't been any publication before 1965. If it is meant that Chomsky and other scholars only started working at that theory in 1957, then this should be made clear. If this is not meant, then there should be cited a publication in 1957. Indeed, Syntactic Structures was published in 1957.

Garrafao ( talk) 16:08, 29 May 2015 (UTC) reply

It is somewhat surprising that this page describes none of the criticisms of the generative approach, which span the beginning (e.g. Hockett, Robert Hall) until now (e.g. Dan Everett). Of course detailed discussion of critics at every moment wouldn't be appropriate, but presenting the whole school as essentially unchallenged is certainly POV. Tibetologist ( talk) 06:06, 7 February 2016 (UTC) reply

Hello fellow Wikipedians,

I have just modified one external link on Generative grammar. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:

• Added archive https://web.archive.org/web/20100919021754/http://chomsky.info/articles/195609--.pdf to http://www.chomsky.info/articles/195609--.pdf

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

This message was posted before February 2018. After February 2018, "External links modified" talk page sections are no longer generated or monitored by InternetArchiveBot. No special action is required regarding these talk page notices, other than regular verification using the archive tool instructions below. Editors have permission to delete these "External links modified" talk page sections if they want to de-clutter talk pages, but see the RfC before doing mass systematic removals. This message is updated dynamically through the template {{ source check}} (last update: 18 January 2022).

• If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
• If you found an error with any archives or the URLs themselves, you can fix them with this tool.

Cheers.— InternetArchiveBot ( Report bug) 04:50, 9 January 2017 (UTC) reply

Whatever about the verb phrase has been said is totally abusive. Nothing is clear-cut. A short of Greshm's rule is applied on the modern learning. Under the name of Generative Grammar every discipline of English Grammar has been polluted. Your so-called Generative Grammar is rubbish. Your Chomsky is rubbish. Birbal Kumawat ( talk) 18:37, 24 December 2017 (UTC) reply

I undid edits made by User:Botterweg14 on March 3, 2020 because these consisted of the deletion of most of the sourced information on the page. Wikipedia is meant to be a site for information on science, and this includes descriptions and criticisms. It is useful to know the background of generative grammar and its differences from the previous structuralist models to understand what it is all about. It is likewise necessary for people to be able to find out whether scientific claims made by generative grammarians are founded on research. While some restructuring is possible, it is not possible to remove well-sourced information only because it is uncomfortable. Weidorje ( talk) 08:37, 4 March 2020 (UTC) reply

I've restored my edits, which did not remove most of the sourced information on the page. Particular points:

• The relation to earlier (and later) approaches is important and valuable, and I would support expanding the section "Historical development of models of transformational grammar" however I think it's undue weight to focus on that in the lede.
• I removed several statements which seemed to suggest that a primary claim of generative grammar involves the object originating inside the verb phrase. Generative grammar isn’t a theory of whether objects originate inside verb phrases. That’s just a minor detail of many particular analyses with the framework (and one shared with plenty of non-generative analyses too.) So making it a major focus of the article is misleading.
• The “lack of evidence” heading is inaccurate. The text there describes some challenges to the foundational assumptions of the approach, but that’s different from a lack of evidence. My edits there were an attempt to clarify actual criticisms of the approach.
• Some neurolinguists have indeed failed to find reflections of generative-style theories in the wetware. But there’s also plenty of neurolinguistic work that builds on generative grammar, including ERP studies. So I think it’s misleading to outright say things like “it is stated that generative grammar is not a useful model for neurolinguistics”.
• “It was later found, however that Chomsky’s linguistic analysis was inadequate”. This is neither a consensus view nor is it an accurate reflection of what Pullum says in the cited article. In the cited article, Pullum argues that poverty of the stimulus arguments aren’t as compelling as they’ve been made out to be, and that there are reasons to think that future work might push in the other direction. That’s a far cry from Chomsky’s personal linguistic analysis being found to be inadequate.
• “Generativists also claim that language is locked inside its own mind module with no connection to other types of information processing” Again, this is inaccurate. Generative grammar assumes a domain-specific module for grammatical knowledge, but no generative research would say that it isn’t connected to other types of information processing. Even in that Chomsky Hauser Fitch article (a relatively strong view about modularity) talks about how the “faculty of language in the narrow sense” hooks up to the sensory-motor system and the conceptual system.

Botterweg14 ( talk) 14:07, 4 March 2020 (UTC) reply

These are all false statements, based on the souces. Since Botterweg14 is not adding anything to the discussion – in fact, he's only broken previous links – I will keep his changes deleted. Weidorje ( talk) 15:08, 4 March 2020 (UTC) reply

Botterweg14, can you rather add a "quote needed" tag or something wherever you'd like it, so this does not develop into an unnecessary edit war. The point from Wikipedia is to give people truthful information, and whatever it is, it must be found here. We're not trying to hide anything. Weidorje ( talk) 15:24, 4 March 2020 (UTC) reply

FWIW:

• Quoting from Chomsky, Hauser, Fitch (2002): “We assume, putting aside the precise mechanisms, that a key component of [the faculty of language in the narrow sense] is a computational system (narrow syntax) that generates internal representations and maps them into the sensory-motor interface by the phonological system, and into the conceptual-intentional interface by the (formal) semantic system.” So the claim isn’t that the mental architecture for language is walled off from other mental systems. In fact, it’s just the opposite— that part of that architecture includes a component which connects up other cognitive systems. Modularity comes from the fact that the computations performed by that component don’t depend on other cognitive systems. So other approaches might try to explain verb inflection and politeness using a single mechanism, but the generative approach would split them up.
• Quoting from Pearl (2019): “In one lively debate, Pullum and Scholz (2002) noted that no one was saying how much data was enough, even though many people were quite convinced PovStim was occurring for various case studies. Legate and Yang (2002) (L&Y2002) took up that challenge and offered one approach to specifying how much data was enough for children to make the constrained generalizations they did” Point being, Pullum and Scholz (2002) didn’t show that it’s all over for poverty of the stimulus arguments. Rather, they raised serious objections which are the subject of ongoing debate.
• For more on the wealth of generative neurolinguistic work, see any of the work coming out of the neurolinguistic labs at NYU and Maryland (note the EEG studies).
• To see that the structural position of objects within VP's is a minor detail in generative syntax, see any standard textbook for instance this one. Point being, sure, that's probably how most research in generative syntax treats objects but it's a minor detail, not a central feature of the approach.

Botterweg14 ( talk) 17:32, 4 March 2020 (UTC) reply

So what's the innate structure then, if not the object placed into the VP? Note that even if there will be further (ubstantiated) claims, this does not mean the first one should not be evaluated in its own right. We're not writing an essay. Weidorje ( talk) 06:44, 5 March 2020 (UTC) reply

This comment still doesn't address any of my concerns, so I'm reverting the article to the previous version. Botterweg14 ( talk) 13:57, 5 March 2020 (UTC) reply

Botterweg14, your concerns do not need to be addressed as such. What you write up above is not based on the cited sources, it's something different. If there's solid evidence somewhere, let's have a look at it. Weidorje ( talk) 14:32, 5 March 2020 (UTC) reply

Each of my concerns above comes with a cited source. Moreover, part of my concerns come from the fact that the current article text does not accurately reflect the sources already cited in the article, e.g. the Pullum and Scholz citation. Botterweg14 ( talk) 14:37, 5 March 2020 (UTC) reply

I'd like to request comment on the "lack of evidence" section as well as the lede. I am concerned about their current form for reasons I have laid out above. Botterweg14 ( talk) 17:47, 4 March 2020 (UTC) reply

For my part I would like to explain that there is no NPOV issue because there is no evidence for the claims made by generative grammar. There are no brain modules, no Chomsky gene, and the object is not in the VP, according to research. I am fully aware that these are more or less actively studied as if they were real, but doing such research itself does not prove the underlying assumptions. Generative grammarians feel that the framework has not been rebuffed properly because there is always a slight possibility the claims are true. This is exactly what the main man Noam Chomsky is saying, which only means generative grammar needs to be treated differently from sciences. Like neuro-linguistic programming, it appears much like a cult of personality and a quasi-science. Weidorje ( talk) 06:37, 5 March 2020 (UTC) reply

• There really isn't much evidence. I'm not sure it's a good idea for the section to be titled this, rather than just, say, "Evidence". However, the lead is required to accurately summarize the article, so the lack of solid evidence backing this theory will have to be in there. That said, I think I see for WP:OR going on in this section, e.g. claims that the theory "has been deemed" this or that, which looks like Wikipedia trying to declare WP:TRUTH. The material does need work, and it cannot cite primary sources, like primary-research papers in journals, for any kind of WP:AEIS claim, but only high-quality secondary ones like literature reviews, and recent books from subject-matter experts and reputable publishers. Weidorje is correct that there'a cult-of-personality problem going on here (in this sub-sub-field, in the real world), just as with NLP and psychoanalysis. Nevertheless, there's a growing acceptance of the idea that there is something innate about human language and that it arose probably somewhere around 75K to 40K years ago (estimates vary). But that doesn't equate exactly to the generative grammar model, it just has a central idea in common. I think we'll find in 20-ish years that g.g. has effectively been replaced by a new model with some aspects in common but shedding a lot of the presumptions and untestables that mar g.g. The entire field of anthropology is becoming more interdisciplinary, including a lot of number-crunching and genetics and other less "interpretative" approaches, and we can expect this shift to overtake linguistics (an anthropology subfield), too.  —  SMcCandlish ☏ ¢ 😼  13:49, 5 March 2020 (UTC) reply

I take it you're talking about Schwarz-Friesel's article. It is a review article and not original research at all. It explains how the brain works, pointing out that there are modules in a certain understanding: language is processed differentlly from other types of information, and in somewhat different locations around the brain. This, however, is not in line with what the generativists are claiming. The passage is too long for the reference, so I add it here: "As in the past, two antagonistic camps dominate the field of cognitive linguistics: the holistic CL1 and the modular CL2. Basically, the modular CL2 position aims at a further development of the generative linguistic theory. Unfortunately, CL2 supporters are also frequently mentally closed regarding the examination of their working premises and the consideration of data from different theoretical accounts. Many CL2 researchers disregard scientific results which are not presented according to formalist rules. The Institute for Cognitive Linguistics (CL2) in Frankfurt defines cognitive linguistics as follows: ‘‘Cognitive linguistics is the branch in the modern science of language that deals with the biological foundations of language and describes them with the help of genetically determined principles.’’8 This sounds like neuropsychology. If so, it should deal with synapse activity, neurotransmitters and neuronal columns in the brain. But instead, CL2 scholars (not only in Frankfurt) usually produce formalist analysis in the Chomskyan tradition. Frequently, formalizations are offered instead of explanations. These formalizations are in many cases nothing else than the translation of one form of description into another one, without any heuristic or explanatory advantages attached. Often, the formalistic ‘explanation’ is little more than pseudo-science, with a wide gap remaining between the theoretical assumptions and the concrete applied methodology. Summing up the ‘state of the art’ in cognitive linguistics, one is forced to remark that CL1 and CL2 behave exactly like the famous input systems (postulated by Fodor (1983)): ‘domain specific’ and ‘informally encapsulated.’ They work autonomously and produce, unfertilized by other scientific resources, a ‘shallow output.’ According to Fodor modules are domain specific processing systems of the mind (they cope with a restricted class of inputs). The outputs of a module are shallow in the sense of being nonconceptual. Hence, shallow means that they have relatively simple outputs. They generate information of a certain sort, but they do not deal with thought. Mandatory in their processing they ignore the input from other domains. Finally, modules are isolated from the remaining domain of cognition (i.e. informationally encapsulated), since they cannot be guided by information at higher levels of processing." Weidorje ( talk) 15:01, 5 March 2020 (UTC) reply

I added this source. Botterweg, can you stop editing this article. Your changes are not constructive. Weidorje ( talk) 15:28, 5 March 2020 (UTC) reply

In this section, I'm carrying down my concerns about the article so that they're all clearly stated in one place.

• This edit removes a claim which is not in the source. The source does not make this claim about generative grammar, but rather a sub-branch of a different framework, namely cognitive linguistics. Unless someone else can point out an explicit place where this article makes this claim about generative grammar, I think it needs to be removed. Botterweg14 ( talk) 21:36, 5 March 2020 (UTC) reply

Botterweg14, you're virtually looking at it. It's here, right above in the previous section. I suppose I'll need to put back the pseudoscience claim, then. Weidorje ( talk) 23:07, 5 March 2020 (UTC) reply

The claim in that quotation is about a sub-branch of cognitive grammar which she calls "CL2". Botterweg14 ( talk) 23:19, 5 March 2020 (UTC) reply

CL1=Lakoff–Langacker, CL2=Chomsky. Weidorje ( talk) 23:37, 5 March 2020 (UTC) reply

According to the source, CL2 is the branch of cognitive linguistics introduced by Bierwisch and Lang which attempts to bridge the gap between CL1 and generative grammar: "From its beginnings, cognitive linguistics was not a homogenous field but rather an agglomeration of diverse approaches under a common title. In the United States, Cognitive Linguistics (CL1) emerged as a result of a radical break from the dominant generative grammar theory (GGT) (Lakoff, 1987; Langacker, 1987, 1999), whereas in Germany and most of Europe, linguists tried to build a bridge between the two approaches (Bierwisch, 1983, 1987; Bierwisch and Lang, 1987), focusing on the mental character of the language system (CL2)." Botterweg14 ( talk) 00:01, 6 March 2020 (UTC) reply

• This edit removes content that is not in the source. As mentioned in the edit summary, the term "generative linguistics" does not occur in the source. Unless someone else can point out an explicit place where this article makes this claim about generative grammar, I think it needs to be removed. Botterweg14 ( talk) 00:01, 6 March 2020 (UTC) reply

True, although I didn't remove it but changed it: "Neuroscientific studies using ERPs have found no evidence for the claim that human mind processes grammatical objects as if they were placed inside the verb phrase.[11] The validity of generative grammar was called into question in 1993,[11] and since then the scientists who carried out the research have no longer made use of the generative model, instead relying on a simple, more general processing model.[12]" Weidorje ( talk) 23:38, 5 March 2020 (UTC) reply

Thank you, I would regard this as an improvement. However, I'm still not seeing the relevance of this study to generative grammar as a whole. Can you clarify which parts of the 1993 article show that? Also, is your point that the later article simply doesn't use generative ideas or that it is arguing that generative grammar shouldn't be used? Botterweg14 ( talk) 00:12, 6 March 2020 (UTC) reply

• This edit does not remove any content or any references. It is intended to (1) clarify what the cited article is actually about, since its main claim seems to be that subjacency effects are processing effects and (2) clarify that this is a particular claim from particular studies, and not a universal consensus in neurolinguistic research. If I'm misunderstanding the papers, I'd be glad to have that pointed out. As for the second point, we would need a separate source to show that, since there is no such claim in the papers. Botterweg14 ( talk) 21:36, 5 March 2020 (UTC) reply

That sounds like too much jargon to me. What happened with the findings was that they were consistant with a simple approach with mere word semantics. The point is that there was nothing to indicate an unexpected/innate element. This would have been found if there had been an easiness of processing elements in the VP even when they are placed far apart. What was found was that the more distant the elements, the more it loads on working memory. The Chomskyan answer is of course that this doesn't prove anything, which is correct. The fact is that any claim, such as that the object is actually (cognitively) in the VP, needs to be proved in order for it to be scientific. This is what the researchers looked into in 1993, but they found no indication or evidence, and consequently abandoned generative grammar. Of course we could cite any book or study and state that there's no evidence for the claim, but this particular article is interesting; besides, they don't normally publish zero results many times. At any rate, Wh-islands were studied, and no specific evidence was found for generative grammar. The new formulation is nonetheless clearer that this view comes from a selected group of researchers. More studies can be added. Weidorje ( talk) 23:58, 5 March 2020 (UTC) reply

I don't see anywhere in the paper where the authors (1) make any claim about the position of objects inside/outside of VP's (2) explicitly say that they abandon generative grammar or (3) challenge the existence of innate grammatical knowledge. All I see is the claim that subjacency isn't a part of innate grammatical knowledge, if it exists. When I have my researcher hat on, I might wonder whether this points to a shrinking role for UG and that could have been the authors' intent for all I know. But for a Wikipedia article we have to stick to reporting what's actually in the citation. Botterweg14 ( talk) 19:28, 7 March 2020 (UTC) reply

• This edit does not remove any content or any references. It is intended to (1) clarify what UG is (2) clarify what poverty of the stimulus arguments are (3) more specifically describe P&S's criticism (4) clarify that this is the subject of ongoing work. My edit adds material, including a citation of a paper which summarizes ongoing work on this topic. Botterweg14 ( talk) 21:36, 5 March 2020 (UTC) reply

I'll have a look into that tomorrow. Weidorje ( talk) 00:02, 6 March 2020 (UTC) reply

• This edit is intended to clarify that the discussion is about the quality of the evidence. Poverty of the stimulus arguments are intended as evidence, whether or not they are high-quality evidence. Botterweg14 ( talk) 21:36, 5 March 2020 (UTC) reply

So now you're on the next paragraph which is about the Chomsky gene, not POS. Weidorje ( talk) 00:06, 6 March 2020 (UTC) reply

Okay, sure, but then I think that paragraph should begin “No evidence has been found for a single specific language gene as hypothesized by XXXX” where XXXX is whatever generative work made such a hypothesis. My objection to the current text is that it seems to imply that there’s no evidence for innate grammatical knowledge which is a slightly different claim. — Preceding unsigned comment added by Botterweg14 ( talk • contribs) 18:23, 7 March 2020 (UTC) reply

• I think the section about the FOXP2 gene would need to be removed unless its importance to generative grammar can be explained and backed with sources. I don't know anything about this particular topic, but a quick glance at the Chomsky and Berwick book seems to suggests that they regard the gene as relating to memory, not the “faculty of language in the narrow sense”. If that's correct (and it might well not be) it would seem to not be relevant to the article. [[[User:Botterweg14|Botterweg14]] ( talk) 21:36, 5 March 2020 (UTC) reply

Generative grammar is the study of an innate structure. How is it innate? According to Chomsky, it is caused by a genetic mutation. However, there is no evidence for the genetic claim. The FOXP2 gene is the closest one can get, but of course we could just take any reference book on genetics and state that there is no evidence for the innatism claim. Weidorje ( talk) 00:06, 6 March 2020 (UTC) reply

Okay, so if I'm understanding the claim is that research so far has failed to identify a single gene responsible for language and that FOXP2 was one candidate which was subsequently ruled out. If that's the argument, I think it could be made clearer in the text. Right now, it reads as if the FOXP2 gene was part of the original empirical motivation for innateness. — Preceding unsigned comment added by Botterweg14 ( talk • contribs) 00:18, 6 March 2020 (UTC) reply

• Quoting from Chomsky, Hauser, Fitch (2002): “We assume, putting aside the precise mechanisms, that a key component of [the faculty of language in the narrow sense] is a computational system (narrow syntax) that generates internal representations and maps them into the sensory-motor interface by the phonological system, and into the conceptual-intentional interface by the (formal) semantic system.” So the claim isn’t that the mental architecture for language is walled off from other mental systems. In fact, it’s just the opposite— that part of that architecture includes a component which connects up other cognitive systems. Modularity comes from the fact that the computations performed by that component don’t depend on other cognitive systems. So other approaches might try to explain verb inflection and politeness using a single mechanism, but the generative approach would split them up. Botterweg14 ( talk) 21:36, 5 March 2020 (UTC) reply

• The current text misstates the generative idea about modularity ("Generativists also claim that language is placed inside its own mind module"). No generative research would say that language is completely isolated from other cognitive systems. Even the Chomsky Hauser Fitch article (a relatively strong view about modularity) talks about how the “faculty of language in the narrow sense” hooks up to the sensory-motor system and the conceptual system. To be clear I'm completely in favor of including criticism of generative claims about modularity-- my point is just that the article should accurately represent those claims. Botterweg14 ( talk) 21:36, 5 March 2020 (UTC) reply

• I am not opposed to including information about how research in generative linguists has responded to these challenges. However, I don't think these replies are fairly represented by the current sentence "Chomsky has answered the criticism by emphasising that his theories are actually counter-evidential". In the cited source, Chomsky talks repeatedly about evidence, saying for instance “you would need evidence for the special assumption that you can only have two things attached to a head”. The basis of the passage in the article seems to be Chomsky's notion of the "Galilean style" where all scientific theories appear to be refuted initially, and it requires deep analysis of particular phenomena to figure out which refutations are real. I strongly suggest removing or rephrasing this passage so that it accurately represents Chomsky's personal position. I also strongly suggest that we include discussion of work that responds to the other challenges discussed in the current article text, e.g. Legate and Yang on PoS arguments, generative-inspired ERP research from Pylkannen, Phillips, and others etc. Botterweg14 ( talk) 21:36, 5 March 2020 (UTC) reply

I'd rather leave these all out of the critical section and focus on what the critical sources say. Weidorje ( talk) 00:40, 6 March 2020 (UTC) reply

If you’d like to remove that paragraph, I’d agree to that. (I do think the article would be stronger if it clearly laid out the back-and-forth on these issues, but I don’t have time at the moment to do that well.) Botterweg14 ( talk) 18:21, 7 March 2020 (UTC) reply

• In response to Weidorje's question about what the innate language-specific structure is, that's the question that research in generative grammar tries to answer. If earlier generative work hypothesized that VP-internal objects are part of that structure, I'm certainly not against including that in the article. However (1) we would need sources establishing specifically that this part of the structure was posited as innate and (2) it should be made clear that this idea is a particular idea proposed within the framework and not one of the constitutive principles of the framework. For sources regarding (2), see e.g. the Chomsky interview that's linked in the current article text or Chomsky, Hauser, and Fitch (2002), which proposes the hypothesis that innate knowledge of language amounts to a very abstract structure building operation plus some conditions on how the resulting structures play nice with other cognitive systems that link to it. Botterweg14 ( talk) 21:36, 5 March 2020 (UTC) reply

• Could I ask you to revert this edit, per our agreement we reached with User:Jayron32 not to edit until we reach consensus? As for your request, I would be happy to avoid mentioning this discussion to my generative colleagues if you have concerns about brigading. Though at some point in the future I do think a full-on generativist should have a look at the article. Botterweg14 ( talk) 00:01, 6 March 2020 (UTC) reply

• My last comment for tonight-- I think the discussion above would be much easier to read if it was in a bulleted list form but when I tried to reformat I kept screwing up the syntax. If you agree and you're better with markup than me, I'd appreciate you reformatting it. Botterweg14 ( talk) 00:23, 6 March 2020 (UTC) reply

Done. Ok, how's this:

"Neuroscientific studies using ERPs found no indication that human mind processes grammatical objects as if they were placed inside the verb phrase.[11] The validity of the innatism claim was called into question in 1993,[11] and since then the scientists who carried out the research have no longer made use of the generative model, instead relying on a simple, more general processing model.[12]"

Here, of course, it is not said that VP is anyhow related to the innatism claim. 'Validity': a claim needs evidence in order to be proven valid; a simpler claim is chosen as 'valid' per Occam's razor. As you pointed out yourself, the 1993 article discusses GG, but not the 2015 article where the research remain 'agnostic' on specific architectural designs, as a reference to the 1993 article. Weidorje ( talk) 00:38, 6 March 2020 (UTC) reply

Thanks for doing the bullet points! I've responded to the subjacency/VP/ERP points up above. Botterweg14 ( talk) 19:31, 7 March 2020 (UTC) reply

Issue 1: linguistic evidence

Goedemorgen, Botterweg14 and y'all! So, we are now making a revision of the Criticism section, and it's been agreed that there are no edits until we've reached a consensus. My vision of the article is that it will give people critical information of Generative grammar because this is an approach that made some interesting claims in the 1960s, but there was eventually no evidence for any of them. The approach however remains quite influential. Many people would wait for GG to fade away in a natural course (Chomsky is now 91 years old), but WP obviously needs to provide information to people on a daily basis. These people would include students and education providers, among others, who would be able to prepare for changes beforehand.

But information needs to be accurate and in line with the WP guidelines. This is somewhat difficult because journals don't publish articles claiming "there is no evidence" for something on a regular basis. Likewise, reference volumes don't usually list things that are not based on research, but rather tell you what are. I think there will be a small number of papers, however, that we'll be able to use as directly critical sources.

Starting with the first 'issue' which is that of purported lack of linguistic evidence for GG. The background here is in Chomsky's rejection of behavioural psychology. He claimed to have found linguistic evidence that proves that children do not learn syntactic structures, but 'acquire' them from 'UG' which is hard-wired in the brain. Note that there was no brain research done.

The current version stands as follows.

Noam Chomsky, the creator of generative grammar, believed to have found linguistic evidence that syntactic structures are not learned but ‘acquired’ by the child from Universal Grammar. This led to the establishment of the poverty of the stimulus argument. It was later found, however, that Chomsky's linguistic analysis was inadequate. ^[1]

Pullum and Scholz's article has been commented by others. This paper explains it as follows.

"Example 2b: Is the man who is eating hungry?

An alternative rule would be to move the first occurring auxiliary, that is, the one in the relative clause, which would produce the form

Example 2c Is the man who eating is hungry?

In some sense, there is no reason that children should favour the correct rule, rather than the incorrect one, since they are both of similar complexity and so on. Yet children do in fact, when provided with the appropriate context, produce sentences of the form of Example 2b, and rarely if ever produce errors of the form Example 2c (Crain and Nakayama, 1987). The problem is how to account for this phenomenon. Chomsky claimed first, that sentences of the type in Example 2b are vanishingly rare in the linguistic environment that children are exposed to, yet when tested they unfailingly produce the correct form rather than the incorrect Example 2c. This is put forward as strong evidence in favour of innately specified language specific knowledge: we shall refer to this view as linguistic nativism.

In a special volume of the Linguistic Review, Pullum and Scholz (2002) showed that in fact sentences of this type are not rare at all. Much discussion ensued on this empirical question and the consequences of this in the context of arguments for linguistic nativism. These debates revolved around both the methodology employed in the study, and also the consequences of such claims for nativist theories. It is fair to say that in spite of the strength of Pullum and Scholz’s arguments, nativists remained completely unconvinced by the overall argument."

Sure they did. However, I think it is quite correct to say that Chomsky's linguistic analysis was inadequate, based on Pullum & Scholz 2002. By 'inadequate' I mean that it was not false per se, but the claim that such sentence structures are not included in the 'positive data' received by the child was pointed out to be false by Pullum & Scholz. Weidorje ( talk) 08:34, 6 March 2020 (UTC) reply

Goedemorgen to you as well. I agree that the article should help readers understand these criticisms. But to do that, it must accurately represent the issues. Not just because of WP policy but also because doing otherwise would give the false impression that non-generative work never seriously engages with generative work. I say this from my own experience. During my bachelor, I would never have imagined that I would end up working on discourse--- isn’t that one of those subfields that’s all about strawman attacks on GG? Well, no, it’s a rich tradition with all sorts of incredibly exciting work going on. I’m very glad I discovered that, and I wouldn’t want this article to serve as a barrier for others.

I'm not happy with the phrasing "Chomsky's linguistic analysis was inadequate" since (1) it's vague (was his syntactic analysis wrong? did he make an incorrect claim about the structure of the learner? did he make an incorrect claim about how much data the child hears?) and (2) I'm still not convinced it's true. Unless I'm misunderstanding (which could very well be) it looks like Chomsky's claim wasn't that children don't hear many sentence with subject-aux inversion, but rather that without innate biases, children would need to hear an infinite number of them to rule out all the simpler candidates for the rule.

Here’s my current suggestion for rewording this section, based in part on the quote you shared from the Clark and Eyraud article. I think it’s an improvement because it (1) explains precisely which generative ideas are being criticized (2) helps the reader understand some of the specific objections P&S raise and (3) conveys that this remains an open issue.

Recent work has challenged the empirical strength of poverty of the stimulus arguments for universal grammar, some of which rely on the premise that certain sentence structures are too infrequent for children to learn without innate biases. For instance, Pullum and Scholz (2002) disputed Chomsky’s claim that children are not exposed to enough auxiliary-fronting constructions to acquire the relevant grammatical rule. This objection has led to ongoing back-and-forth between nativist and non-nativist researchers concerning how much evidence would be necessary to learn the relevant rule. Botterweg14 ( talk) 18:18, 7 March 2020 (UTC) reply

No, such discussions are not interesting from a scientific view. It makes no difference what they say because their underlying assumptions are not based on evidence. There's a page for the Poverty of the Stimulus for discussions of the complexities of the issues. What people need to see here is that the whole claim is simply made up. This helps you understand why it is said to be based on guru teachings; and also why claims made by generativists cannot be rejected (they are unfalsifiable). To have evidence for something being hard-wired in the brain, you would obviously need brain research (this is where the generativists will start discussing the complexities of the issue). Chomsky thought to have provided linguistic evidence, but as he was never a scientist, his linguistic claims are also false. This was pointed out by Sampson in 1989 and Feidin in 1991. Generativists will see it as a victory that they continued with their claims despite the counter-evidence. But from an objective point of view, this is how a scientifically rejected framework gradually turns into a pseudoscience.

Pullum & Scholz demonstrate that Chomsky's claim means children are not exposed to sentences such as:

(24) a. Will those who are coming raise their hands?

b. Can the people who are leaving early please sit near the door?

c. Is the man who was showing you the pictures still here?

d. Would anyone who is interested see me later?

e. Can a helicopter that has lost its tail rotor still fly?

f. Will the owner of the bicycle that is chained to the gate please move it?

g. Could the girl who has lost her ticket come to the desk?

h. Could a tyrannosaurus that was sick kill a triceratops?

(25) a. If you don’t need this, can I have it?

b. Since we’re here, can we get some coffee?

c. When you’re done, could I borrow your pencil?

d. Given that I’m not needed, can I go home?

e. While you’re getting cigarettes, could you get some more milk?

f. Though you won’t like me asking, did you brush your teeth?

The linguistic claim (potential weak evidence) is quite obviously unfounded. Therefore Chomsky's analysis was inadequate in the dictionary sense 'lacking the quality or quantity required; insufficient for a purpose.' It was too narrow. Chomsky provided a sentence that could convince a number of people, but the full list above would not have convinced anyone. Weidorje ( talk) 06:47, 8 March 2020 (UTC) reply

Also, as a comment to Botterweg14's comment: "Unless I'm misunderstanding (which could very well be) it looks like Chomsky's claim wasn't that children don't hear many sentence with subject-aux inversion, but rather that without innate biases, children would need to hear an infinite number of them to rule out all the simpler candidates for the rule." This is quite clearly wrong based on the original discussion, but of course some people would have made new interpretations of the old claims as a distraction from the embarrassment, a bit like the Jehova's witnesses since the World didn't end in 1917 or any other year predicted so far. Some guru might say that a child needs to hear an infinite number of sentences in order to learn them (which would obviously be impossible), but such a claim is neither based on research nor common sense. Do the readers of Wikipedia need to be aware of all statements that may or may not have been made? No. Weidorje ( talk) 11:09, 8 March 2020 (UTC) reply

I'm not sure I understand. What's your objection to my rephrasing of that paragraph in the article? Botterweg14 ( talk) 13:52, 9 March 2020 (UTC) reply

Because it's not relevant. You're thinking that the criticism section should reflect the position of the generative community, but it is not true. The Wikipedia page of generative grammar does not owe anything to generativists, and generativists do not have any claim on the page. It is in no way their intellectual property or advertisement. The heading says: 'No evidence', and this will be elaborated in the section. There has been progress so far in that we both now agree that there's no linguistic evidence (for the relevant claim). So, the statement will start by pointing this out. Next, the fact should be stated that Chomsky did attempt to provide linguistic evidence but failed. These two last ones are the points you haven't agreed with. But if you will, we could then expand the passage by stating that many generativists do not consider this as problematic. That would reflect both the objecctive view and the discussions among the generative community. Weidorje ( talk) 14:30, 9 March 2020 (UTC) reply

I'm a little confused by your comment. I haven't agreed to your statement that there's "no evidence" and I haven't in any way suggested that this page should be an advertisement for generative grammar. My rewrite is an attempt to more clearly articulate what specifically Pullum and Scholz claimed and to situate their arguments in the ongoing discussion in the literature. I don't see how that information could be irrelevant. I don't think the article can state that Chomsky's analysis was inadequate given that sources disagree about that. Botterweg14 ( talk) 18:34, 9 March 2020 (UTC) reply

So there is evidence? Weidorje ( talk) 14:43, 10 March 2020 (UTC) reply

Our personal opinions don't matter for the Wikipedia article. The sources currently cited in the article clearly show an ongoing debate about poverty of the stimulus-- as do many others (e.g. Pearl 2019). Botterweg14 ( talk) 15:09, 10 March 2020 (UTC) reply

Sure, the sources currently cited in the article clearly show many kinds of things. The topic here is 'evidence'. OK? According to the sources: is there evidence or is there not evidence? Weidorje ( talk) 08:01, 11 March 2020 (UTC) reply

Once again, sources disagree about the state of the evidence. That was one of the things I was trying to capture in this rewrite, the other being the actual objection raised by P&S.

Recent work has challenged the empirical strength of poverty of the stimulus arguments for universal grammar, some of which rely on the premise that certain sentence structures are too infrequent for children to learn without innate biases. For instance, Pullum and Scholz (2002) disputed Chomsky’s claim that children are not exposed to enough auxiliary-fronting constructions to acquire the relevant grammatical rule. This objection has led to ongoing back-and-forth between nativist and non-nativist researchers concerning how much evidence would be necessary to learn the relevant rule. Botterweg14 ( talk) 13:09, 11 March 2020 (UTC) reply

So basically you're saying that Sampson and Freidin didn't really criticise Chomsky's analysis because they were only going back and forth with their arguments? Also note that we're not talking recent work, but 1980s. Weidorje ( talk) 07:58, 12 March 2020 (UTC) reply

No, I'm not saying that. I'm saying that the article should reflect the fact that subsequent work has replied to the criticisms in the Pullum and Scholz (2002) paper, leading to an ongoing back-and-forth between different researchers. Pearl (2019), which I mentioned above, summarizes a lot of the debate. My other point which I would like to reiterate is that the article should clearly convey the content of those criticisms. Do you disagree with either of these points? And do you have any specific objections to the text I've proposed above? Botterweg14 ( talk) 15:04, 12 March 2020 (UTC) reply

I hope I'm being sufficiently specific. Your wordings are not fine at all. The reader primarily wants to know what the criticisms were and what kind of problems there may have been. Unfortunately, your idea is that people don't need that type of details, but should instead rest assured that whatever problems someone may have said there were, this is now under control: there's nothing to see here, keep calm and carry on thinking GG is a very scientific thing.

Our discussion is exactly like talking to an official of the Jehova's Witnesses (yes, I do have experience): "There may have been something about the end of the world in 1914, but you don't need to worry about it because you wouldn't understand it anyway, so leave it to those with Knowledge of things because they will explain it in a way that is good for you." Now, this is not how Wikipedia works. Even though the movement is very powerful, WP reflects an objective view, see Watch Tower Society unfulfilled predictions. What this means for our section is that we'll apply the following structure:

(1) Clearly point out that there was criticism as concerns generative grammar, especially the nativism argument or the so-called Poverty of the Stimulus argument.

(2) Clearly point out that the critics (whether right or wrong) suggested that Chomsky's analysis was inadequate (choose between wanting and misleading).

(3) We can add that this led to 'discussions' in early 2000s with no consensus achieved.

(4) Since (3) is now rather old information, add an update of the situation. Weidorje ( talk) 09:43, 13 March 2020 (UTC) reply

Thank you for suggesting that structure-- I think we're making progress here! I agree with your (1-3), though for (2) I'd like to add that I think the text should be specific about which analysis they argued to be inadequate, i.e. that the idea that children don't hear enough subject-aux inversion. (I don't quite agree with the word "analysis" since I think it's more the raw empirical claim that Pullum and Scholz challenged, but I won't push the point if you prefer that word.) With (4), I'm not sure what you mean though I'm very much in favor of adding recent work-- what do you have in mind? Botterweg14 ( talk) 10:53, 13 March 2020 (UTC) reply

Sure, the word claim can be used instead of analysis, although Chomsky's claim was obviously based on his analysis. Any expression is fine as long as the reader gets the information that Chomsky argued for linguistic evidence that there are sentence types which the child is able to produce despite not having been exposed to them in the positive data. He gives an example which he believes is too rare to be assumed as having been heard by most children. They can nonetheless produce the structures correctly, and this is purportedly evidence that children are born with the structure. Pullum and Scholz 2002 write:

4.4. Auxiliary-initial clauses

The apparently strongest case of alleged learning from crucially inadequate evidence discussed in the literature, and certainly the most celebrated, concerns auxiliary-initial positioning in polar interrogatives in languages like English and Spanish: sentences like Are you happy?, the polar interrogative (yes/no question) corresponding to the declarative You are happy. The generalization concerning the formation of such sentence types is structure-dependent: it is based on structural relations (dominance among constituents), not just temporal sequence (precedence among words). SpeciÞcally, it is the matrix-clause auxiliary verb that is assigned initial position, not, e.g., whatever is the leftmost auxiliary in the corresponding declarative clause. Chomsky (1971b: 29–33) gives these examples:

(21) a. The dog in the corner is hungry

b. Is the dog in the corner hungry?

c. The dog that is in the corner is hungry.

d. Is the dog that is in the corner hungry?

e. *Is the dog that in the corner is hungry?

I think these details and what follows in Pullum and Schulz 2002 will be more relevant for Poverty of the stimulus, but if you have a good formulation, why not add a little here.

The discussions that followed are now old news. Traxler's Hanbook of Psycholinguistics from 2006 reports that a 'paradigm shift' followed:

3. A PARADIGM SHIFT: NEW PERSPECTIVES ON LANGUAGE LEARNING Although nativist views of language acquisition are forceful and still widely endorsed, there has been ongoing controversy about the adequacy of such theories as an account of how children develop competence in language. Some critiques directly challenge the logic of arguments made by Chomsky, Pinker, and like-minded theorists, questioning such core assumptions as the universality of generative grammar, the autonomy of syntax in language processing, and the fundamental unlearnability of language (e.g., Bates & Goodman, 1999; Braine, 1994; Pullum & Scholz, 2002; Tomasello, 1995). Other critiques focus on empirical evidence inconsistent with particular nativist assertions. For example, the claim that negative evidence is not available when children make grammatical errors, an assumption central to the “poverty of the stimulus” argument at the heart of Chomsky’s theory, is not supported by a recent analysis of parents’ reformulations in speech to children (Chouinard & Clark, 2003). These diverse challenges, both philosophical and data-driven, have fueled debate over four decades about the explanatory adequacy of nativist theories of language learning. However, in recent years this debate has begun to change in focus and tenor, not only in response to explicit critiques within linguistics and developmental psychology, but also in response to research findings and theoretical insights from farther afield. An alternative perspective on language learning has been gathering force, amplified by new developments in research areas that formerly made little contact with theoretical debates on the nature of language development (see Kuhl, 2004; Seidenberg & MacDonald, 1999; Tomasello, 2003). We focus on four such developments that have begun to change the direction of research on early language acquisition: first, the emergence of more “user-friendly” theories of language and language use; second, the contribution of computational approaches to modeling language processing and learning; third, provocative findings from experimental research on learning and cognitive processing by infants; and fourth, insights from studies with children and non-human primates on the role of social cognition in communication. In different ways, these diverse areas all motivate and support an emerging alternative view of language learning.

That was the situation then. In 2015 de Bot reports in A History of Applied Linguistics:

GG is generally seen as a declining paradigm and its proponents now tend to stay away from conferences like AAAL (the American Association of Applied Linguistics) and University of Boston Child Language Development conferences, as a cursory count of papers on the basis of abstracts shows ... In my view, it is no coincidence that in Levelt’s (1989) ground-breaking book on language production there is only one reference to Chomsky, and that is not N. but C. Chomsky. In the psycholinguistic community, the idea of innateness and a Language Acquisition Device (LAD) were seen as problematic, and the leading institute in the world, the Max Planck Institute for Psycholinguistics in Nijmegen was ostensibly anti-GG. This is also reflected in Levelt’s early criticism of the GG model in his 1975 work What Happened to LAD? There is also a sociological component in the debate about the position of GG at universities. Starting from the 1960s, GG was seen as the new way to go and many linguistics departments turned Chomskian, which led to serious conflicts with linguists of other persuasions who felt sidetracked. Now that generation of GG linguists is retiring and there is a tendency in many universities not to replace them with younger scholars of that school, but rather appoint UB oriented linguists. There is almost a euphoria that the grip of the nativists on what constitutes linguistics is gone and that other approaches and more social orientations are seen as meaningful alternatives. Others try to explain the reasons for the decline of GG ... Some informants are quite outspoken about the role of GG in AL. William Grabe states: “Fundamentally Chomsky is wrong and we wasted a lot of time. In 1964 Chomsky’s Aspects was published. Now, in 2014, we are 50 years later. What impact has all of that had in real world language use? This is an overstated theoretical direction.” Jan Hulstijn summarizes: “Generative linguistics has had no noticeable (or durable) impact.”

I think this shows the POS debate is old news and GG has lost its previous status in applied linguistics, now operating from the margins. Weidorje ( talk) 09:27, 16 March 2020 (UTC) reply

In the article, it is stated that there is no proof when it comes to the existence of generative grammar. However, it is a misconception that confuses both pyschology and neurology. That is, generative grammar does NOT try to be the theory of neurology. Instead, it aims to be a part of cognitive psychology, which is consequently fair to ask whether these rules really exist in the brain. To be more precise, how can one find a real description of GG in the brain, meanwhile there are tons of neurons firing away there? That means, neurology examines concreteness, but cognitive psychology aims to model the abstract events in the mind whose results can be realized concretely(for example, memory, perception and so on) and this is exactly what GG wants to do, since one can of course not find the real concrete language rules by studying in the field of neurology. Kuzeytaylan ( talk) 00:03, 24 June 2020 (UTC) reply

Absolutely, I think your definition of cognitive science is quite correct. However, whatever kind of model one constructs, it will need to be based on something. Rules can exist in the brain in many different ways, but we want to know whether what it is that Chomsky precisely says is true. We can agree there is no evidence without saying that generative grammar has been proved wrong. It has not and never will be. But some critics have pointed out that the evidence is not there while Chomsky believes it will appear one day. Fingers crossed. Weidorje ( talk) 04:32, 24 June 2020 (UTC) reply

Exactly, one can not find the real evidence in the brain concretely. Notwithstanding, we can, for example, provide an evidence for how subjects are derivated for the first time(or for non-generativists, seen for the first time) under the tense phrase and undergo a movement to the Spec,TP position. For example, “The kids are all in the garden.” and “There are all the kids in the garden”. When there is an expletive subject in the sentence, the real subject cannot undergo a movement to that position(2. Sentence), but when there is not one, then the subject can go up there(1. sentence) and may there leave “all” or bring it too(for example, all the kids are in the garden.). What I mean here is that this is how the modelling is. That is, this is the concrete result of movement, but how this processing is really represented in the mind is for now unknown. There may be something different, but the result can be seen in the examples. Kuzeytaylan ( talk) 00:11, 25 June 2020 (UTC) reply

I think it's clear that different languages have different rules, and that these rules are represented in the cognition or the unconscious mind. This is what de Saussure argued although the idea is age-old. Chomsky's idea is however much more specific because he argues that rules are innate and therefore universal (i.e. manifested in all languages). GG models are different from others in that they have the binary branching NP VP (NP) structure which is not semantically motivated, so it is claimed to be innate and caused by a genetic mutation. When we look at the evidence, there is nothing for innateness or a specific language gene so far. Also, there have been some studies on how English speakers process the NP - NP distinction, and there was nothing specific supporting the idea of VP (NP), that the object is processed as part of the verb phrase. The section and the article is however unfinished because there's been an agreement that it will not be edited until me and Botterweg have agreed on the details. The sources need to be reviewed and partially changed to better ones, and it will be added that many scholars argue that GG is pseudoscience and a cult of personality. In the history section it needs to be added that the GG idea is derived from Thomas of Erfurt's 'Speculative Grammar' (ca. 1300–1310). Weidorje ( talk) 08:27, 25 June 2020 (UTC) reply

Concerning recent edits made by Kuzeytaylan. Some of it was good because now it's clear what the problem with the research was. Next, you're claiming that the research however proved that constituents are cognitive. I'm not arguing that they aren't, but I don't suppose the research you're citing was on that topic. I'll check it soon. Next, you're claiming that the sources don't contain the information that GG was rejected (and pushed into the margins, as it said before your edits). Now, why are you doing that because the matter was already discussed above? I'm duplicating some of it here:

GG is generally seen as a declining paradigm and its proponents now tend to stay away from conferences like AAAL (the American Association of Applied Linguistics) and University of Boston Child Language Development conferences, as a cursory count of papers on the basis of abstracts shows ... In my view, it is no coincidence that in Levelt’s (1989) ground-breaking book on language production there is only one reference to Chomsky, and that is not N. but C. Chomsky. In the psycholinguistic community, the idea of innateness and a Language Acquisition Device (LAD) were seen as problematic, and the leading institute in the world, the Max Planck Institute for Psycholinguistics in Nijmegen was ostensibly anti-GG. This is also reflected in Levelt’s early criticism of the GG model in his 1975 work What Happened to LAD? There is also a sociological component in the debate about the position of GG at universities. Starting from the 1960s, GG was seen as the new way to go and many linguistics departments turned Chomskian, which led to serious conflicts with linguists of other persuasions who felt sidetracked. Now that generation of GG linguists is retiring and there is a tendency in many universities not to replace them with younger scholars of that school, but rather appoint UB oriented linguists. There is almost a euphoria that the grip of the nativists on what constitutes linguistics is gone and that other approaches and more social orientations are seen as meaningful alternatives. Others try to explain the reasons for the decline of GG ... Some informants are quite outspoken about the role of GG in AL. William Grabe states: “Fundamentally Chomsky is wrong and we wasted a lot of time. In 1964 Chomsky’s Aspects was published. Now, in 2014, we are 50 years later. What impact has all of that had in real world language use? This is an overstated theoretical direction.” Jan Hulstijn summarizes: “Generative linguistics has had no noticeable (or durable) impact.”

You should not have removed the criticism. Weidorje ( talk) 07:01, 10 September 2020 (UTC) reply

First of all, I did not understand why you deleted the information in the first section that says that neurology is not concerned with how mind works(concerning GG). This is a significant misconception. That is why it was crucial to point out that it is pyshcology under which the theory is considered to be real. Otherwise, how is it possible to find a concrete process of merging of a verb and a noun etc.? It is as if GG claims to be a theory of neurology but neurology did not find any evidence so GG is not real. Also, how is it possible that you don’t suppose the article which i cited is about words being realized as constituents? You should not have deleted it without reading the article. Kuzeytaylan ( talk) 10:31, 10 September 2020 (UTC) reply

The piece of information you are referring to is out of place in the GG article. If it was a quote from a published source, a generativist explaining why neurologists cannot falsify their claims, it could be considered maybe somewhere else on the page. Also, neurology is basically another name for neuropsychology, so it is relevant for psychology. It is specifically claimed that UG is hard-wired in the brain. The question is how such claims have been founded in evidence. There could be many kinds of evidence for GG: psycholinguistic, genetic, maybe even linguistic. However, there is nothing to be seen. It is very clearly stated by generativists that their approach is a particularly scientific one. To my understanding, scientific only means the same as being based on hard evidence. As such, GG is not supported by research. This is what the reader wants to know. Weidorje ( talk) 11:33, 10 September 2020 (UTC) reply

As regards the idea that words are realised as constituents in the mind, I haven't seen such research. Maybe it's your interpretation of the cited study. If there's one it could be discussed in the article, but not as concerns the criticism of POS. It doesn't belong there. Weidorje ( talk) 11:33, 10 September 2020 (UTC) reply

I even cited GG’s being a theory of psychology from a book, but you also deleted it, which was an important distinction. It therefore seems that you want GG to be considered a totally fake theory, although there are supportive evidences. There are of course critics about them, but these are also another point of views, but you write them as if the critics were right, so GG was false. That is why I kept changing your writings. Kuzeytaylan ( talk) 10:43, 10 September 2020 (UTC) reply

I at the same time want to thank your for your attention. Kuzeytaylan ( talk) 10:48, 10 September 2020 (UTC) reply

Thanks, Kuzeytaylan, it is a matter of improving this article. (Also see my answers above - there was an edit conflict because we were writing simultaneously.) The cited source was removed because the statement only repeated what was already said. The word 'cognitive' is a reference to cognitive psychology. If there's a POV bias it can be fixed, although if the topic is unfounded, the POV can be mainly negative. See e.g. Neuro-linguistic programming. Generative Grammar is no worse than Lakoff's cognitive linguistics or Tomasello's usage-based linguistics as comes to scientific evidence. There are claims of evidence, but critics argue that these claims are false. To me it seems that you want to show that language is organised as a rule-based grammar in the mind, and not like Lakoff claims. I think this has been the kind of juxtaposition that is fully avoided by the article as it is. You could start a section discussing the matter, but don't say anything has been proved if it hasn't, and don't say that Chomsky and Lakoff are the two alternatives that researchers must choose from - they don't. Thanks. Weidorje ( talk) 11:33, 10 September 2020 (UTC) reply

Also notice that the current wording in Evidence is generous. It now says it is a a proven fact that children are able to manipulate the constituent structure, implying that rules are real. However, it is also says it is not a proven fact that they are innate. Weidorje ( talk) 12:29, 10 September 2020 (UTC) reply

I read all of your comments, thank you again. I am unfortunately unable to continue writing since i am on the phone and it is hard to discuss like this. I’d really like to talk to you about these issues. Kuzeytaylan ( talk) 12:40, 10 September 2020 (UTC) reply

Hi all! Currently, the lead of the article states that "[Generative Grammar] is a sociobiological modification of structuralist theories", and cites the article "Sociobiology and You" to support this claim. I don't think that source supports that claim. The closest that the source gets to to supporting this claim, in my opinion, is seen here:

"Trained as a linguistic psychologist whose specialist work has focused on irregular verbs, Pinker has made a name for himself as one of evolutionary psychology’s most appealing ambassadors, with a series of bestselling books summarizing an astonishingly wide body of work: The Language Instinct (1994) looked at the mental module for language processing, first proposed several decades ago in Chomsky’s idea of a generative grammar underlying all human language, and refined by subsequent developments in neuroscience and evolutionary theory."

So the article only supports a weaker claim, there exists a mental module for language processing, rather than generative grammar as a whole being supported; and even that module is merely supported by neuroscience and evolutionary theory, rather than these being primary motivators for Generative Grammar, which I think is suggested by the phrasing "[Generative Grammar] is a sociobiological modification of structuralist theories". Because of this, I'm leaning towards removing that reference (as it does not seem relevant to the claim made here), but I wanted to check what other people think before removing a properly formatted citation. Alternatively, we could rephrase the claim to something like "Some aspects of generative grammar, such as the idea of a language module, have been influenced by developments in neuroscience and evolutionary theory". Thanks for reading! JonathanHopeThisIsUnique ( talk) 02:17, 15 December 2020 (UTC) reply

Hi User:JonathanHopeThisIsUnique. Generative grammar is usually considered a biological approach and not so often sociobiological. In the context of human sciences, though, these would be synonymous. Chomsky was probably first identified as a sociobiologist by Ullica Segerstråle in the 1970s. It's true though that the article on Pinker is not a good one here, so it can be removed.

As for a better source, Croft 2006 discusses generative grammar's relationship with structural linguistics and its place within (socio-)biological and evolutionary approaches. John A. Hawkins is another source. From his books (and elsewhere) you will see that the generative movement is divided in two. One camp (Chomsky, Pinker, Newmeyer) supports the idea of an innate, genetically determined view of language. The other (Hawkins and various advocates of LFG, HPSG and OT) advocates the evolutionary view that language is a system which has adapted to neurological processing biases of the human brain. Chomsky has discussed his view of the biological evolution of language in many places (well, all of them basically), for example here. Seuren's book 'Western Linguistics' discusses thoroughly the development of generative grammar from Hjelmslev's glossematics through Harris's and Hockett's model to Chomsky's Syntactic Structures. Weidorje ( talk) 20:37, 15 December 2020 (UTC) reply

I changed the source. Still, GG is not so often called sociobiological or biologistic, but biological. I just think plain 'biological' is somewhat misleading because it seems to suggest people doing GG are biologists. Weidorje ( talk) 13:48, 17 December 2020 (UTC) reply

Generative grammar and generative linguistics should be two separate articles. Generative grammar is just about the tecnical concept, while generative linguistics is a more general article about the linguistic approach/paradigm, its epistemiological/philosophical assumptions and foundations, history, subfields, and relationship with other approaches.-- OpenNotes1 ( talk) 22:56, 25 January 2021 (UTC) reply

The content from this revision should be restored as a working point, as it contains some proper information about the approach of generative linguistics in general, not about the tecnical concept.-- OpenNotes1 ( talk) 23:39, 25 January 2021 (UTC) reply

I agree. The article as it is now seems to conflate the two (and suffers from WP:NPOV and WP:BALASP besides). Compare with Model-theoretic Grammar, which provides a better (and more concise) definition of generative grammar than this page does. EstraKC ( talk) 18:02, 12 April 2021 (UTC) reply

I agree as well. Botterweg14 ( talk) 23:20, 12 April 2021 (UTC) reply

Who says generative linguistics is different from generative grammar? I haven't heard that before. OpenNotes1, the link you posted is your own redirect from generative linguistics to "formalism". But formalism is a reference to philosophy of mathematics, see https://plato.stanford.edu/entries/formalism-mathematics . The term is quite ambiguous because Chomsky's point is that language is not purely axiomatic. I do understand advocates of GG would like to change this article to a more favorable one, but I'm not sure what OpenNotes1 wants to do. Is your point that the mathematical calculus, or what is also called transformational grammar, should be separated from GG? Weidorje ( talk) 07:31, 13 April 2021 (UTC) reply

Weidorje ( talk · contribs), I'm drawing from Müller's "Grammatical Theory: From transformational grammar to constraint based approaches", particularly Chapter 14 where he makes the comparison between the two clear. There's definitely a tendency in the field to use "generative grammar", "generative linguistics", "chompskian linguistics", etc. interchangeably. Perhaps there ought to be a disambiguation page, one for "Generative-Enumerative Grammar", the technical concept (a formal grammar that contains the rules that generate the set of grammatical sentences), and one for "Generative Linguistics", which "Generative Grammar" can redirect to. EstraKC ( talk) 22:51, 13 April 2021 (UTC) reply

I agree there's Chomskyan GG, other GG models, and then there are non-GG approaches to mathematical grammar in general linguistics. The terminology is tricky. Weidorje ( talk) 06:48, 14 April 2021 (UTC) reply

Is 'generative grammar' written with or without caps? This article changes randomly between the two; there are even instances of one element being capitalized but not the other.-- Megaman en m ( talk) 21:49, 23 July 2021 (UTC) reply

Wikipedia guidelines tell us to avoid unnecessary capitalization, so I think it should not be capitalized although I suppose it's usually capitalized elsewhere. There's also the possibility that Generative Grammar refers to the Chomskyan idea in particular, and generative grammar to the general concept. I find it currently only once capitalized, depending on different editor styles. I wouldn't mind if it was consistently capitalized, though. What do you think? Weidorje ( talk) 19:40, 24 July 2021 (UTC) reply

I'd be for ditching all capitalization of this term, I don't see a use for it.-- Megaman en m ( talk) 20:56, 24 July 2021 (UTC) reply

Okay, done. Weidorje ( talk) 14:06, 25 July 2021 (UTC) reply

I understand that this theory is mostly obsolete and is not accepted/ proved by linguists in general anymore, but the language of the page is highly negative (with uses of words such as "so-called", for example). This feels unnecessary and the comment of Chomsky saying that "it's only a theory that will be understood later on" feels more like a personal opinion rather than anything that is pertinent to the theory itself. 267 17:06, 26 March 2022 (UTC) reply

Agreed. For example, the neuroscience stuff from 1993 citing Kluender & Kutas seems trivial and makes it seem like the editor was trying to pile up criticisms. Overall a pretty poor article as it stands. D emcee ( talk) 07:50, 13 October 2022 (UTC) reply

There should be plenty of criticism cited here. The paper cited at the end of the opening section, Modern language models refute Chomsky’s approach to language, is not representative of it, nor does it hold up to any scrutiny. It's not published anywhere and clearly motivated by some extra-scholastic factors.

To serve the topic the link tries to provide: I would suggest there can even be a section disambiguating the goals and results of LLMs vs. the goals and results of human language theory. But the current article cited doesn't even address the same question the theory is attempting to: Chomsky's theory tries to answer how do humans arrive at their grammar that produces language, whereas this paper is focused on how a different computational process simulated human language - the result, not the how. 100.2.102.27 ( talk) 23:29, 21 February 2024 (UTC) reply