Why can't we talk to each other?: Why can't we listen?

In a very thoughtful review of Fritz Newmeyer's Language Form and Language Function, Martin Haspelmath (2000) poses the (partially rhetorical) question Why can't we talk to each other? which is also the title of the piece. The "we" in question are formalists* and functionalists, who—despite the best efforts of linguists like Newmeyer and Haspelmath—seem locked in a Middle East-style conflict without compromise or meaningful concession. One of the reasons for non-talking, Haspelmath proposes, is that our fundamental goals are incompatible. This may be so, but I think there is a simpler explanation: a conversation requires listeners as well as talkers, and we just don't listen.

Let me climb off the fence for a change, to make clear which kind of non-listening I'm referring to.
It may be true that, by and large, Chomsky doesn't listen to anyone who proposes an alternative perspective on language and language acquisition, and that on the rare occasions he does, his response is infuriatingly dismissive. As an example, one need only listen to the first 20 seconds or so of this lecture (Chomsky not addressing the question of statistical learning). Depending on one's perspective, this failure to engage with—or even cursorily acknowledge—the theoretical and empirical contributions of others to our understanding of language diversity, acquisition, or processing may be interpreted either as arrogant provocation, or simply as reflecting his frank appraisal of the value of such work. The truth often hurts. Whichever is correct, there is no question that Chomskyan rhetoric sticks in the craw of many people, and is provocative to some. But the relevant point here is that Chomsky doesn't bash other researchers, he simply ignores them. An unproductive stance, to be sure, but not a counter-productive one. And not a wasteful strategy either: the refusal to countenance alternative approaches has freed up a lot of time for internal theory development that his detractors regularly steal from their own pockets.

By contrast, these same detractors who lay continual siege to the Chomskyan keep—a vast, uneasy heterogeneity of opponents coming from a bewilderingly diverse range of theoretical schools and academic disciplines, more reminiscent of the attacking hordes in Narnia or Lord of the Rings than of a coherent opposition—are passionately engaged in a different kind of non-listening, which—not to put too fine a point on it—wastes their own time, and everyone else's too.

Chomsky, by general consensus, is an intellectual genius of the first order; there are few human beings of his calibre alive in the world at the moment. This doesn't make him right about everything—notwithstanding the assertions of his more fawning students and disciples, who treat his writings as ex cathedra bulls—but his status does make his work worth reading carefully, and his lectures worth listening to with the same attention. So why does the opposition not listen (carefully)?

Their difficulty cannot be intellectual: many of the most prominent nay-sayers are considerably more successful academically—and smarter—than I have ever hoped or wished to be. Nor are they lazy. Quite the opposite: I would wager that the work of the average functionalist linguist, or typologist, or developmental psychologist shows more scholarship, academic rigor, and empirical responsibility than any paper by Chomsky one might choose at random. This fact only demonstrates what we know already from other fields, namely, that creative genius and craftsmanship are not closely correlated.

So if it’s not intelligence or diligence, it can only be willful ignorance, or relativist perceptual failure…or perhaps a third factor (!) I’m not a fan of either of the specific alternative explanations: as for the first, surely the scientific world has something better to do than to deliberately misconstrue Chomsky’s claims for the purpose of debunking something that he has never claimed—this would be a perverse, Quixotic Strawmanism; as to the second, I don’t want to believe that our ideological discourse and practices so cloud our perception that we can only view the world through our particular perceptual filters. So I don’t really have an answer to this question.

Whatever the reason, the empirical observation is that we don’t listen; at least, we don't hear what others say. The latest example of this to come to my attention is found in an April 14th Nature podcast, in which Michael Dunn, from the Max Planck Institute for Psycholinguistics, adds his load to the band-wagon of claims to have shown that Chomsky was wrong about Language Universals…

Michael Dunn‘…The various universalist ideas about language change have a very strong hold on linguistic theory. They predict that language change is determined by human cognitive structures and what all these theories ignore is the contingencies of language evolutionary history…

(Over the years, I’ve read quite a lot by Chomsky, but nothing significant about any predictions for language change; indeed, I cannot recall any work of his dealing with language change. On the basis of the lecture below, though, I’d guess he’d say there has been no language change in our evolutionary history since the ‘Great Leap Forward’).

Michael Dunn: …So we found that if you take language evolution into account many of the things that we thought of as being universal properties of language change turn out to be specific to particular families. So, there are particular historical contingencies within families that make these various correlations between linguistic features happen.

Charlotte Stoddart: These are things like, and things to do, with word order for example in a sentence?

Michael Dunn:All the features that we examined were word order characteristics of different thoughts, yes? [I think their transcription service could do with an overhaul, in the last few words, but the rest is accurate enough].

A lot hinges here on that word “we” (as in “many of the things we thought of as being universal properties of language turn out to be specific to particular language families”). Perhaps Dunn is saying that he and his colleagues previously imagined these properties to be universal, but I don’t think that is what he means by “we” at all. Rather, I’m fairly confident that “we” means “they”: the clear implication here is that 'we'—exclusive we, that is—know better! The thing is though, Chomsky has never proposed that are Language Universals in this Greenbergian sense—unanalyzed properties of surface word-order—either. Indeed, Chomsky has been consistently dismissive of the idea that languages—in the plural, and in the lay sense of the word—have any place in a scientific theory of Language (in the singular, with a capital, a term of art). He calls them E-Languages: E for external, extensional—also, if one wanted to wickedly co-opt an opposition term, epiphenomenal. If E-languages are not even deemed coherent objects of inquiry, why would Chomsky make any claims about their properties? He wouldn’t, of course, and doesn’t, but this hasn’t prevented generations of opponents from asserting that he does, and then working tirelessly to debunk these claims. What he does claim is that there is Universal Grammar, that language acquisition is biologically constrained, and that underlyingly—at a very abstract level—I-language grammars are determined by a invariant set of syntactic principles and mechanisms. These may be universal properties of language, but these do not map in any reliable way to surface properties, like word-order. Indeed, as a matter of logic, they cannot be reflected in surface properties: no generativist denies that surface word-order varies cross-linguistically—we are not fools—so if it is true that languages are ‘cut to a common pattern’ then that pattern must be a very abstract one (DNA, rather than surface morphology, to use a genetic analogy). Universal Grammar has been re-branded in several ways over the past 50 years (LAD, the language faculty, Knowledge of Language, I-language, Faculty of Language), but it has never been called Language (or linguistic) Universals. In this extract from the interview, Chomsky is unusually explicit in distinguishing the two:



Chomsky Video (around 2:35)

The theory of the genetic endowment is often called Universal Grammar, UG…that’s kind of borrowed from a traditional term, but adapted to a new framework, so it actually means something different, it doesn’t refer to properties that are found in all languages, which is the traditional usage…just the theory of the genetic endowment, whatever it is. And if language is a real object, we can be confident that that exists. Now it’s often called a controversial hypothesis, but…the alternative is…magic!’

Lest this is insufficiently clear, let me spell out in two lines, shouting:

UG is not a theory of Language Universals

Language Universals are not part of Universal Grammar

In Sapir-Whorf Redux, I elaborate on this: if you are interested, please click here. But that should do for now. Given this, the problem with the remaining part of the Dunn interview should be clear:

Michael Dunn: so the kind of universal the linguists talk about, I think a lot of them are like feathers on birds that these different word order features do tend to go together and there are good reasons for them going together but there are not universal reasons for them going together.

Charlotte Stoddart: You're not the first to suggest that Chomsky's idea of a universal grammar is wrong, how much support does his theory still have and how do you think your paper will be received?

Michael Dunn: Our results are not consistent with any kind of simple view of a specialized language module. If there was a universal language model, then all languages should follow a single set of rules and the historical development and our phylogenetic approach shows that this is not the case. I think, like any sort of challenge to scientific orthodoxy then people who have more invested in the theory who resisted more than people who have not invested in the theory, I think there is a big change underway in how we view the role of language within cognition and I think there is an emerging consensus that language is much more integrated in general cognition than people had previously thought and that ideas of a simple language module separate from every other aspect of cognition just don't hold water. There is no evidence for them and some evidence against.

Charlotte Stoddart: So the picture that's emerging is that our language is much more a product of more generally the way our brain works but also of our culture than we previously thought?

Michael Dunn: That's right, yes.

There are main three points to observe about this extract. First, the discovery that apparently universal word-order patterns emerged independently, and for different reasons, in different language families, is genuinely interesting, and an important contribution to our understanding of evolutionary change in the world’s languages. It’s hardly earth-shattering stuff, though: I should have thought that it was the null hypothesis, given the massive explanatory contribution of external historical contingencies in the development of languages. (To take a relatively modern example, close to home, consider word-order in Present Day English: but for the influence of Old Norse through Viking-Anglo-Saxon language contact in the Danelaw, and later Old French, as a result of the Norman Conquest, English (the E-language!) would most likely be an unremarkable West Germanic variety, with verb-second and underlying OV word-order. But it’s not: typologically, English is a bastard language, the world’s most sophisticated, lexically resplendent, creole (see, e.g., Bailey & Marold 1977).

The second observation is that Dunn does not correct the interviewer’s assertion that he is suggesting ‘that Chomsky's idea of a universal grammar is wrong’. Charlotte Stoddart can be forgiven this confusion, but Dunn cannot, so either he believes it himself, or knows the difference and is happy to perpetuate the error. Neither conclusion is greatly exculpatory.

Finally, observe that, contra Dunn, none of the evidence for UG—which has always come from internal considerations or from language acquisition, in any case—is any less secure after this body of research than it was beforehand. The theory of UG has never been motivated by, nor grounded in, inductive generalizations about language diversity, and that is not about to change.

The shame of it all is that this research, like that of Dunn’s mentors and colleagues, Stephen Levinson and Nick Evans (Evans & Levinson 2009) is important, valuable and inspiring stuff. Moreover, to the extent that large-scale typological studies of this kind foster interest in language diversity and promote language documentation, such work has far more beneficial social impacts than anything achieved by Minimalists, as it does not greatly matter to UG researchers whether there is one language spoken on Earth or 10,000: at the epistemological depths, there is only one anyway (Faculty of Language, that is). But to the rest of us, it is diversity that makes life interesting, not just academically, but in every phenomenological, experiential respect. In common with many people, a principal reason that I got into linguistics in the first place was my fascination with languages—not Language—and with exotic diversity, not genetic uniformity. Beneath the surface, ‘I’m a deeply superficial person’ (as Andy Warhol is supposed to have said).

Perhaps because of my own career path, I am also a compatibilist: I really believe that it is possible to pursue a theory of Universal Grammar, while remaining open to the possibility that there are no Language Universals, as Evans & Levinson assert, and that, as Dunn says, language is much more integrated in general cognition than people had previously thought. In fact, if you listen to the whole interview, it is clear that Chomsky thinks so, too.

Linguistics doesn’t have to be a zero-sum game.



*For formalist, read generativist—I know at least one cognitive linguist who resents our appropriation of the term, there are probably many others.)