Evidence Rebuts Chomsky.

I had seen links to Paul Ibbotson and Michael Tomasello’s Scientific American piece “Evidence Rebuts Chomsky’s Theory of Language Learning,” but having recently posted yet another anti-Chomsky piece I thought I should probably let this one go by, much as the topic attracts me. However, reader elessorn sent it to me with this cover note:

Nothing in it will be new to you or most people on Languagehat, but I can’t recall seeing many articles like it before — lucid, anti-Chomskian without the ranting (not that I don’t like a good rant!), and with an upbeat outlook. The last one especially. All too often academic fights seem to involve a lot more “death to theory X” than “look what we can do now without theory X in the way.” This was a refreshing exception. Either way, I bet you’ll like the last paragraph.

He was absolutely correct about all of that, and I heartily recommend it. To give you a sample, after a good history of the dispute and the evidence against the Chomsky position, we get:

All of this leads ineluctably to the view that the notion of universal grammar is plain wrong. Of course, scientists never give up on their favorite theory, even in the face of contradictory evidence, until a reasonable alternative appears. Such an alternative, called usage-based linguistics, has now arrived. The theory, which takes a number of forms, proposes that grammatical structure is not in­­nate. Instead grammar is the product of history (the processes that shape how languages are passed from one generation to the next) and human psychology (the set of social and cognitive capacities that allow generations to learn a language in the first place). More important, this theory proposes that language recruits brain systems that may not have evolved specifically for that purpose and so is a different idea to Chomsky’s single-gene mutation for recursion.

In the new usage-based approach (which includes ideas from functional linguistics, cognitive linguistics and construction grammar), children are not born with a universal, dedicated tool for learning grammar. Instead they inherit the mental equivalent of a Swiss Army knife: a set of general-purpose tools—such as categorization, the reading of communicative intentions, and analogy making, with which children build grammatical categories and rules from the language they hear around them.

The new approach (which of course is discussed in greater detail) sounds sensible to me. And here’s that last paragraph:

Universal grammar appears to have reached a final impasse. In its place, research on usage-based linguistics can provide a path forward for empirical studies of learning, use and historical development of the world’s 6,000 languages.

Comments

  1. Chomsky’s retort:

    >Grammar? Did I say universal grammar? I meant universal diction:

    http://languagelog.ldc.upenn.edu/nll/?p=28143

  2. Just for people who’d like to read a response, here’s a nice trilogy of them from Jeff Lidz:

    http://facultyoflanguage.blogspot.com/2016/09/the-generative-death-march-part-1.html
    http://facultyoflanguage.blogspot.com/2016/09/the-generative-death-march-part-2.html
    http://facultyoflanguage.blogspot.nl/2016/09/the-generative-death-march-part-3-whose.html

    Briefly: Tomasello has been announcing the impending death of Chomskyan generative syntax for decades now, and he’s no more persuasive now than he’s ever been. But your mileage may vary, I guess.

  3. Thanks! My mileage does vary, but I’m always glad to have the opposing point of view represented.

  4. While I am no fan of many of Chomsky’s more specific ideas, saying, “The theory, which takes a number of forms, proposes that grammatical structure is not in­­nate. Instead grammar is the product of… human psychology (the set of social and cognitive capacities that allow generations to learn a language in the first place),” sounds a lot like a distinction without a difference.

  5. J.W. Brewer says:

    It seems to me that it’s a distinction without a difference only if you think the Chomskyan notion of a “language module” in the mind is simply a loose and poetic metaphor with no substantive content. There is no dispute that somehow or other human beings are born with (or at least naturally develop at a very early age absent unusual trauma or developmental problems) the “innate” cognitive capacity to learn and master language. But as far as I can tell the Chomskyans have spent the last five or six decades making a more specific and thus controversial claim about the details of that innate cognitive capacity, not least the claim that it’s language-specific as opposed to it being a stroke of luck that the complex interaction of a bunch of other bits of our swiss-army-knife cognitive skill set which individually have other non-language-related functions can be McGyvered together into the ability to use language.

    It’s not clear to me FWIW that the plausibility or usefulness of a broadly generativist approach to grammar necessarily stands or falls with Chomskyan theories about language acquisition, or universal grammar.

  6. David Eddyshaw says:

    For some reason the title struck me as a “Man bites dog” story …

  7. Heh.

  8. a stroke of luck — everything is the way it is because it became that way. Including language — if our swiss-army-knife cognitive skill set whose bits individually have other non-language-related functions had been constituted differently, a different sort of language ability would have been McGyvered together when the need for knowledge transmission arose, and we would have been communicating in something quite unlike the human languages we know. Very possibly at an earlier or later stage in our evolution, too.

    In other words, the odds against apes evolving to be able to use human language as we know it may be astronomical, but the a-priori odds that those clever apes would find an efficient way to communicate were probably quite good.

  9. If the Chomskyans truly believe that our capacity for language did not evolve by co-opting brain structures that originally evolved in response to other evolutionary pressures, then they don’t understand evolution at all.

  10. Trond Engen says:

    That’s been an argument against “the language organ” for as long as the idea’s been around. I gather that these days nobody ever expected or even mentioned a language organ, just the idea of the existence of a trait that at some point emerged and made co-opting of other structures possible. A peace agreement after 50 years of language wars are within reach with the carefully worded treaty text “Well, duh.”

  11. Trond Engen says:

    Not that I necessarily think one can pinpoint such a trait either, or that the hunt is fruitful for the understanding of language. It’s just about which adaptation happened to take place last. It might not even be anything involved in production and processing of language, but some catalyst feature since lost.

  12. J.W. Brewer says:

    FWIW I tweaked my earlier comment before posting to talk about a language “module” in the “mind” rather than a “language organ” in the “brain” in order to give the Chomskyan position (again, assuming there’s an actual controversial position rather than just an idiosyncratic way of talking about stuff that no one actually disagrees about substantively) as much wiggle room as it might want with respect to how the “mind” and its internal structure relates to something so vulgarly material as the brain of some semi-randomly-evolved primate.

  13. Trond Engen says:

    No, I’m unfair. Such a trait is not part of the current position, and haven’t for a long time, if understood as a biological trait, only the emergence of some unique little mental capacity that made language possible.

  14. Yes I’m with the “Well, duh.”/”distinction without a difference” conclusion.

    What strikes me in the Lidz rebuttals is the number of appeals to syntactic models formed in the mind of the language learner (“building representations”). But we’re never going to find the (wonky) ganglions for agreement-attraction or island-violations. We can experimentally assess language processing ability, but that doesn’t show how the learner acquires or retains the ability, nor whether it’s language-specific.

    Can’t performance errors be equally explained by over-generalisation of patterns? Which is certainly a general cognitive failing.

    All of Lidz’s examples in Part 1 I found of borderline comprehensibility. (They’re certainly not everyday sentences.) So how does this constitute evidence that language learning or processing is different in nature to other highly abstract mental modelling skills like calculus or subatomic physics?

  15. But it might be a bit more complicated, no? It is possible that communicative mode that we can recognize as a (proto-)language emerged when some mutation changed Homo‘s cognitive capacity such that some particular way of communication became adaptive. After that, many of the relevant cognitive abilities evolved toward supporting this mode of communication, that is evolved to create language as we know it. This co-evolution of otherwise not particularly related cognitive abilities can be construed as creation of the “language organ”. For theoretical physicists among us, it is rather like long-range theory that is determined by renormalization and dimensional transmutation more than by the underlying small-scale structure.

  16. All of Lidz’s examples in Part 1 I found of borderline comprehensibility.

    “Every sailor loves any sailor.”

  17. >All of Lidz’s examples in Part 1 I found of borderline comprehensibility.

    Lidz:
    > (2) a. Valentine is a good value-ball player and Alexander is too
    > b. Valentine is a better value-ball player than Alexander is

    I was sure Lidz was going to draw conclusions from the distinction he had set up between a ball player of better value and a good player of a game called value-ball. Alas, that wasn’t his point at all.

    I was also primed and waiting for Valentine to be a holiday in the second sentence. He manages to construct sentences with more oddness and ambiguity than he has recognized, obscuring the point he’s trying to make.

  18. Thanks @ryan, I’m glad it’s not just me. What do you mean oddness? 😉

    value-ball is apparently a thing at the University of Maryland Linguistics Department (where Lidz is) http://ling.umd.edu/tags/semantic-valueball/ Valentine and Alexander are players (apparently).

    Reminds me of whoever-it-was in love with their toothbrush and kissed her.

    If you strip language of familiar semantic context such that our poor monkey brains short-circuit then all claims are off about what “we” understand or process or “fill in”.

    To me, this makes any attribution of purely linguistic processes or mental constructs (as opposed to general cognition) just vacuous. These armchair linguists should do something useful like getting out into the field and recording endangered languages. [Hat said no ranting, sorry.]

  19. I’m quite prepared to believe that the capacity for language learning was present long before language emerged.

    You could probably teach a Neanderthal to drive a bus too, if one was somehow brought forward in time. For a less silly example, consider the glacial pace of innovation in prehistory — ‘modern man’ has existed for 200 000 years or so, with the same brain capacity as we have, but it’s only in the last 10 000 that things like horse riding and wheels appeared.

    Also consider that it takes a modern human child 5-10 years, maybe 15, to learn a language to a level where it can usefully transmit cultural knowledge. If pre-modern humans and humanoids were slower at it, most of them would be dead before they could learn to speak, and the sort of language that could be transmitted between generations might not give a group enough of an edge to be worth the bother.

    Maybe you could even teach real ASL to a chimpanzee if you had 100 years to do it.

  20. Trond Engen says:

    D.O.: But it might be a bit more complicated, no?

    Yes, it very much might, which is why I say one single trait may never be found. I may be wrong, but I don’t understand a set of mental skills co-evolved into a new capacity as a “language organ” in the early chomskyan sense, and the current minimalist approach definitely has its focus elsewhere.

  21. @D.O.: Are you a theoretical physicist too?

  22. Hat said no ranting, sorry.

    No I didn’t, I love ranting! I just said this particular piece doesn’t rant, so those who don’t care for ranting might enjoy it.

  23. Brett, no. But I studied it.

  24. Lars:

    there is a half-serious theory which stipulates that humans became sapient considerably more recently than it is thought. In the most extreme version, the end of the last Ice Age (circa 9000 BC) is claimed as the moment when human intelligence emerged.

    All of human culture in the Paleolithic period is explained away as complex animal behaviour. (Making stone axes of the same model for seven hundred thousand years surely more resembles instinctive animal behaviour than conscious human activity).

  25. David Marjanović says:

    That idea has a hole you can drive a car through, like the concrete “coffin” around Chernobyl:

    How on Earth is “human intelligence” supposed to have spread around the world and reached every single human population so recently?!?

  26. @SFReader, I remember that someone published a book based on that idea some decades ago, but I never looked into the specifics.

    About the same time that ‘cusps’ and branching phenomena could explain everything in nature…

    Personally I think that there probably is a feedback effect between the level of material culture and longer life + leisure + sustainable population that means that the pace of invention behaves like a fourth or fifth order differential equation — there isn’t one specific magic point where the nature of cultural development changes, but if your sampling is too coarse (which it is in this case) it looks like an abrupt domain change.

    (There is also the slightly darker version that says that the guys with the stone axes actually had it made and going beyond the primitive hunter/gatherer stage was a mistake, because it drove population increase which drove pressure to invent and to organize, which allowed warfare which fed back to the others and then we had nuclear bombs when we could just have been sitting around the fire every night telling tall tales).

  27. But didn’t deglaciation play a big – and abrupt – role? Given how quickly agriculture and settled society took off after the Pleistocene, it seems like we were just waiting for a temperature rise.

  28. David Marjanović says:

    Deglaciation had very little impact outside of the northern continents. Agriculture & stuff took several millennia to spread even just to the places where it could spread given the technology of the time.

    In other words, I’m rhetorically asking: why are Australian aborigines, Andamanese etc. etc. psychologically unremarkable modern human beings?

  29. Deglaciation had very little impact outside of the northern continents.

    I can easily find several papers that disagree, such as this one. I don’t think it’s by coincidence that we’re typing these words during an interglacial.

    On the other point, I agree – it seems like a crude refashioning of the old multiregional hypothesis.

  30. -How on Earth is “human intelligence” supposed to have spread around the world and reached every single human population so recently?!?

    Human intelligence (AKA sentience) is spread by contact.

    Specifically, by process of language learning.

    You see, we are talking about non-sentient creatures who had basically same hardware for thinking as us, they just lacked the software (packaged in form of language).

    Hence, when a pack of non-sentient anatomically modern humans entered contact with a tribe of sentient humans, within few years or decades, they would have acquired the full culture package (what in archaeology is called Upper Paleolithic culture revolution), including language and human level intelligence.

    Incredibly, same process actually happened with hominids which weren’t anatomically modern humans. I am talking about traces of Upper Paleolithic culture associated with late Neanderthals. Even they, despite much wider gap, adapted human language and became sentient beings. For a few millenia anyway, but then they were exterminated by Upper Paleolithic anatomically modern human racists.

    The process is usually dated to circa 40 000 BC, but it is well known that in most of Asia or Australia, lower Paleolithic survived to the end of last Ice Age when the last non-sentient humans finally acquired language and intelligence (while non-sentient hominds were eaten)

  31. In fact, we all can see same historical process unfold in our own families.

    Every human baby age two or less is, by definition, a non-sentient anatomically modern human (I am oversimplifying, of course, technically speaking newborn babies are not anatomically modern humans at all, it actually takes years to advance beyond the hominid stage). It acquires human intelligence in the process of learning human language through sustained contact with sentient humans nearby (ie, parents and older siblings).

    We know of cases when abandoned babies survived without contact with other humans and were raised by animals (usually wolves). And as a result, they failed to acquire human intelligence, retaining however, basic capability to do so in the future (because it’s in our hardware, ie, anatomy and neurophysiology)

  32. It is possible that communicative mode that we can recognize as a (proto-)language emerged when some mutation changed Homo‘s cognitive capacity such that some particular way of communication became adaptive.

    Certain cognitive capacities may be necessary for language to arise. After that, it may be a bit like a virus (or perhaps like life itself): once it exists, it’s hard to eradicate.

    Valentine is a good value-ball player and Alexander is too
    Valentine is a better value-ball player than Alexander is

    Yeah, why couldn’t he have chosen something that a child is more likely to say, like:

    You’re a fast runner and he is too
    You’re a faster runner than he is

    Using simpler sentences, the difference between the two leaps off the page. Reduced to basics:

    You’re a fast runner. He is also a fast runner
    You’re a fast runner. He is a fast runner. You run x fast, he runs y fast, x is greater than y

    The complexity of the comparative becomes apparent with the later sentence:

    You’re a faster runner than I think he is
    You’re a fast runner. He’s a fast runner. You run x fast, I think he runs y fast, x is greater than y

    Taking the following “parallel examples” as the next step is almost disingenuous:

    c. You’re a fast runner and I heard a rumour that he is too
    d. * You’re a faster runner than I heard a rumour that he is

    Haspelmath’s questioning of the use of the English comparative makes sense. Patterns like this involve a complexity that is missing from more straightforward language like “You’re a fast runner”, or better, “You run fast”.

    Having restraints on increasing complexity in sentences like these does not seem counterintuitive.

  33. Also, using a comparative particle like than is an SAE feature very rare in the world’s languages: see WALS.

  34. @SFReader: You’re using sentience in the old sci-fi sense? I think sapience is the apter term: as I understand it, my cat, having the ability to feel and perceive, is sentient.

  35. sentient cats are new to me. I somehow missed that moment

  36. This exchange in the very first comments to Lidz’ initial post::

    Martin Haspelmath September 11, 2016 at 1:40 AM
    It’s true that IT do not address phenomena such as these, which are indeed intriguing and worthy of in-depth study. However, it is completely unclear how innate knowledge could explain these facts, because (als far as we know) they are specific to English. I would be impressed if someone showed that there is an abstract pattern that is valid for all languages that helps explain These English facts.

    Omer Preminger September 11, 2016 at 5:14 AM
    [Pirahã expert Daniel] Everett’s Fallacy: x is an innate feature of the human linguistic capacity, therefore x must be observable in every human language.

    Unless you have an explanation for how “analogy”, “intuiting”, and other ill-defined notions lead to this knowledge ending up in the heads of English speakers, the only remaining explanation is that they were there in the first place, and perhaps in some languages, specific properties conspire to make them unobservable.

    In other words, to Haspelmath’s “But this might only be a feature of English. We can’t just assume it applies everywhere”, Preminger replies, “Unless you can explain how it could only work in English, the only remaining explanation is that it must be present everywhere, even if we can’t observe it.” This is willful epistemic insanity, and gets the burden of proof entirely backwards. It’s also why I’m suspicious of “distinction without a difference” arguments that try and bridge the gap between, well, more flexible versions of UG and something like what Ibbotson and Tomasello were arguing. However the metaphor is finessed–language “instinct,” “organ,” whatever–it seems to me the basic problem is an unresolvable disagreement about epistemologically sound procedures of weighing evidence.

    Your Chomskian knows or intuits that his habits of arguing require that language function like physical matter, universally subject to constant, ubiquitous laws. He might be willing to admit that the nature of the Laws of Language are beyond our current grasp, be willing to switch out metaphors in referring to them, etc., but in the end he needs them to exist, and to function like gravity–as observable on Mars or Alpha Centauri as it is here on Earth.

    Of course anyone who has any experience of any language’s change over historical time would or should find this crazy. Learning children may experience their mother tongue in one synchronic rush, but they observe, acquire, and learn to correctly employ (or refrain from employing) a ton of fossilized and fossilizing bits of language as well, essentially diachronic survivals that have no necessarily systematic connection with the majority of the active grammar. (If indeed there is any set of systematically, mutually necessitating principles at the core of any grammar at all.) Even within individual languages it’s hard to set out a satisfying set of consistent laws, much less between languages, to say nothing of all of them.

    Absent that sensibility, we get discussions like this from Lidz in the body of the post (bolding mine):

    (3) a. Valentine is a good value-ball player and I think Alexander is too
    b. Valentine is a better value-ball player than I think Alexander is
    c. Valentine is a good value-ball player and I heard a rumor that Alexander is too
    d. * Valentine is a better value-ball player than I heard a rumor that Alexander is

    The fact to be explained here is why the child learner when building representations for (2) doesn’t treat the silent predicate in the same way in the two cases. Both can be interpreted as identical to the main clause predicate in (3a/b); however, this dependency can only hold across the expression “hear a rumor that…” in the coordinate (3c) but not the comparative (3d). It is an analogy that could be drawn but apparently isn’t. Moreover, it seems that (2b) has a structure analogous to the structure of interrogatives.

    4) a. What do Valentine and Alexander like to play together?
    b. What do you think that Valentine and Alexander like to play together?
    c. * What did you hear a rumor that Valentine and Alexander like to play together?
    In (4) there is a dependency between the wh-phrase “what” and the verb play. In (4b), we see that this dependency can be established across multiple clauses (just like the coordinate and comparative ellipses), and in (4c) we see that it cannot be established across “hear a rumor that” (like the comparative ellipsis and unlike the coordinate ellipsis).

    Evidently the analogy that the child learner draws when acquiring English is that comparatives have the same kind of structure as wh-questions. Why do they draw this analogy and not the analogy between the comparative and the coordinate ellipsis, which shares more obvious surface features?

    Why? Because children have a world of occasions to learn that “and” functions quite differently than “than,” a difference in large part motivated by the two words’ different histories. Lidz’ reasoning seems be: (1) both are conjunctions that, in certain cases, can take identical full sentences as arguments, ergo (2) it’s weird that children don’t analogize to allow the than(x) to mimic the and(x) function in its behavior, despite their shared “surface features.” But of course, the schematic surface similarity isn’t nearly as salient for child learners as the well-known non-analogous behavior patterns of the words “and” and “than.” Of course they’re not going to make that analogy. And it’s not that comparatives work like wh-questions, it’s that both work very differently than and-coordination. And don’t even get me started on the implicit assumption that “I think that” and “I heard a rumor that” are somehow equivalent constructions, just because they lead up to “that + [sentence].” And… but why go on?

    I’ll admit that as a general practice it’s reasonable to doubt massive challenges to reigning theories based on a single scholar’s unique knowledge of a single case/organism/language. That’s a healthy skepticism. But it’s not nearly as worrying as this voodoo epistemology that assumes the whole variety of human languages is intuitable from observation of one (group) of them. I think this is why, ultimately, the “one extra step” of the “language organ” has got to go. Without the a priori assumption of some kind of sovereign universal “out there”, English-centered “reasoning” like Lidz’ here doesn’t have a leg to stand on.

  37. sentient cats are new to me. I somehow missed that moment

    Sentient: “Able to perceive or feel things.” I think sentience and sapience are important enough, and different enough, that it’s useful to have separate terms for them.

  38. Agree with Elessorn.

    Geoffrey Pullum wrote a short paper on the fact that grammar is not as regular as assumed and doesn’t follow absolute rules, which I can’t seem to find. This observation shafts the generativists since ad hoc rules need to be made up to account for the (arbitrary) holes. Language is more of a pastiche than a seamless garment of whole cloth, which fits the idea of language as usage.

    As for the “You’re a faster runner than I heard a rumour that he is” type of sentence, it’s possible that all you need is a prohibition on inserting an object in this structure, since the object screws up the way the sentence hangs together. With an object, various structures are possible, but not using “that”:

    You’re a faster runner than he’s rumoured to be
    You’re a faster runner than rumour has him to be

    Incidentally, I’ve heard a lot of sentences in spoken English that attempt to stretch the boundaries of grammar, and I wouldn’t be surprised if someone actually said “You’re a faster runner than I heard a rumour that he is”. What is clear, however, is that such a sentence isn’t acceptable in proper written English. One day it’s quite possible that this restriction will break down and people will start saying it and even writing it.

    (For instance, I’ve heard radio broadcasts that use sentences along the lines of “It’s important not to spend more money than can be afforded”. According to my understanding, “afford” originally didn’t take the passive in this kind of construction.)

  39. Also, using a comparative particle like than is an SAE feature very rare in the world’s languages: see WALS.

    I’d call 22% “uncommon” or “unusual,” but hardly “very rare.”

  40. David Marjanović says:

    I don’t think it’s by coincidence that we’re typing these words during an interglacial.

    Oh, indeed not. I agree that agriculture requires the kind of climate stability only found in interglacials. However, agriculture left much of the planet untouched (directly or even indirectly) for several more millennia.

    Hence, when a pack of non-sentient anatomically modern humans entered contact with a tribe of sentient humans, within few years or decades, they would have acquired the full culture package (what in archaeology is called Upper Paleolithic culture revolution), including language and human level intelligence.

    What about absence of contact? What about, say, the Tasmanians?

    For a few millenia anyway, but then they were exterminated by Upper Paleolithic anatomically modern human racists.

    It looks more likely that they, and the anatomically modern humans of western Europe who had mixed with them, were done in by the Last Glacial Maximum. When it was over, Europe was resettled from the east by Ancient North Eurasians.

    Incidentally, I’ve heard a lot of sentences in spoken English that attempt to stretch the boundaries of grammar, and I wouldn’t be surprised if someone actually said “You’re a faster runner than I heard a rumour that he is”.

    I’m quite prone to doing this, actually. I don’t like the results, but I keep coming up with this kind of sentence anyway.

  41. David Marjanović says:

    I’d call 22% “uncommon” or “unusual,” but hardly “very rare.”

    “Very rare” fits once you look at the map:

    4. Geographical distribution

    As the map demonstrates, the areal distribution of the various types of comparative constructions is striking (see also Heine 1994). For one thing, the Exceed Comparative appears to be almost exclusively restricted to two areas, viz. sub-Saharan Africa, and China and Southeast Asia. No less limited is the distribution of the Particle Comparative, which turns out to have its base in the modern languages of Europe; instances of this type outside Europe (such as the Uto-Aztecan languages in North and Central America) may well be cases of influence from English and/or Spanish. The Conjoined Comparative has a stronghold in Australia and New Guinea, and is also prominent in the Amazon basin. Finally, the Locational Comparative is the rule in northern Africa and in the vast landmass of Eurasia (including the Middle East and India, but excluding Europe), and can also be found in Eskimo languages and scattered instances of languages over the Americas, in Polynesia, and in Australia and Papua New Guinea.

  42. I guess it depends on how you define “very rare”; I can only repeat that 22%, in my view, does not fit.

  43. I’m slightly confused by the classification of Mandarin as an “exceed comparative” language. In standard Mandarin the format uses 比 ‘to compare’: 他比你高 tā bǐ nǐ gāo ‘he compare you tall’.

    In Southern Chinese (such as Cantonese), I believe the “exceed comparative” is common. I’m not sure if this is an accurate or correct sentence, but 他高过你 tā gāo guò nǐ ‘he tall exceed you’ would belong to this type.

    Then there is the written form 他高于你 tā gāo yú nǐ (‘he tall than you’), which gives me the impression of being a Classical-style sentence that has become common under the influence of English. I’m not sure what 于 could be characterised as in the scheme given, but presumably it’s a “locative”.

  44. I meant to say: very rare among the other (i.e. non-SAE) languages.

    grammar is not as regular as assumed

    Probably “The Truth about English Grammar: Rarely Pure and Never Simple”. I’ve quoted a bit of it here at the Hat.

  45. I meant to say: very rare among the other (i.e. non-SAE) languages.

    Ah, OK, that makes sense.

  46. “instances of this type outside Europe (such as the Uto-Aztecan languages in North and Central America) may well be cases of influence from English and/or Spanish”—this is not convincing. Malagasy was not influenced by English or Spanish. All three Uto-Aztecan languages in the sample (Comanche, Tümpisa, Tohono O’odham) have this construction. Nevertheless you can still argue that the construction is rare in the sense of having arisen independently only a few times. Also note that it’s 22 languages, amounting to 13% of the sample.

  47. Also note that it’s 22 languages, amounting to 13% of the sample.

    Whoops, very bad reading on my part! Thanks for the correction. I’ll move my preferred characterization down a notch to “rare.”

  48. I too agree with Elessorn’s excellent comment and also with Bathrobe’s point that the asterisks in sentences like 3c quite possibly shouldn’t be there. In fact I’m willing to bet that native English speakers do regularly, if rarely, produce such sentences in natural speech. Their rarity might simply be due to the corresponding rarity of discourse situations in which they could naturally be produced (can anyone describe an plausible set of actual circumstances in which one might want to say “Ted is a faster runner than I heard a rumor that Tim is?”). Their unacceptability in the standard written language could then simply be due to their rarity in speech, but even if not, obviously a universal theory of grammar that can only account for written language is, well, slightly misnamed?

  49. That “rumor” sentence is in an interesting category. For me, it is straight-up ungrammatical, but I can also see that another native speaker might not agree. It would be interesting to know more about how our personal language processes modules come to grammaticality decisions about such rarely-produced sentences.

  50. -What about absence of contact?

    In the absence of contact, I presume non-sentient humans remained non-sentient, unable to speak a word.

    -What about, say, the Tasmanians?

    Beginning of full isolation of Tasmania from Australian mainland is dated c. 12000-6000 BP. By that time, they certainly would have acquired language.

  51. More on the ‘rumor’ example:

    1) Ted’s a faster runner than rumor has it Tim is. (fine)
    2) Ted’s a faster runner than I’ve heard from everyone Tim is. (a bit awkward, but passable)
    3) Ted’s a faster runner than I’ve heard rumors Tim is. (more awkward, but still better than “a rumor”)

    (All of these, too, would I think work better if we said “Ted’s an even faster runner than..,” since it would imply that Tim was indeed so fast people widely made talk about it, making the utterances feel more natural.)

    If I’m right (and it’s a dicey thing since we can always talk ourselves into grammaticality judgments), there’s a pragmatic problem with “heard a rumor.” Maybe it’s because the turn of phrase (as opposed to “heard rumors that”) is usually used not so much to “judgment-source” the information following, but rather just to tentatively introduce it. In other words, the two register as (if not inherently, at least usage-habitually) belonging to different categories of utterance purpose, causing a mismatch that feels “wrong” in “than I heard a rumor that.”

    Perhaps a similar pair might be:
    1) Ted’s a faster runner than there’s a possibility Tim is.
    2) Ted’s a faster runner than chances are Tim is.

    I think (1) isn’t impossible, but feels off somehow. And it could be improved by saying “there’s a good possibility,” at least to me.

    Either way, if pragmatics is the answer, it lends weight to Tomasello’s usage primacy theory of language– which is certainly my experience as a learner. It is *possible* to speak (repurpose) Chinese or Japanese in a more SAE manner, after all, but it’s not natural, and there’s no way I know of to express the unnaturalness by sentence diagram.

  52. there’s a pragmatic problem with “heard a rumor.”

    I think that’s the nub of the problem.

    “I’ve heard a rumour that Tim is x fast” is a declaration of having heard a particular rumour.
    “I’ve heard rumours that Tim is x fast” is a declaration of hearing rumours in general.

    “Ted’s a faster runner than I’ve heard a rumour that Tim is” embeds knowledge of a particular rumour in a sentence that makes a general statement about Ted’s speed. To be acceptable, the declaration of a particular rumour needs to be expressed separately from any general statement of Tim’s or Ted’s speed. What about:

    “Ted’s a faster runner than the rumour that I’ve heard Tim is.”

    This yields a marginal improvement of acceptability.

    Methinks the generativists are barking up the wrong tree.

    @ John Cowan

    That is indeed the paper I was thinking of!

  53. That should have been:

    You’re a faster runner than I heard the rumour that Tim is

    Marginally better, but still not the best.

  54. David Marjanović says:

    “The Truth about English Grammar: Rarely Pure and Never Simple”

    Downloaded, bookmarked, and read. Thank you!

    Beginning of full isolation of Tasmania from Australian mainland is dated c. 12000-6000 BP. By that time, they certainly would have acquired language.

    …OK, that shifts the problem: the continent of Australia as a whole (with New Guinea) was settled some 40,000 or 50,000 years ago. The next evidence for contact with Asia is the introduction of the dog/dingo just 3,000 years ago, IIRC. You need to postulate additional contact events that were intense enough to teach language to children sometime within this period. At some point it all becomes rather unparsimonious.

  55. unparsimonious

    Which is how you say fuckwitted in Scientistish.

  56. Trond Engen says:

    Let the non-native come to rescue:

    “Ted’s a faster runner than rumour says of Tim.”

    Doesn’t this work better? I think the reason is that stacked prepositional phrases can be kept apart by slight pauses:

    “Ted’s a faster runner – than rumour says – of Tim.”

    I still don’t like the start “Ted’s a faster runner”. It works better by itself than as the beginning of a dissertation. I think it’s about how far ahead you realistically plan a sentence. One way out is to provide some of the information first and then start again:

    “Ted’s fast. He’s a faster runner – than rumour says – of Tim.”

    “Ted’s a fast runner. He’s faster than rumour says – of Tim.”

    But is this grammar?

  57. @Trong Engen: I agree that that phrasing is grammatical, although I would not use the second pause you have indicated.

  58. J.W. Brewer says:

    Just for kicks, I put “Valentine is a better value-ball player than I heard a rumor that Alexander is” into google translate. It came out as “Valentine yn chwaraewr gwerth-bêl yn well nag yr oeddwn wedi clywed si bod Alexander yn” in Welsh, “Valentine I Aleksandr bir söz eşitdim daha yaxşı dəyər top oyunçu” in Azerbaijani, and “Valentine dia zava-dehibe-baolina tsara kokoa mpilalao noho izaho nahare filazana fa Alexander dia” in Malagasy. Then I got bored, but I invite comment from anyone who knows one or more of those languages well enough to have insight into whether the translations should also be marked with a * for alleged ungrammaticality.

    (FWIW, google translate thought my English input was odd enough that it asked me “Did you mean: Valentine is a better value-ball player then I heard a rumor that Alexander is” [with “then” italicized and boldfaced for emphasis] — which doesn’t really seem like the most plausible way of fixing the questionable phrasing of the original!)

  59. J.W. Brewer says:

    The problem with the “rumor” examples is the object, as noted upthread. “Valentine is a better value-ball player than I think Alexander is” is the example given as uncontroversially grammatical, but “Valentine is a better value-ball player than I’ve heard Alexander is” is equally uncontroversially grammatical, as would be, e.g. “blah blah blah than they say Alexander is” or even “than it’s rumored that Alexander is” or “than rumor has it Alexander is” or many other possibilities that don’t have the specific structural problem (awkwardly-placed presence of direct object) of the asterisked example. But this seems trivially obvious. Why the fellow thinks it’s an “aha” point for any sort of thesis at all that two given strings of words you can connect with “and” in English can’t always be felicitously connected with “than” is what I’m failing to follow. What strawman theory out there implies a contrary expectation that’s now being shown false?

  60. David Marjanović says:

    Which is how you say fuckwitted in Scientistish.

    Heh. That wasn’t my intention this time. I really did mean to say “OK, this is all possible, but we don’t need this hypothesis”.

  61. What strawman theory out there implies a contrary expectation that’s now being shown false?

    The argument in the linked posts is essentially that only Chomsky-ryu grammarians have offered a coherent account of why children learn to generalize in one way and not another. I don’t believe that this is true, but I know better than to argue on this topic without all terms rigorously defined, engraved on gold plates, and sworn to in the presence of forty-four impartial witnesses beforehand.

  62. @J. W. Brewer, I tried with Danish since I can’t really give native speaker reactions to anything else: Valentine er en bedre værdi-ball spiller end jeg hørte et rygte, Alexander er.

    As a point of usage it should be et rygte om at Alexander er, the preposition and conjunction aren’t optional as in English — and that makes me think that GT doesn’t have correspondences that actually cover the grammatical construction, but is just splicing things together.

    Other than that it’s fine for relaxed style, but Danish is notorious for allowing deeply nested arguments to be lifted. Even better is end rygterne siger at Alexander er, which is equally nested but uses a more common construction — would than rumour has it that Alexander is be objectionable in English?

  63. would than rumour has it that Alexander is be objectionable in English?

    Sounds fine to me.

  64. @John Cowan: Thanks for repeatedly linking to Rarely Pure and Never Simple; I had missed the other times. And wow, only a measly 4 references on scholar.google, and apparently (I can’t be certain due to paywalls) none of them looking like real engagement! Why isn’t anyone talking about this‽

  65. Probably because it’s an emperor-has-no-clothes paper: it says that grammar isn’t really rule-bound, and that’s not playing any of the current games. The best and safest thing for scientists to do is to ignore it at the current stage of scientific acceptance. J.B.S. Haldane listed them as follows:

    1. This is worthless nonsense.
    2. This is an interesting, but perverse, point of view.
    3. This is true, but quite unimportant.
    4. I always said so.

    This paper is hovering between 1 and 2.

    As for paywalls, appending “.sci-hub.cc” to the domain name in the URL is usually your friend. The Davis et al. hit cites Pullum only for the claim that “lexical categories [are] purely semantic phenomena […] is well known to be false”; the Zyman hit (a senior thesis) says Pullum’s claims are irrelevant to his work; the Payne hit doesn’t cite Pullum at all; the Huang hit is all squiggles and squoggles to me.

  66. Stephen C. Carlson says:

    The argument in the linked posts is essentially that only Chomsky-ryu grammarians have offered a coherent account of why children learn to generalize in one way and not another.

    I wonder how they deal with Classical Greek then. Agbayani et al. have an article Phonological Movement in Classical Greek where they demonstrate that many of the old generative constraints and conditions–including the Left Branch Condition, the Coordinate Structure Constraint, the Adjunct Consition, binding consitions on anaphor, and freezing islands–simply don’t apply to Classical Greek with its hyperbaton. (They propose instead [or in addition to] a whole set of movements in the phonological module, using optimality theory as their formalism. I express no opinion on that.)

    What I had been taking from this is that generative-style syntactic movement isn’t very helpful for explaining classical Greek syntax, but it now occurs to me that it could also challenge the poverty of the stimulus argument because whatever constraints are inherent in the syntax, presumably due to universal grammar, are violated anyway in a post-syntactic module. If possessing a universal grammar is necessary for an ancient Greek child to acquire the language and make the right generalizations about grammaticality and ungrammaticality, I wonder how this child is to do this when the language data contain utterances that the syntactic constraints supposedly won’t permit.

    I realize that for a long time the study of grammar has been inordinately dominated by classical Greek and Latin models (e.g. the ban on splitting infinitives). But here it seems that the pendulum has swung far in the other direction where grammatical theories are now a poor fit to the classical languages. It is as if linguists aren’t trained in the classical languages anymore. Indeed, it is necessary now to transliterate the Greek in order to have a reasonable expectation of reaching that audience.

  67. David Marjanović says:

    It is as if linguists aren’t trained in the classical languages anymore. Indeed, it is necessary now to transliterate the Greek in order to have a reasonable expectation of reaching that audience.

    [ˈsɪkˑˈtʀanˑsɪtˈg̊loˑɐ̯ʀɪaˈmʊnˑd̥ɪ]
    (not Greek)

  68. It was Peter Matthews back in 1974 who reintroduced transformationalists to the world of Latin morphology in his book Morphology. It appears to have brought the Word and Paradigm model back into favour.

  69. @John Cowan: Well, he does say that construction grammar (among others) are compatible with the main point (language as a historically-determined library of forms, not necessarily reducible to abstract minimalist principles). I’ve been meaning to get into CG for some time now; this essay gave me extra motivation.

Speak Your Mind

*