CHOMSKY.

I have occasionally made offhand remarks indicating my dislike for Noam Chomsky’s linguistic theories and, still worse, his effect on the field, but I have not had the heart to go into detail; I’m worn out from all the arguing I did about him back when I was an actual linguist (now, I just play one on the internet). Fortunately, my procrastination has paid off (as it so often does), and Scott Martens has done the job for me. I direct anyone who wants to know exactly how wrong and destructive Chomsky has been to go forthwith to Pedantry and scroll down to “Friday, July 25, 2003: My carefully considered and well earned aversion to Noam Chomsky” (I won’t even try to provide a permalink, Blogger being what it is). Quick summary: “His principles ultimately produced nothing, and may well have set linguistics back decades. The day will come when his legacy is compared to Skinner’s, and when historians of the social sciences will debate which one ultimately caused the most damage.” But there’s much, much more.

Addendum. Scott expands on the subject in the comments to this entry.

A caveat: after sixteen paragraphs, you will reach the sentence “There was some more stuff I was going to say about Cambodia.” You can stop there, unless you’re particularly interested in the still-simmering argument over exactly what Uncle Noam said about the Khmer Rouge 25 years ago and whether it was justified given what was known at the time. As Scott says, “There’s something about that country that seems to drive its students mad.”

Comments

  1. *does happy dance*
    Just goes to show that my husband’s linguistics department is deeply retrograde in its Chomskybottism — or is that Chomskybotulism? Which is encouraging, I must say.
    (No worries; my husband is NOT a Chomskybot.)
    The bit about proof-shifting was evident to me even in the basic syntax class. Pfeh.

  2. thanks for the link! As a BA (cum laude) in linguistics, now doing an MA in TESOL, and thinking about a MA/PH.D program in linguistics again, I am glad to see I’m not the only one who despises Noam.
    I teach ESL to adults, and many of them like his politics. Whenever they mention Chomsky, I have the unfortunate reaction of blurting out, “I hate him!” and they are shocked. Then I have to explain why, but I don’t know if many of them believe me that my distaste — gut-level distaste -is caused exactly by what is discussed in that well-reasoned post. Plus a couple of other points, too..

  3. Maybe I missed something, but the post seemed to make lots of sweeping claims without any support for any of its points. Not that I like Chomsky – as a linguistic anthropologist I’m quite put-off by the idea that linguistics is so removed from how language is actually used – but still, having studied a little linguistics, I find it hard to believe the claim that Chomsky contributed “nothing”. For one thing, there does seem to be some empirical verification of some of his theories within the fields of child language acquisition as well as studies on people with brain injuries, not to mention comparative work across langauges. For instance, the fact that children never make certain kinds of linguistic “mistakes” and that certains structures don’t exist in any langauge seems to support some of Chomsky’s basic claims. I’m not at all familiar with the arguments within linguistics against Chomsky – but, unfortunately, reading this link didn’t really do much to change that. So, LH, I’m afraid you are still not “off the hook”… something still needs to be written which makes the case against Chomsky to a non-specialist. Like I said, I personally come from a very different tradition, in which “dialog” not “the sentence” is the basic unit of analysis and my gripe with Chomsky is that he considers actual language use to be outside the scope of linguistics. But it seems that the linked post is arguing that at the level of the “sentence” there are significant failings with Chomsky, I’d like to know what those are specifically…

  4. Kerim, I am not familiar enough with the studies you cite to try dealing with them; I can only assure you that from within the linguistic tradition in the strict (non-anthropological/sociological) sense there is every reason to suppose that whatever data may seem to support him are likely to have been chosen for that purpose. I’m afraid I can’t improve on Scott’s formulation. But if you dislike Chomsky anyway, you may as well just take our word for it that there are even more reasons to do so!

  5. Kerim, you’re point is fair criticism. I did not offer supporting grounds for my case against Chomsky’s linguistics. On a blog, in short format, when I know that most of my readers are not in the language business, if I had written such a post it would have taken more of my time and resources than I can offer it, and no one would have read it. But you probably shouldn’t take my word for it.
    I can’t just point to a book on linguistics for the non-specialist that really dedicates itself to dismantling Chomsky and Chomsky-inspired approaches. Frankly, most of the books in English about linguistics for non-specialists are written by people from within the Chomskyan tradition (e.g. Pinker). The most important single exception I can think of is George Lakoff’s books. In Philosophy in the Flesh, he spends a fair amount of time attacking a number of key Chomskyan principles. It is the only resource off the top of my head that I can point you to.
    I can only assure you that there are contesting interpretations of all those studies. I can offer you a few brief pointers.
    For the brain lesion studies, I personally have a soft spot for the notion of “emergent modularity” advanced by Annette Karmiloff-Smith in Beyond Modularity. The primary data in support of her conclusion is that young children who suffer brain damage to the “language centres” of the brain are very often capable of learning language just as well as children without lesions. The MRI evidence shows fairly conclusively that they just use a different part of the brain to do language. This suggests that even if language is identified to some degree with one area of the brain (to what degree is still an open question), localisation seems to be the result of learning a language, not its precondition. This undermines the idea of innate modularity in language which is so central to Chomsky and to post-Chomskyans like Pinker.
    As for the types of morphological errors children make as they learn their first language, I wrote my second last school thesis on neural network solutions to this problem, starting with Jeff Elman’s work in the early 90’s. The debate ran through the late 90’s but seems to have faded out of view recently. In 1997, Gert Westermann showed how a certain kind of neural network produced results that explained those errors in English quite effectively without pre-existing linguistic knowledge or any kind of explicitly encoded rules. His conclusion claims to preserve one aspect of Pinker’s model: the “two-track” model of morphology. However, he does not seem to understand that Pinker means something very different from Westermann when he says “two-track.”
    I work mostly in empirical computational linguistics. How language actually gets used is exactly what I study. If I didn’t compeletly ignore the competence/performance distinction, what I do would be impossible.

  6. Cliff Crawford says:

    Hi Kerim,
    You might also be interested in Geoffrey Sampson’s “Educating Eve: The `Language Instinct’ Debate”, which criticizes Chomsky’s and Pinker’s theories of language innateness.

  7. One of Chomsky’s strengths, too, is his utter inability to write well. It’s very tedious to criticise a work if halfway through you find yourself obliged to kill yourself from exposure to ugly syntax and an absolute disregard for standard idioms. (a comment)
    BWAH!
    Excuse me, I thought I was the only one who had ever had that reaction to Chomsky.

  8. Thanks Scott,
    From what I know, Chomsky himself is quite upset with Pinker’s characterizations of his arguments. I think the “modular” aspect and the “instinct” aspect does need to be separated. Chomsky clearly makes an argument about the former, but the latter seems somewhat incidental to his theory. But it seems as if Annette Karmiloff-Smith’s book attacks the modularity argument as well (hence the title) so that seems worth looking into.
    I’m also somewhat familiar with nural network approaches to phonology, having studied phonology under someone who was very into optimality theory – but that wasn’t presented as being anti-Chomskian. In fact, it was simply presented as a better tool to do the job. I would need to understand more to see how one invalidated the other.
    I guess my problem comes down to the fact that I see “empirical” as good, but “empiricist” as bad. The competence/performance distinction far predates Chomsky and goes back at least as far as Saussure. Chomksy’s work certainly re-engergized attemts to describe what competence means and to try to formally describe what such competence entails. That he has had to try several times and keeps starting again with some new model is certainly a good reason to be sceptical of the whole enterprise – and I do think it needs a more empirical focus. However, I worry about tossing the baby out with the bathwater and denying the very existence of such a distinction or its utility.

  9. It’s true that some of the ideas I have attributed to Chomsky do predate him – not just the competence/performance distinction but headless phrase structure grammar as well. Chomsky’s argument is very much for an account of language which is quite strictly modular and innate to the point where there ought to be no genuine distinction in linguistic abilities between healthy people. Where he and Pinker go their separate ways is on the evolution of language. Chomsky thinks – if I understand him correctly – that innate linguistic competence is the result of some other cognitive capacity being turned to some novel end, while Pinker advocates a strictly adaptationist account.
    The constraint satisfaction school of neural networking is borderline compatible with some aspects of the Chomskyan approach. It is fairly neutral with regard to modularity and innateness. However, there is no easy way to take a neural network approach and explicitly encode rules of language. There are a couple of really hard ways. Pinker has, as a result, rejected the usefulness of connectionist modelling completely. I have to admit, I’ve never seen a word from Chomsky about neural networks per se, but I just can’t see anything that can be salvaged of his work in syntax without explicit rules, and neural networks just don’t afford themselves to explicit rules.
    As for the competence/performance distinction – I’ll agree that there is a principled position that only admits that perhaps the distinction has been overblown. The thing is, I just don’t think one can separate any aspect of language from communicative context. People communicate in a manner predicated by their abilities and the communicative demands placed on them. There are several different ideas of what competence means, but at the core of all the ones I can think of is a notion that there exists a kind of language-as-abstraction in people’s heads, and any failure to live up to that abstraction is due to flawed communication channels.
    I agree with you about empiricism. I’m not an empiricist. Most of the time I’m a sort of environmental engineer, designing software to feed texts to translators in the manner most conducive to them, and the stream of text flowing through the system I work on is a sort of environmental restriction on my activities. Even when I did more theoretical work (and in the future, when I hope to be doing so again), there is an engineering aspect to it. I build machines, and I like theories that help me build machines, not merely data-driven observations. I know it’s not the only way to work, but it’s the one that works for me.

  10. As a recovering linguist, I, too, have always had to explain my aversion to the world of “Norm” (as an old friend at Berkeley refers to him.
    The Chomskyan revolution, to my mind, is simply part of a larger (lemming-like) drive throughout the social sciences and humanities to introduce “rigor”–hence, the concentration on theory and theory construction to the exclusion of all else.
    Remember, most of Chomsky’s fiercest critics were once his disciples.

  11. How about a list, AP? 🙂
    Something to cheer my husband up when he’s talked to his Chomskybot department chair.

  12. Dorothea: Post-Chomskian Feuding in linguistics? How long have you got?
    Paul Postal honed his debating style (aptly summarised as “calling everyone else stupid”) in the service of the Chomskter before his apostacy. Jim McCawley, George Lakoff and other Generative Semanticists acromoniously split from the Mothership – there’s a book called “The Linguistics Wars” about this. Pullum’s “The Great Eskimo Vocabulary Hoax (and other essays)” is repeatedly scathing. Gary Gazdar of the Second Great Heresy (renouncing transformations in favour of a return to PSGs) is dismissive of Chomsky’s contribution in this interview.
    Instances could, as they say, be multiplied.

  13. This debate reminds me of the unfortuantely unattributed quote, “people get ahead in other disciplines by standing on each other’s shoulders, but linguists get ahead by standing on each other’s necks.”

  14. I am an undergraduate linguistics student at the University of Arizona, and, as a beginning student, it is odd for me to read such vitriolic statements against Chomsky and his theories. Of the three professors I’ve had in class thus far, all of them are Chomskyan (Chomskian?) supporters with two of them being avid Chomsky freaks. I can’t tell you how many times I’ve heard ‘colorless green ideas sleep furiously’! Very irritating. The consensus among my classmates seems to be mostly students who are sick of hearing about him but accept his theories as valid and a few who would (desperately, in some cases) like to be pointed in the direction of some anti-Chomsky literature, if only for a change of pace. As for myself, I think I’ll pick up a copy of ‘Educating Eve’ since ‘the Language Instinct’ was “highly recommended” (ie required) reading for my 201 class.

  15. Of the three professors I’ve had in class thus far, all of them are Chomskyan
    And you wonder why people are vitriolic? The man has cornered the market! Of course the students “accept his theories as valid”; students (except for the usual malcontents) accept what they’re given. And in this case it’s bull hockey.

  16. No scientifical founding is beyond contestation, Chomsky contributed a lot to the development of linguistic: He raised questions which we must be thankful of 🙂 Eventhough chomky’s work generates lots of “whats” and “the hecks” they all brought linguistics to a majuscule scale.
    Just like when Einstein came and step on Newton, someone might come and do the same thing to Chomsky. So, rather than yack whether his ideas are valid or not, it will be better if we do a research or study then come up with something that can counter chomsky’s foundings, and the same time contribute to the linguistics society.

  17. hmmm. i think you are all a bit unfair to chomsky to say he has contributed nothing goes a bit far, everyone in any field of study has contributed something whether it be simply a theory for people to reject, they have still contributed to people coming up with alternative theories.
    i myself dont totally agree with chomsky’s innateness theory but i think it cant simply be rejected as it does make sense on some levels and attempts to explain how babies and small children learn correct grammar despite deficient data around them to learn from (eg adults who talk usuing slang, abbreviations, make grammatical errors, use jargon and produce incomplete or abbreviated sentences. Behavioural theories cannot explain this…
    i think chomsky should be given some credit at least for attempting to explain what others haven’t and for coming up with a feasible theory.

  18. Richard Weavers says:

    CHOMCIDIA
    I’ve just stumbled across this site, maybe two and a half years too late. Y’aull still there?
    I’m particularly interested in Scott Martens’s comment:
    “Friday, July 25, 2003: My carefully considered and well earned aversion to Noam Chomsky”…,
    which I think is a prevalent enough feeling to deserve a name. If you’re there, Scott Martens, could you email me, please?
    Richard Weavers
    spamequalsdeath@yahoo.com

  19. We are of course still here, and while Dorothea refers to human Chomskybots, she did not link to the One True Chomskybot:

    Of course, the theory of syntactic features developed earlier suffices to account for nondistinctness in the sense of distinctive feature theory. If the position of the trace in (99c) were only relatively inaccessible to movement, the notion of level of grammaticalness is to be regarded as an abstract underlying order. However, this assumption is not correct, since the natural general principle that will subsume this case is not subject to a corpus of utterance tokens upon which conformity has been defined by the paired utterance test. Presumably, this analysis of a formative as a pair of sets of features is not quite equivalent to the levels of acceptability from fairly high (eg (99a)) to virtual gibberish (eg (98d)). Notice, incidentally, that the appearance of parasitic gaps in domains relatively inaccessible to ordinary extraction appears to correlate rather closely with a general convention regarding the forms of the grammar.

    As the bot says about itself, look on his works, ye mighty, and despair. The code was written originally by John Lawler and rewritten for the Web by Kevin McGowan, and the underlying phrases collected by John F. Sowa, but the phrases themselves — they are the work of the Master.

  20. Thanks for reviving this; I replaced the Pedantry link in the post with an archived one. Thank goodness for the Wayback Machine!

  21. Stu Clayton says:

    My eyes got to “parasitic gap”, and stuck there as to the Tar Baby. It represents to me the essence of advanced flapdoodle French prose – and in English even ! It means, and it don’t.

    Here be much to ponder. Or maybe I’ll just blow it off. My rural network can take only so much.

  22. David Marjanović says:

    Knowing the bot had composed this, I watched how far I would get in before encountering the first phrase I could recognize as nonsense.

    It is “subject to a corpus”.

    Checking again, I find a grammar mismatch in “if […] were […], the notion […] is to be regarded […]”. But other than that I can’t tell the first five lines from the real thing.

  23. David Eddyshaw says:

    @Stu:

    “Parasitic gap” does actually mean something.

    https://en.wikipedia.org/wiki/Parasitic_gap

    The problem (IMHO) is not so much the peculiar terminology as the elaborate theoretical superstructure erected on the basis of such linguistically contingent and frankly marginal oddities. My own response to Chomskyism in this, its Blue Period, is basically: “Fascinating. So what?”

  24. Stu Clayton says:

    @David Eddyshaw:

    Thanks for the link, but the entire article is a mystery to me. What I find is this: one takes a complete sentence, removes one word, and calls it a “gap” in the sentence. Then decades are spent in search of a theoretical acount of how gaps can arise when they are introduced. This deserves a gold WTF.

    A footnote says “Parasitic gaps have been studied most using English and Swedish data”, and I see another one referring to Romanian. Does all of this make more sense with examples from German, French or Spanish ?

    This is parasitic on the intelligence, that’s for sure. Now where are my tick tweezers ?

  25. Wow, I wonder what happened to poor B. Cummings, Sept 3, 2003 10:07pm. Disillusioned, of Arizona.

  26. January First-of-May says:

    What I find is this: one takes a complete sentence, removes one word, and calls it a “gap” in the sentence.

    I don’t think it’s quite that – my impression was that “gaps” were located where a restructuring of the sentence had plucked a word out of its place, kind of like a syntactic version of those runaway vowels from Russian grammar.

    The runaway vowels are, of course, the places where the fall of the yers (or, rarely, another sound change) had left a vowel in some parts of the paradigm but not in others. Presumably the Chomskyan linguists were trying to figure out what the syntactic yers that fell to leave the gaps might have been (if they ever existed, which they probably didn’t, but good luck telling this to the Chomskyans).

  27. David Eddyshaw says:

    @January:

    Yes, that’s quite right. This business of “gaps” left by words being moved around is no mere Chomskyan fad but a perfectly real phenomenon in lots of languages, and the idea is actually often quite useful in real-life grammatical description. The gaps often preserve a sort of ghostly afterlife, interfering with the grammar of the sentence in ways which a mere absence would not. I’d give examples but it’s too boring.

    Gaps do indeed often show puzzling behaviour in English in ways which really do cry out for explanation. The trouble was that the Chomskyites early on got very adept in spinning ever more elaborate theories on the basis of such behaviours, and they soon developed a bad habit of elevating these interesting but frankly peripheral aspects of particular languages into supposed keys to the structure of Language in general. Since those days, Chomsky himself has generalised the program to the point where to muggles like me it seems to have no discernible predictive power at all. A cynic might suppose this has something to do with so many past predictions having turned out to be, like, wrong, but I prefer to think that it’s more a case of future-proofing against any possible refutation.

  28. Gaps are very straightforward and ordinary and appear in many languages. “I wanted to go to the store” is a simple example. Who do I want to go to the store? Myself, of course, although the sentence doesn’t say so. We know this is a gap because there are similar sentences where the gap is filled, as in “I wanted Alice to go to the store.” So we can show the gap as “I wanted ___ to go to the store”.

    English and Swedish are odd because they can have two gaps in a sentence, one of which (the parasitic gap) depends on the other. The Wikipedia example “Which explanation did you reject without first really considering?” has a gap after reject, which is what was rejected, but there is another one after considering, which is what was not really considered. Of course, they are the same thing, so we can notate it as “Which explanation did you reject __1 without first really considering __1?”, where the numbers show that the two gaps refer to the same thing.

    In this sentence, we can replace one of the gaps with the pronoun it, but not the other. We can say “Which explanation did you reject without first really considering it?”, but “*Which explanation did you reject it without first really considering?” is ungrammatical, at least for me. So the first gap is parasitic on the second gap. Easy-peasy.

    Of course, I am not a Chomskyite, or a theorist of any sort, and don’t draw any theoretical conclusions from this. But the data is clear and descriptive adequacy is plenty for me.

  29. We know this is a gap because there are similar sentences where the gap is filled, as in “I wanted Alice to go to the store.” So we can show the gap as “I wanted ___ to go to the store”.

    We don’t “know” any such thing; this is an analysis, and not an especially convincing one if you ask me. I do not feel anything missing in the original sentence; the fact that you can have a similar sentence with an object like “Alice” does not prove anything.

  30. January First-of-May says:

    I personally was only thinking of gaps like “Whom did you want ___ to go to the store?”, or perhaps slightly more realistically “Where did you want to go to ___?”. I’m not sure if there are non-question examples of that sort of thing.

  31. David Eddyshaw says:

    Yes there are, for example in relative constructions and clefting.
    The Cambridge Grammar of the English Language, by no stretch of the imagination a Chomskyan work, uses the notion quite extensively. I’ll give some examples from English and Kusaal presently. (Not at a computer just now.)

  32. I can see how the concept can be useful for analysis, but that’s not the same as claiming it clearly exists. Cf. the notorious contortions quantum theorists go through.

  33. David Eddyshaw says:

    I have no axe to grind re the ontological status of gaps (I’m a thoroughgoing philosophical nominalist when it comes to grammar), but I think it really does simplify not so much analysis as the statement of grammar. For example, CGEL p1082 talks about the syntactic function of gaps. using the examples

    What did you buy []?
    What are you referring to []?
    Where did you see them []?
    Who do you think [] was responsible?
    *Whose did you borrow [] car?
    *How many did they receive [] applications?
    *How serious will it be [] a problem?
    *Who have they shortlisted [] and KIm?

    … saying that the gap can only function in such constructions (in English) as a post-head dependent or a subject.

    There are certainly alternative ways of putting the same facts, but within the already existing descriptive framework of CGEL, this is a very neat way of doing it.

    Kusaal is consistently SVO, but there are nevertheless two ways of getting the object first. One happens with objects which are long and complex (“heavy shift”): the object just gets plonked first, and then must be picked up by an object pronoun:

    Gbigimkanɛ sa kʋ niigi la, ti na kʋ o.
    “The lion that killed the cows yesterday, we’re going to kill him.”

    The other way is by putting the object first, followed by ka, which is a foregrounding construction. The object may then not be picked up by an explicit object pronoun:

    Gbigim la ka ti na kʋ.
    “The lion, we’re going to kill him.”

    This despite the fact that kill is obligatorily transitive: the commandment “Thou shalt not kill” in Kusaal has to be expressed:

    Mid ka ya kʋ nid
    “See that you don’t kill a person.”

    So it seems natural to say there is a “gap” functioning as the object in Gbigim la ka ti na kʋ.

  34. True. I just have scars left from my long-ago experiences in the Gulag mandatory transformational-ling class.

  35. Indeed, for me as well a concept exists just in case it is useful for analysis. There is oxygen, and there isn’t phlogiston, because it’s more helpful to understand combustion in one way than in the other. LIkewise, with Dennett I take the self to be a theoretical posit too, myself just as much as any other self.

  36. David Eddyshaw says:

    Man tɛn’ɛsid la pa’al ye m bɛ!

    (“That I am thinking has shown that I exist.” Arini Dikaatu, Kusaasi philosopher.)

  37. January First-of-May says:

    On the subject of Kusaasi philosophy – are lizards actually supposed to be masculine in Kusaal?

    Because I keep occasionally quoting that one Kusaasi proverb about lizards, trousers, and tails, but I have to change the lizard to feminine, because that makes much more sense in the context where I’m quoting it, and I’m not sure how much is this a disruption of proper Kusaasi philosophy.

  38. David Eddyshaw says:

    Like all advanced modern languages, Kusaal does not grammaticalise sex.
    However, women do not wear trousers. Where would it end?

  39. Maybe, lizard is the one who got into someone else’s trousers? January, will you be so kind to enlighten us?

  40. January First-of-May says:

    The context has to do with a maybe-sorta-urban-fantasy-ish story that I regularly comment on. TL/DR is that there’s a kinda-lizardy teenage girl, and, IIRC, she wears trousers because that’s what teenage girls do in the USA in the 2010s, but because she’s kinda-lizardy, she has to include a tail space in said trousers.

    I don’t recall the exact details offhand, but the existence of said tail space was actively plot-relevant at some point, and this is when I started quoting the proverb.

    But the version of the proverb that I knew referred to the lizard as “he”, so I had to change that word to “she” to fit the context. Sometimes I remembered to mention that the original had “he”, sometimes I didn’t.

     
    Lizards do have a grammatical gender in some languages (but not in others, such as English) – but which grammatical gender they have might as well be random; just quoting from Wiktionary, they’re feminine in Russian, German, Italian, and Hebrew, but masculine in French, Spanish, Serbian, and Lithuanian.

    Obviously, I didn’t know enough about Kusaal to tell whether it was plausible for lizards to be masculine in that language.

    In any case, I sincerely apologize for any misuse of Kusaasi philosophy; it certainly was not intended to be disrespectful.

  41. Thanks, but what is the proverb? My google search ended up with such pearls of wisdom as “Just because the lizard nods its head doesn’t mean it’s in agreement” (I guess, it could serve as demonstration of its/it’s distinction) and “a lizard can be caught with the hand, yet it is found in kings’ palaces”

  42. David Eddyshaw says:

    “If you see a lizard wearing trousers, you can be sure that he knows what he’s done with his tail.”

    It is a proverb warning against the Fundamental Attribution Error.
    While human females do not wear trousers, I do not think the rule is strictly applied to lizards. The matter never came up.
    Kusaal pronouns do not distinguish sex.

  43. Lounge lizard is masculine in English.

  44. Wow, I wonder what happened to poor B. Cummings, Sept 3, 2003 10:07pm. Disillusioned, of Arizona.

    He became an Electrical Technician at the LA Times printing facility.

    That knowledge was shockingly easy to acquire, btw.

  45. We know this is a gap because there are similar sentences where the gap is filled, as in “I wanted Alice to go to the store.”

    This seems to me like an example of how Chomskyites are led astray by their lack of knowledge of foreign languages. “I wanted to go to the store” has no gap, and the construction is the same in any Indo-European language – “Ich wollte zum Laden gehen”, “Volevo andare al negozio”, “Chociałem iść do sklepu”, etc. The fact that you can want someone else to go, rather than want THAT someone else goes, is what is odd about English.

    I also cannot see where the gaps are in David Eddyshaw’s examples. That just looks like word order.

  46. “gaps” left by words being moved around is no mere Chomskyan fad but a perfectly real phenomenon

    I would say that already the notion of “words being moved around” IS precisely the core of “the Chomskyan fad”.

    The two basic ideas that all this presumes are (1) one sentence might be more primitive and another being derived from it, and (2) this happens synchronically within the grammar. The first is quite plausible, but then the second not very: syntactic derivation must certainly exist also a diachronically, and then that seems to be an entirely sufficient account of the facts already. If all derivation is instead crammed into synchronic grammar, then “gaps” end up as a sort analogue of null morphemes that can be inserted at will if the evidence doesn’t fit one’s essentially internally reconstructed generalizations about word order.

    There is oxygen, and there isn’t phlogiston, because it’s more helpful to understand combustion in one way than in the other.

    Combustion as such is just as easy to understand either way, it is other concepts of chemistry such as specific gravity or the isolatability of substances where oxygen fits into the established picture and phlogiston does not.

  47. David Eddyshaw says:

    This seems to me like an example of how Chomskyites are led astray by their lack of knowledge of foreign languages. “I wanted to go to the store” has no gap, and the construction is the same in any Indo-European language

    No, it’s not. Check out modern Greek or Welsh for a start. It seems that you may be being led astray by lack of knowledge of foreign languages.

    I am no Chomskyan and neither are the very much more learned editors of CGEL. Them’s fighting words.

    Interestingly, the English modal verbs are often analysed these days as taking “catenative complement” stripped-down clauses, which may or may not have their own independent subjects. I was struck by this because it has interesting analogies with Kusaal constructions which have been traditionally (but wrongly in my view) described as serial verb constructions.

    In any case, however, just because one construction does not lend itself to a useful analysis with “gaps”, it hardly follows that no construction does. You might as well say that the fact that there are no elephants native to Greenland proves that there are no elephants in Africa.

  48. Yes, I forgot Welsh, sorry. Still, even in a sentence like “Rydw i eisiau mynd i’r siop”, is there really a gap? A gap is an ellipsis, and I don’t perceive any missing information in that statement.

    I understand that an utterance like “She went to the party and Tom too” demonstrates a gap. I’m not saying gaps don’t exist, just that I don’t see any in Cowan’s example or in yours and it is unclear to me how those are supposed to be examples. The examples on Wikipedia make a lot more sense to me.

    Is the claim that “I wanted [] to go the shop” contains a gap just because you could add a direct object after “wanted”?

  49. David Eddyshaw says:

    I would say that already the notion of “words being moved around” IS precisely the core of “the Chomskyan fad”.

    Would that the Chomskyans had indeed stopped at that point!
    Far from the case, unfortunately. And they’ve long since abandoned that particular idea, anyway.

    Personally, I find the idea quite intuitive, but I take absolutely no stand on whether (for example) interrogative sentences are synchronically “derived” from assertions by a process involving movement in English, not least because I can’t see how taking a stand would make a blind bit of difference to anything that matters. On the other hand, it seems a bit perverse to refuse on ideological grounds to accept some sort of synchronic derivational connection between e.g.

    “I am not a linguist.” and
    “He says he’s not a linguist.”

    To do so has been a standard grammarian’s explanatory manoeuvre since long before Noam met Zellig. I don’t see the harm in it unless you arbitrarily object to all abstraction in syntactic description on principle.

    However, the idea of “gaps” needn’t actually logically imply any sort of movement anyway. Whether there’s any value in talking about gaps boils down pretty much entirely to whether it elucidates your description of syntax or not. It’s at least worth pondering why people with absolutely no ideological axe to grind might have come to the conclusion that it does. Explaining syntax is hard.

  50. David Eddyshaw says:

    @Vanya:

    The supposed gap represents the embedded subject of the infinitive:

    “I wanted [me] to go to the shop.”

    Gaps are not ellipses, or at least not what is normally meant by that slightly slippery term.

    CGEL (source of all wisdom) has a whole section on “catenative complements” which goes into their rationale for reanalysing English constructions like these in great depth. I found it entirely convincing, not least because of cross-linguistic parallels. I can’t do justice to the argument here, what with (a) not being clever enough and (b) typing on a phone, but there is no whiff of the Chomskys about it at all.

    https://www.thoughtco.com/what-is-catenative-verb-1689832

  51. Lot lizard is feminine in English.

  52. I wanted to go to the store” has no gap

    I protest that to me gap means no more than a place where something could be but isn’t, like a gap in the teeth. No implications of word-shuffling in the brain or elsewhere.

    I wanted to go to the store” has no gap, and the construction is the same in any Indo-European language

    I think this might be part of the pervasive effects of the European Sprachbund on linguists’ thinking. German, along with French, is a nuclear Sprachbund language, and the others you exemplify are at least part of the core. (Greek is also core, but Welsh is definitely outside.) You really have to look at the Indo-Iranian languages before you generalize like that (not that I have).

    The fact that you can want someone else to go, rather than want THAT someone else goes, is what is odd about English.

    Although there are a handful of earlier examples, English seems to have borrowed the accusative and infinitive construction directly from Latin, probably as part of the revival of learning.

  53. English seems to have borrowed the accusative and infinitive construction directly from Latin, probably as part of the revival of learning.

    I thought of that, but it seems very unlikely. Isn’t that construction more likely a Celtic substrate?

  54. Yeah, I find it a little hard to believe that a learned grammatical borrowing from Latin would become part of basic grammar.

  55. David Eddyshaw says:

    The construction is not the same in modern Greek, which doesn’t have an infinitive. I believe it’s a Balkan Sprachbund thing.

  56. David Eddyshaw says:

    The English “accusative and infinitive” (so-called) can hardly be from a Celtic substrate: there’s nothing comparable in Insular Celtic.

    The (colloquial modern) Welsh example Vanya gave is actually calqued on English, btw: eisiau is actually a noun, not a verb, modal or otherwise.

  57. David Eddyshaw says:

    (And I’ve just discovered it’s a loanword from the Latin exiguus. I Did Not Know That.)

  58. David Eddyshaw says:

    Before the Romans came, my forebears evidently not only did not have children (plant, from Latin) but also lacked the concept “want” (eisiau, also from Latin.) An earthly paradise. Curse you, Julius Caesar!

  59. Stu Clayton says:

    But they wanted for children, by your own analysis. Caesar deserves a break here.

  60. How did one say “I want you to go to the shop/well/smithy” in medieval Welsh?

  61. On the other hand, it seems a bit perverse to refuse on ideological grounds to accept some sort of synchronic derivational connection (…) I don’t see the harm in it unless you arbitrarily object to all abstraction in syntactic description on principle

    I have little objection to any arbitrary amount of epicycles as long as this is kept within syntactic description and not reified to an entity existing in language itself.

    I protest that to me gap means no more than a place where something could be but isn’t, like a gap in the teeth.

    Ah, following Subtractive Universal Grammar you will then no doubt agree that e.g. “I protest that…” also has an adverb gap as can be proven by the feliciticity of “I loudly protest that…”, a conjuction gap as proven by “I and my mother protest that…”, which in turn has, among others, two adjective gaps for each member of the conjunction, a grand- prefixation gap in the latter member, an /or gap in the conjunction itself…; also several of these are in fact indeed parasitic gaps, as in e.g. “I and my mother protest in unison that…” but **”I __ protest in unison that…”

  62. I am agape!

  63. PlasticPaddy says:

    @DE
    I do not know if this is relevant to Welsh but in Irish “want” is a bit elusive. You never “want” someone else to do something: you “would like” this.
    Ba mhaith liom go ndéanfá suas d’intinn féin = i want [would like] you to make up (dreadful english calque but ok) your own mind
    In the past you have to either use
    “I asked/was looking for you to” (polite) or “I needed you to” (rather less so) or something like “I would have liked” or more pragmatically “I wanted whatever it is done, without mentioning you at all”.

  64. Stu Clayton says:

    I am agree !

  65. David Eddyshaw says:

    @Vanya:

    I am not fluent in Mediaeval Welsh since I left Arthur’s court, but I’d say it’s something like

    Mi a vynnaf vynet ohonot i’r ffynnawn.
    “I want you to go to the well.”

    The main point is that Welsh doesn’t have infinitives; it has a ubiquitous verbal noun, which turns up all the time and is the basis of all sorts of periphrastic tenses, but the subject of it is either expressed as a possessive or with a following preposition.

  66. David Eddyshaw says:

    I have little objection to any arbitrary amount of epicycles as long as this is kept within syntactic description and not reified to an entity existing in language itself.

    We are then in complete agreement (see my remark on nominalism above.)

    I am glad to see that you are progressing in the way of SUG, my Padawan. But beware of the Dark Side!

  67. ə de vivre says:

    But “eisiau” even less so, since (at least in ‘good’/not-too-English-calqued Welsh) it doesn’t fill the “yn” slot after the main verb: “Dw’n hoffi caffi” (I like coffee) vs “dwi eisiau caffi” (I need [a] coffee).

    But whither dialectical “moyn”?

  68. David Eddyshaw says:

    (Y)mo(f)yn seems to be a well-behaved verbal noun, at least. Not like those camouflaged Englishmen eisiau and angen. I imagine it’s related to gofyn and thus to the highly respectable mynnu that I mediaevalised above.

  69. David Eddyshaw says:

    Quirk and Wrenn’s Old English Grammar describes “accusative and infinitive” constructions:

    ne hyrde ic snotorlicor guman þingian
    “I have not heard a man speak more wisely”

    do hit us to witanne
    “make us learn it”

    het his heafod ofaslean
    “ordered his head to be cut off”

  70. If you are really interested:

    Alternative expressions of “want” complements.

    It doesn’t work in Japanese…

  71. ə de vivre says:

    Ah, the GPC gives the chain go-mynnaf > gofynnaf; ym-gofynnaf > ymofynnaf > moyn. All from the PIE root *mendh-.

    Also with respect to moyn.

  72. David Eddyshaw says:

    @Bathrobe:

    Nor in Kusaal. Bɔɔd “want” takes either a noun-phrase object or a purpose-clause, and in Kusaal such clauses never delete the subject pronoun regardless of its reference. There are plenty of cases where Kusaal consistently does omit same-subject pronouns: after ka “and” in narrative this is so consistent that if a pronoun is present it signals a change of subject by default. The Kusaal facts just fall out inevitably from the syntax of the construction that bɔɔd takes, and semantics doesn’t come into it. (Same as modern Greek, in fact.)

    I doubt that speakers of Uto-Aztecan languages really have a radically different notion of “wanting” from speakers of Kusaal (or Greek.) It seems to me that the author is in danger of “explaining” different “want” constructions by attributing subtly different meanings to them without any very clear independent evidence for those meaning differences. Looking for that difference in terms of whether the subjects are the same or different looks awfully like a circular argument.

  73. In Japanese different forms are used.

    ikitai (I) want to go.

    itte hoshii (I) want him/you to go.

    Mongolian doesn’t have a single standard form to express wanting.

    I suspect the Anglocentrism of the generativists is somehow involved in all this.

  74. David Eddyshaw says:

    You’ll know orders of magnitude more than I about it, but I seem to recall that Japanese regards mental states in general as inaccessible to everyone except their owner, and this is grammaticalised, so “I am happy” but “he seems happy.” So the “wanting” difference in Japanese could reflect a broader phenomenon.

    I wonder if some of these Uto-Aztecan languages do something similar?

  75. That’s ikitagatte iru you’re talking about.

    It means ‘he is showing signs of wanting to go’. Yes, ikitai does express one’s internal state, and you can’t express someone else’s internal state (because you aren’t them), but I think there are grammatical contexts where you could use ikitai of another person (surmise, quoting). Ikitagatte iru means only that you have witnessed expressions of interest and you’re pretty sure that that represents his internal state. It does not mean ‘He wants to go’; it means ‘He has shown strong signs of wanting to go (or he has said he wants to go)’.

    It seems to me that you could say Kare wa ikitai deshō ne ‘I guess he would want to go’, surmising about that person’s feelings, which would be different from Kare wa ikitagaru deshō ne ‘I guess he would indicate a desire to go, be keen to go’.

    The Japanese difference I was talking about is that iki-tai expresses the desire of the person speaking using the suffix -tai, while itte hoshii uses the verb hoshii meaning ‘to want’. Other verbs could be used, like itte moraitai. Morau means ‘receive from someone’, -tai is ‘want’. So the meaning is ‘I want to receive that [you] go’. Or Itte morau, which could virtually be a command meaning ‘you go’ (literally ‘I will receive that [you] go’).

    So the aspect that you mention is relevant but in terms of form they are quite different.

  76. I should add to the above, which I wrote as I was thinking:

    Hoshii is similar to -tai in that it represents the internal feelings of the speaker.

    What is significant is that a different construction, including a different verb, is used for the two (‘I want to go’, ‘I want you to go’). When speaking of someone else’s desires you would have to add a distancing mechanism such as quoting them or describing their behaviour objectively. Hoshii is no different from -tai in that respect.

    And the other point is the use of the verb morau or similar, in the expression itte moraitai. You could say that it is really the same as English if you strain out all the inessential parts, and that the only difference is the existence of the suffix -tai that attaches directly to the verb, obviating the need for a verb meaning ‘want’, but it’s a bit different from debating whether the subject of the clause you go is ‘gapped’ or not.

  77. David Marjanović says:

    The construction is not the same in modern Greek, which doesn’t have an infinitive. I believe it’s a Balkan Sprachbund thing.

    “I want that I go” is Balkan.

    Quirk and Wrenn’s Old English Grammar describes “accusative and infinitive” constructions:

    The first works exactly the same way in German. Same for “see”, and I think that’s the complete list of verbs that tolerate AcI in German.

    The second… “it” is the accusative, and “us” is a dative. In the very archaic German cognate tu es uns zu wissen, I’ve always interpreted uns as a dative. “Put it to us to know”, basically.

    The third – no, that’s active, not passive: “he told [his audience in general] to cut his head off”. Er befahl (archaic: hieß), ihm den Kopf abzuschlagen. The head is not the A of an AcI, and the order is given to a missing dative, not a missing accusative.

  78. David Eddyshaw says:

    Quirk and Wrenn cite het his heafod ofaslean specifically in a context where they say “we often find the infinitive used with passive meaning”, cross-referencing other examples of passive use like heht hine læran “commanded him to be taught” and þas þing sint to donne “these things are to be done.” Of course, they may be wrong; the first example in particular looks susceptible to the same kind of reanalysis you suggest.

  79. David Marjanović says:

    Diese Dinge sind zu tun “these things are to be done” isn’t grammatically passive in German either. They probably contain a gap where some “for”-construction or dative has fallen out: “these things are for us/somebody to do”.

    hine is accusative, right? In that case it must be “ordered to teach him” – hieß, ihn zu lehren.

  80. David Eddyshaw says:

    it’s a bit different from debating whether the subject of the clause you go is ‘gapped’ or not.

    Sure. I brought it up because I thought it might shed light on the sort of meaning differences Guerrero asserts in “want” concepts which she tries to correlate with different constructions, some of which involve “gapping” if you want to call it that. There doesn’t really seem to be enough data there to say: More research needed!

    Given that Japanese personal pronouns are poor marginalised creatures avoided in polite conversation anyway, any invocation of “gapping” in Japanese would have to involve quite stratospheric levels of abstraction, and J Pystynen would be gravely displeased. I’m not prepared to take that risk.

  81. David Eddyshaw says:

    In Kusaal you can use bɔɔd “want” with an abstract noun regularly derived from a verb as object (Kusaal doesn’t have infinitives either), but it means “be about to”, not “want to”, and the subject doesn’t even have to be capable of volition:

    Tiig la bɔɔd liig.
    “The tree is about to fall.”

    Bɛog bɔɔd nier.
    “Morning is about to appear.”

    not *Bɛog bɔɔd ye li nie *”Morning wants to appear.”

    [Actually, you probably could say Tiig la bɔɔd ye li lu “The tree wants to fall” in an appropriate context; in the traditional Weltanschauung trees can be both personal and spiritually powerful. I expect living in Sudanian savanna gives you a different perspective on these things.]

  82. Stu Clayton says:

    Diese Dinge sind zu tun “these things are to be done” isn’t grammatically passive in German either. They probably contain a gap where some “for”-construction or dative has fallen out: “these things are for us/somebody to do”.

    That damned “gap” again. What happened to “ellipsis” ?

    I see sinning against Pystynen principles here, reification right and left. It’s easy to imagine some top dog courtier inventing these German locutions from whole cloth: not as a gapping or ellipsiotomy of an eternal sentence, but as a polite (veiled, threatening) equivalent of one – equivalent in the sense of conveying the same information and commands. There is no need to reify an “original sentence” that has acquired holes.

    Diese Dinge sind zu tun can be seen as an equivalent to, say, Diese Dinge sind euch aufgegeben zu tun. The latter may precede the former in time, but not in being and dignity. It can be taken out for disciplinary or explanatory purposes, otherwise it’s kept in the attic.

    It’s just as easy to imagine the “shorter” form as the original form, traditional in the days of oracular rule. Prove it ain’t so. Maybe there’s no temporal precedence at all, at least not one measurable in units of length. I could live with that.

  83. David Eddyshaw says:

    @ə:

    V moyn T

    Bah. Kids today.

  84. but it means “be about to”, not “want to”

    Cf. will in English vs. German.

  85. Khalkha Mongolian -маар байна -maar baina originally meant something like ‘about to’. Now it means ‘want to’. (Still means ‘about to’ in Inner Mongolian.)

  86. David Marjanović says:

    hine is accusative, right?

    *facepalm* Completely irrelevant, if Old English was like 18th-century literary German in this respect: heißen goes with the accusative anyway, so heht hine læran is perfectly ambiguous between hieß ihn(,) zu lehren “told him to teach” and hieß, ihn zu lehren “said/ordered to teach him”.

    That damned “gap” again. What happened to “ellipsis” ?

    I’m perfectly fine with “ellipsis” here. We’ve just been talking about “gaps” much lately, and I didn’t sleep well. 🙂

    It’s just as easy to imagine the “shorter” form as the original form, traditional in the days of oracular rule. Prove it ain’t so. Maybe there’s no temporal precedence at all, at least not one measurable in units of length. I could live with that.

    Me too.

  87. Stu Clayton says:

    Could someone please pass me a reliable hint as to whether “Pystynen” is stressed on any particular syllable(s), and if so is it the first, second or third, or a combination of the aforementioned ?

  88. David Eddyshaw says:

    That damned “gap” again. What happened to “ellipsis” ?

    Ellipsis usually means omitting something redundant, in the sense that the omission makes no difference to the original meaning. It varies from very informal (“Can’t see the point”) to quite stylised (“Thank you”), where you could reasonably argue that calling it ellipsis at all is not appropriate synchronically.

    Gaps aren’t like that: if you fill the gap you either get an ungrammatical sentence or you change the original meaning.

    So in English, “I want me to go to the shops” is weird, whereas “I want you to go to the shops” is normal (and less effort.)

    In Kusaal (I checked this specifically with a very clued-up informant):

    Pu’a la da’ daka ka [GAP] kɛŋ Bɔk.

    is “The woman has bought a box and gone to Bawku”, whereas

    Pu’a la da’ daka ka o kɛŋ Bɔk.

    is “The woman has bought a box and it’s gone to Bawku.”

    with the implication of subject change being so strong that it overrides the fact that boxes are inanimate and would normally be referred to with the pronoun li “it”, not o “he/she” (even this residual natural gender system is tending to break down in modern informal speech, which is how this interpretation gets to be possible at all.)

  89. Trond Engen says:

    Pystynen is Finnish. Initial stress.

  90. January First-of-May says:

    Could someone please pass me a reliable hint as to whether “Pystynen” is stressed on any particular syllable(s), and if so is it the first, second or third, or a combination of the aforementioned ?

    Wikipedia says that stress is non-phonemically initial in Finnish, so presumably the first. I’m not entirely sure, though.

    [EDIT: ninja-ed by Trond Engen.]

  91. PlasticPaddy says:

    There is a sort of inverse gap in German, where the verb at the end of a clause is uniquely constrained by some of the preceding words. For comic effect a different verb can be substituted.

  92. Bɔk … Bawku

    Presumably the former is a reduced form of the latter; is Bawku from a different language? Wikipedia tells me the kusaasi are the indigenous inhabitants of the area.

  93. David Eddyshaw says:

    Ah! I’m glad you asked me that! Plagiarising my own work with a shameless cut-and-paste:

    When speaking English or French, Kusaasi normally cite Kusaal personal and place names without apocope [ref to relevant subsection of grammar]: À-Wɩ̄n from Wɩ̀dɩ̀-n̆yá’aŋ will introduce himself as “Awini” from “Woriyanga.” Similarly “Kusaasi” for Kʋ̄sâas, “Bawku” for Bɔ̀k etc. “Woriyanga” also shows a Mampruli rather than Kusaal form for the initial combining form “horse”: Mampruli wuri-, Kusaal wɩ̀d-. This reflects the origin of the convention in the use of Mamprussi guides and interpreters by the British in their initial explorations of the area. A parallel development took place earlier in Mamprussi country when the British arrived with Dagomba guides: thus “Gambaga” (Dagbani Gambaɣa) for the Mampruli place name Gambaa (Naden.) The convention has been generalised by analogy, and many forms show distinctively Kusaal phonology, morphology or vocabulary. Simple reproduction of Kusaal forms is also sometimes seen, e.g. “Aruk” for the personal name À-Dʋ̄k, and the language name “Kusaal” Kʋ̄sâal itself.

  94. Thanks! So the spelling Bawku represents /bɔku/?

  95. David Eddyshaw says:

    The actual Mampruli form is /bɔkku/; in English, people say [bɔ:’kʰu:]. Like most local place names, its meaning is transparent: it has the uplifting translation “pit.” Mampruli and Kusaal are closely related, so the relationship between a Mampruli word and its Kusaal equivalent is usually obvious to locals. However, some placenames meaning nothing in Mampruli and the English-context forms are just artificially Mamprulicated by analogy: “Tilli”, for example, is not a Mampruli word; the name is from Toende Kusaal til “tree trunk.”

    The ‘apocope’ thing I refer to is the major distinguishing mark of Kusaal over against its close relatives: in most contexts, including citation, words appear with their final vowels deleted and any resulting final consonant clusters simplified. However, this is a synchronic thing: the full forms still appear in certain contexts, such as the end of clauses involving a negation:

    Pu’a la pʋ kɛŋ Bɔkɔ.
    “The woman hasn’t gone to Bawku.”

    This means that the underlying final vowels are still accessible to Kusaal speakers, making it that much easier for them to match up the forms with Mampruli, which doesn’t have apocope. (Most Kusaasi cannot actually speak Mampruli.)

  96. Fascinating, thanks very much!

  97. Sounds sort of like the relationship between Finnish and Estonian (as I think somebody already remarked in another thread).

  98. David Eddyshaw says:

    In many respects Kusaal seems to be particularly conservative within Western Oti-Volta (though this may be partly the illusion someone here once remarked on: the language you know best always turns out to be the most conservative.) It preserves the original vowel system almost completely intact in root syllables, and the Agolle Kusaal dialect and Mooré are the only languages of the whole group which preserve inherited /r/ as a separate phoneme (but its loss elsewhere is probably late, as the reflexes differ even between very closely related languages.) I think Kusaal is probably unique in preserving a whole distinct conjugation of mostly-stative verbs, though it’s difficult to be certain of this given the underdescription of most of the languages. Mooré has kept more of the original noun class system intact though, and Farefare, Talni and Mampruli have partly maintained the old grammatical gender system, which all the other languages have ditched.

    It would make some sense for Kusaal to be conservative overall if the spread of Western Oti-Volta was largely the result of the expansion of the Mossi-Dagomba empires; the original capital was in what is now Kusaasi territory (though it was relocated south a generation later after successful Kusaasi and Bisa revolts.) The timescale doesn’t seem to work out very well: the original empire was probably founded about seven hundred years ago, and the Western Oti-Volta languages are collectively about as different from one another as the Romance languages. Still, there’s no law that languages have to change at the exactly the same rate in West Africa and in Western Europe. There might well be substrate effects involved too.

  99. It would make some sense for Kusaal to be conservative overall […]; the original capital was in what is now Kusaasi territory

    It seems to vary. Italian is certainly conservative, but on the other hand RP is innovative.

  100. David Eddyshaw says:

    Yeah, I’m not dead set on the idea myself, and there are quite a number of problems with it in this particular case.

    The Dagomba and Mossi kingdoms by common consent are cadet branches of the Mamprussi state, and while Dagbani and Mampruli are strikingly innovative phonologically, Mooré is very conservative. Mooré is also on the other side of the major split in Western Oti-Volta between northern and southern groups. So the political and linguistic histories don’t match well.

    Furthermore, the original Mossi-Dagomba chiefly elite probably spoke a different language (or languages) from their subjects (still the case with the neighbouring Gonja kingdoms.) Mossi-Dagomba chiefs still speak to their subjects on official occasions via a “linguist”, despite there being no actual language difference; however, this is a pan-Ghanaian thing (the Ashanti do it too) and one should probably not draw too many conclusions from it.

  101. David Eddyshaw says:

    I see that the Wikipedia entry for “Ellipsis” cites Anne Toner’s Ellipsis in English Literature: Signs of Omission; while this is not in the same league as the philosopher J Austin’s decision to call his work Sense and Sensibilia, I feel it deserves an honourable mention. The only consistent fix of horrible intellectual puns I get for the most part is picture captions and article subheadings in the Economist.

  102. Quirk and Wrenn: “we often find the infinitive used with passive meaning”

    Does that explain why LIHT MEC HEHT GEWYRCAN means “The Light ordered that I should be made”? (Written on the Sign of Fire in The Dark Is Rising, inspired by a real Anglo-Saxon artwork inscribed AELFRED MEC HEHT GEWYRCAN.)

  103. David Eddyshaw says:

    Well, I would say yes; David M will presumably say that there is a dative pronoun referring to the actual artisans lurking in the ginnunga-gap.

  104. David Marjanović says:

    Heh. Where else!

    Yes, I do think this is “Ælfred ordered to make me” – not idiomatic in modern English, but still unremarkable in German.

    (…other than the fact that it has fallen almost completely out of fashion for objects to talk about themselves.)

    Fascinating, thanks very much!

    + 1

  105. Stu Clayton says:

    Gaps are mere visions fugitives, fleeting glimpses of the Ur-Dative from which all explanations flow.

    Glad I got that figured out. Couldn’t have done it without you guys, though !

  106. David Eddyshaw says:

    the Ur-Dative

    It’s ra.

    https://en.wikipedia.org/wiki/Sumerian_language#Nominal_morphology

    If only Thor Heyerdahl could have known!

  107. David Eddyshaw says:
  108. Stu Clayton says:

    I see a steep learning curve that appears to terminate abruptly. This is not for the masses.

    # Additional spatial or temporal meanings can be expressed by genitive phrases like “at the head of” = “above”, “at the face of” = “in front of”, “at the outer side of” = “because of” etc.: bar udu ḫad2-ak-a = “outer.side sheep white-genitive-locative” = “in the outer side of a white sheep” = “because of a white sheep”. #

  109. Stu Clayton says:

    How did white sheep get on the agenda ? Help me out here, David. I had believed the completive of “omit unnecessary” to be “white sheep”, yet now they appear to have been paradigmatic in Sumerian. Not a completive but a becausative.

  110. Omit needless sheep!

    There is only one god / He is the sun god / Ra! Ra! Ra!

  111. Stu Clayton says:

    Sis Boom Ba !

  112. David Eddyshaw says:

    The People’s Republic of Wales indignantly repudiates the hegemonic Saxon concept of “Needless Sheep”, which is offensive to the feelings of our citizens.

  113. a reliable hint as to whether “Pystynen” is stressed on any particular syllable(s)

    You can hear Tiina Pystysen at the very beginning of this video:

    https://yle.fi/aihe/artikkeli/2010/01/14/kirjailija-ja-graafikko-tiina-pystynen

  114. Stu Clayton says:

    Thanks, juha. In speaking she seems to use a number of words that end in “ta”, this being uttered fairly forcefully after a brief pause/stop. Like a kind of emphasis, but at the other end of the word from what I’m used to.

  115. David Marjanović says:

    I think that’s -tta.

  116. which is offensive to the feelings of our citizens

    Do you mean the citizens belonging to Homo sapiens, or the others?

  117. Surely all citizens of the People’s Republic of Wales find it offensive, humans and ruminants alike.

  118. David Eddyshaw says:

    Indeed. We in the PRW have no truck with reactionary ovinism.

  119. a number of words that end in “ta”

    https://en.wikipedia.org/wiki/Partitive_case#Finnish

  120. David Eddyshaw says:

    In re my vague speculation about the spread of the Western Oti-Volta languages perhaps being related to the political expansion of the Mossi-Dagomba states: I thought that one difficulty was that the foundation of those empires was only about seven hundred years ago, and it seems difficult to square that with the current Romance-level diversity of the languages.

    However, I was just reading Juha Janhunen’s Mongolian, wherein he points out that despite the various Mongolian languages all being descended from the koine of Chinggis Khan’s empire, they differ very considerably from one another, including two (Mongolian and Santa) “structurally almost as different […] as is possible within the context of a single language family.” So I spoke too soon.

    It also occurred to me somewhat belatedly that the author may very well be our Juha (see above.)

  121. I don’t think it’s our Juha… but I could be wrong.

  122. Lars (the original one) says:

    Juha was the John of Finnish in the mid and late 19th, so a coincidence is very likely. (Before and after that it was Juho, it seems).

    Most Danes (or Germans) called Lars are not me, and most Germans (or Danes) called Hans aren’t our Hans either.

  123. Stu Clayton says:

    Most Danes (or Germans) called Lars are not me

    I see. That explains the need to distinguish yourself as “the original one” !

  124. SFReader@yahoo.com says:

    Santa is a seven centuries old Mongolian-based creole. You might as well wonder why French and Haitian Creole are so different. Or English and Saramaccan.

  125. (Mongolian and Santa) “structurally almost as different […] as is possible within the context of a single language family.”

    Santa (and a few other neighboring languages) is a seven centuries old Mongolian-based creole.

    It is spoken by a racially mixed group of Sunni Muslim farmers. Their ancestors are believed to have been brought by Mongols to the Gansu province of China from Muslim Central Asia in 13th century (as slaves I presume) and originally they probably spoke various Iranian or Turkic languages. In Gansu they further were intermixed with people speaking Tibetan (and likely Tangut too) and several varieties of Chinese. And as Muslims, they were of course heavily influenced by Arabic.

    All these languages on top of imperfectly learned Mongolian which initially served as common lingua franca.

  126. Re: Juha Janhunen

    The prominent Finnish Mongolist for some reason looks Mongolian.

    Maybe if you study a language long enough you’ll start looking like a native speaker.

  127. David Eddyshaw says:

    Hasn’t worked for me so far. I could have saved a fortune on sunblock.

  128. David Eddyshaw says:

    Santa is a seven centuries old Mongolian-based creole. You might as well wonder why French and Haitian Creole are so different. Or English and Saramaccan.

    I got the impression from Janhunen’s book that even the core Mongolian languages differ quite markedly from one another, but several Hatters will Actually Know and can tell me.

    In the case of Western Oti-Volta, a fair bit turns on what one means by the all-too-malleable term “creole” (exhibit A, John McWhorter.) None of the languages is creole-like in the way that Haitian or Tok Pisin or Nigerian Pidgin or KiNuba is; on the other hand, the entire group shows loss of several noun classes morphologically, verb flexion is the simplest in all of Oti-Volta after Buli/Konni (which hasn’t got any) and shows abundant signs of levelling, and most languages have lost grammatical gender altogether. The trouble is that the most striking simplifications apply to all the Western Oti-Volta languages (except the geographically isolated Boulba, as far as I can tell), so they either were already there in the protolanguage or have spread across the distinct languages later, which seems to imply a rather more complex dynamic than the sort of language-contact-induced simplification that McWhorter likes to posit. “Contagious creolisation”, as it were.

  129. Trond Engen says:

    Same-lexifier creoles can have similar results on different substrates. I’ll suggest a two-step creolization where the lexifier of the attested local creoles was itself a creole, e.g. an imperial army language or a regional lingua franca. I think this is pretty mainstream in the explanation of the various Romance-based creoles, and also Afro-Caribbean English-lexifier creoles and the various West African “Pidgins”? Wihout modern communications, local langiuages in Papua New Guinea might have been creolized with Tok Pisin rather than replaced by it. Actually I suspect they are, even if these local creoles end up being absorbed rather than establishing themselves as independent systems.

  130. even the core Mongolian languages differ quite markedly from one another

    Not really. Just Scots/English, Scandinavian, East Slavic, Spanish/Portuguese level of differentiation.

    Everything outside of Gansu-Qinghai languages, Dagur and Moghol is basically one language (or not very wide dialect continuum with several codified literary languages).

  131. David Eddyshaw says:

    I’ll suggest a two-step creolization where the lexifier of the attested local creoles was itself a creole

    With the proviso that Proto-Western-Oti-Volta wasn’t very creole-like sensu stricto, and that the various modern Western Oti-Volta languages aren’t either, maybe yes.

    Not only is the diversity of the languages within Oti-Volta concentrated in the east, so is (broadly speaking) the morphological complexity of individual languages. The (perhaps) left-behind-in-Benin Western-Oti-Volta language Boulba hasn’t participated in the reduction of noun classes that the rest of Western Oti-Volta has, and as far as I can tell from the minimal data available it has full-blown gender agreement to a degree unknown even in such of the other languages as retain grammatical gender at all. So the Western-Oti-Volta branch that went to what is now Ghana and Burkina Faso (if that’s what happened) was already somewhat stripped-down morphosyntactically.

    It’s tempting to attribute the sweeping vowel system simplifications seen in Mampruli and Dagbani to speakers of Gurunsi languages balking at getting their tongues round nine distinct vowel qualiities with distinctive length, nasalisation and glottalisation. The Gurunsi group is divided in two by a wedge of mainly Mampruli-Dagbani type Western Oti-Volta, so it’s no great stretch to imagine Gurunsi as a substratum there. However, most extant Gurunsi languages have distinctly more complex vowel systems than Mampruli-Dagbani.

    Gurma languages would be another possibility for substrates, but again, most have more complex vowel systems than Mampruli-Dagbani. The Gurma language in geographical contact with Kusaal and Mampruli, Moba, looks very much like it’s been influenced by Western Oti-Volta, rather than the other way about. The Gurma languages, moreover, all have grammatical gender systems which are very much alive and well.

    Mooré also rather spoils the picture by being spread over a wide area in the north, and also being strikingly conservative in phonology; however, there are independent historical reasons for thinking Mooré has expanded relatively rapidly and relatively recently, and that the area was not previously heavily populated, so you could argue that substratum influences would be less anyway.

    My main concern is that although it’s fairly easy to make quite nice just-so stories, which may even be true, it’s difficult to see how any of it could be really testable (much the same objection as I have to John McWhorter’s “explanation” of Kwa, in fact.)

  132. David Marjanović says:

    What are all these vowel systems like? I’m reminded of East Yiddish having precisely those 5 vowels that Polish (6) and German (10 minimum) have in common.

    Similarly, how do the gender systems work? McWhorter’s argument about how the Vikings burned the English one to the ground is that lots of obvious cognates (usually slightly different derivatives of the same roots) have unpredictably different genders in North and West Germanic, so people basically got confused and gave up except, notably, for social gender, where the systems of course agreed.

  133. David Eddyshaw says:

    The commonest vowels systems locally have seven basic vowels, usually more or less a e o ɪ ʊ i u, very often actually written a ɛ ɔ e o i u respectively. Proto-Western-Oti-Volta had the nine a ɛ ɔ e o ɪ ʊ i u. Agolle Kusaal still does, structurally; the original ɛ ɔ are actually realised as [ja] [wa] when short, [iə] [uɵ] when long, but nevertheless behave throughout like monophthongs. The original e o have expanded into the vacated phonetic space and are realised [ɛ] [ɔ].

    All the languages have distinctive vowel length; many have distinctive nasalisation. Glottalised vowels appear exclusively in the contiguous Western Oti-Volta languages Farefare, Talni, Nabit and Kusaal, but were definitely features of the Western Oti-Volta protolanguage, and correspond regularly to sequences in the much less closely related Nawdm; despite the absence of anything like this in any other Oti-Volta languages, the feature must go all the way back to the Oti-Volta protolanguage: it’s just been repeatedly independently lost in the vast majority of languages – unsurprisingly in view of its unusual character.

    All the languages show very marked positional prominence, with the full set of vowel distinctions found only in root syllables. Affixes typically show only a three-way a i u contrast, where even the i/u contrast may be secondary. Epenthetic vowels usually show no contrasts at all, but may develop them secondarily, as they do in Kusaal because of allophone-conditioning final vowels being deleted by apocope.

    The gender systems are all inherited and ultimately cognate; cognate nouns usually have the same gender in the different languages unless simplification of agreement has led to complete loss of that gender. The system is well preserved in Gurma, much reduced and often absent in Western Oti-Volta, and variably reduced in Gurunsi largely as a result of the tendency of that group to fairly marked phonological attrition.

    Gurma is quite far from Western Oti-Volta in terms of obviously cognate vocabulary, not to the degree that distinct branches of Indo-European are, but much more than French vs Romanian, say; Gurunsi is much more distant, about on the level of Germanic vs Slavonic. So it is indeed the case that if (say) a Gurma speaker was imperfectly acquiring a still gender-rich Proto-Western-Oti-Volta, the gender systems would not be at all helpfully congruent; all the more so for a Gurunsi speaker.

    The equivalent of “social gender” would be the distinction between the “human” ba-plural noun class and all the others, which occurs in some form throughout all the languages; languages like Kusaal which have abandoned grammatical gender have a natural gender system contrasting “human” (or sometimes “big and animate”) gender with pronouns based on the old human gender/class, and “non-human/inanimate” with pronouns based on the numerically largest of the remaining noun classes.

  134. David Eddyshaw says:

    I forgot to add the fairly central fact the the Mampruli-Dagbani-Hanga-KaMara-UncleTomCobbley subgroup has reduced the Proto-Western system to a five-vowel a e i o u with contrastive length but no nasalisation or glottalisation; it’s a particularly small system in local terms.

    Dagbani has partly lost contrastive vowel length, too; long vowels in originally open syllables have become short, so you get sana “stranger” vs plural saamba, and maani “okra” vs plural mana, where the singular has simplified an older consonant cluster: *maanni.

  135. Stu Clayton says:

    the original ɛ ɔ are actually realised as [ja] [wa] when short, [iə] [uɵ] when long, but nevertheless behave throughout like monophthongs. The original e o have expanded into the vacated phonetic space and are realised [ɛ] [ɔ].

    A gap !

  136. David Eddyshaw says:

    No longer. Phonology abhors a vacuum.

  137. Stu Clayton says:

    So “words” are the segments between gasping for air ?

    I’ve just realized that phonology explains how vacuum cleaners work.

  138. David Marjanović says:

    long vowels in originally open syllables have become short

    That is amazing.

  139. In speaking she seems to use a number of words that end in “ta”

    0:33 tota part. sg. of toi ‘that’ [used here as a filler word]
    0:39 julistaa ‘to proclaim’
    0:40 että ‘(so) that’
    0:50 pitää 3PS of ‘to hold’
    0:51 huolta part. sg. of huoli ‘care’
    0:53 niistä elative pl. of se ‘that’ (nom. pl. ne)
    1:05 sitä työtä part. sg. of se ‘that’, työ ‘work’
    1:06 niitä part. pl. of ne
    1:08 joilta ablative pl. of joka ‘who’
    1:09 alta lative of ala ‘under’
    1:17 tosta kirjasta elative sg. of toi, kirja ‘book’
    1:20 mielestä elative sg. of mieli ‘mind’
    1:43 kauneutta part. sg. of kauneus ‘beauty’

    (No relation, incidentally, and even her ex-husband Harri who died young is from a fairly different branch of the Pystynen family tree.)

    long vowels in originally open syllables have become short

    That’s amusingly backwards, but sure, why not (something something bimoraic feet).

  140. David Eddyshaw says:

    What I think happened in Dagbani was that a three-way length distinction has been collapsed into two: long vowels in closed syllables having been extra-long. That’s less far-fetched than it sounds: in Dagbani only long vowels in closed syllables can carry two tonemes, all other vowels, whether originally short or long, carrying just one.

    The Kusaal tone system is actually quite similar, though I’ve found it leads to a simpler synchronic description if you make the Kusaal tone-bearing-unit the syllable. Kusaal has also contrived to create a three-tone system out of the local industry-standard two-tone-with-terracing-and-emic-downsteps system, which Dagbani still keeps.

  141. Stu Clayton says:

    Thanks, JP. But

    1:09 alta lative of ala ‘under’

    seems to be slightly defective. I can’t fit it in with this. What is the missing part of “lative” ? This is my first Finnish lesson.

  142. It is indeed defective: -ta nonproductively appears in this and a few other postpositions in its old local case sense, in this case giving the meaning ‘from under’. There is no standardized case terminology for these really, sorry for the confusion.

    (Other examples that remain in use: kaukaa ‘from afar’, kotoa ‘from home’, luota ‘from the vicinity of’, takaa ‘from behind’, ulkoa ‘from outside’, yltä ‘from above’; they have also likewise fossilized locatives in -na: kaukana ‘afar’, kotona ‘at home’, luona ‘at, in the vicinity of’, takana ‘behind’, ulkona ‘outside’; also alla ‘under’, yllä ‘above’ from *alna, *ülnä).

  143. Stu Clayton says:

    No problem, grammatical categories are just stupid crutches as far as I’m concerned. The sooner you cast them aside, the better. The patterns I see at work in Finnish make good sense.

  144. David Eddyshaw says:

    As I have no doubt that many (perhaps double figures?) will be interested in how you can get a three-tone system (in this case, only apparently so) with only two tones (and just because it’s a nice paper and I felt like plugging it):

    http://llacan.vjf.cnrs.fr/PDF/Mandenkan56/56morris.pdf

    Bisa is the language to the immediate north of Kusaal; it’s a Mande language, and thus related to Kusaal only distantly, if at all. Kusaal has produced three tonemes from two by a different route. In both languages, though, the underlying L tone always surfaces as a falling tone before pause, whereas the underlying H tone is always level.

    The Bisa word sáánà “stranger” cited in the paper, is a loanword, appropriately enough (from Mooré.)

  145. I don’t know why I find the Mande languages so attractive, but I do. I once started teaching myself Bambara, but got sidetracked.

  146. David Eddyshaw says:

    There’s a superb reference grammar of its Great Manding sister-dialect Mandinka by the excellent Denis Creissels, and a nice short grammarette here:

    http://www.deniscreissels.fr/public/Creissels-sketch_of_Mandinka.pdf

  147. David Eddyshaw says:

    … from which, on rereading, I discover that the Mandinka word for “window” is palantéeri. Spooky …

  148. There’s a superb reference grammar of its Great Manding sister-dialect Mandinka by the excellent Denis Creissels, and a nice short grammarette here

    Thanks!

  149. Trond Engen says:

    I was a little short on time. By two-step creolization, I don’t necessarily mean twice full-blown creole, as the Tok Pisin example would suggest, but new speakers acquiring an already morphologically simplified (and not necessarily internally consistent) version of the language — something like the second-language Hausa that was used by soldiers in British service in Northern Nigeria.

    Let’s say that a Benine (para-Bouma) ruler starts expanding his realm, and as the empire grows, his kinsmen become ministers and military commanders while large numbers of young men from the conquered areas are enrolled in the army. Maybe it’s even a deliberate policy to break down old tribal loyalties and forge a new imperial identity, because the recruits are put together in bands and garrisons where no single group can dominate and sent on missions far from their own native land. When the imperial system breaks down, as it often does after the momentum of the initial expansion wears out, the local imperial garrisons become independent tribal aristocracies. In that way, morpho-syntactic levelling may happen across a wide range of substrates, and still just be in its initial phase when the military jargon is split into several local varieties. Some of these varieties acquire local speakers and develop into daughter languages, others surely die out. Those that survive keep developing, but without the imperial community of speakers this plays out differently in different varieties, each being an uneven mix of innovative and conservative features.

    I agree that it ‘s a just-so story and probably untestable.

  150. David Eddyshaw says:

    It’s an intrinsically perfectly possible scenario.

    The expansionary Mamprussi state set up by Naa Gbɛwaa certainly did set up local hereditary aristocracies, which exist to this day. But the traditional accounts pretty uniformly say that the original local clan leaders (who had religious rather than administrative roles) remained in place, and the story is much more Norman conquest than Anglo-Saxon invasion. The traditional accounts don’t suggest Völkerwanderungen, just warrior bands waltzing in and making themselves comfortable. Throughout the region, too, your ethnic group membership comes from your father, but your mother tongue comes from your mother: the many Mamprussi in the Bawku area are mostly in fact Kusaal-speaking (my Kusaal New Testament was a gift from a Mamprussi colleague who himself could not speak Mampruli at all.) The Western Oti-Volta languages do share quite a lot of distinctive vocabulary not seen elsewhere in Oti-Volta, but there’s zero evidence that it’s of foreign origin and it includes words as basic as “water.”

    What I wonder is whether these new statelets, having been reorganised on more military lines, would have successfully spread the languages of the non-chiefly commoners far and wide. In some cases, like Dagaare and Mooré, there’s actually supporting evidence for this in traditional histories. This could in principle have led to the sort of simplifications that McWhorter conjures with; but reconstructing Proto-Western-Oti-Volta along orthodox lines suggests that much of the simplification goes back to a period which seems too early. On the other hand, this could be artefactual (I’ve seen an obvious Hausa loanword reconstructed to Proto-Eastern-Oti-Volta by a careful scholar who just didn’t realise that the forms all corresponded so neatly because he was dealing with a recent borrowing.)

    I suppose the simplification-by-largescale-imperfect-acquisition scenario works pretty much regardless of the language of the acquirers; but that is part of my difficulty with it: it’s just unfalsifiable. I’d be happier with it if I could point to plausible substrate languages with features that made particular imperfect acquisitions plausible (like speakers of languages without tones failing to acquire tones.)

  151. Other examples that remain in use

    +tyköä

  152. John Cowan says:

    Some huge fraction of accepted Proto-Berber reconstructions are in fact Arabic loans. Berber is somewhat like Basque: we have never seen it except when it was subaltern to some other language, albeit related in the Berber case and not in the Basque case.

  153. Trond Engen says:

    SFReader: Santa […] is a seven centuries old Mongolian-based [C]reole.

    This is the basic premise of this year’s animated blockbuster from Disney..

  154. Trond Engen says:

    But to bring this back to the original subject, in a new blow to chomskyist linguistics, Dan Everett claims that there’s no Santa clause.

  155. Ow!

  156. David Marjanović says:

    Day saved, I should go to bed.

Speak Your Mind

*