To the editors:
Noam Chomsky’s “The Galilean Challenge” draws on the revolutionary ideas about language that he has been developing for sixty years, seen through the window of the Minimalist Program that, since the early 1990s, has attempted to reduce the language in the human mind to its simplest form. These are combined with his recent interest in language in the enterprise known as the Biolinguistics Program that spans linguistics and neurosciences.1
The article is an admirable example of Chomsky’s non-technical writing, like, say, The Architecture of Language, that explains his ideas without revealing more than glimpses of the syntactic fist of iron inside his velvet glove.2 Such writing can be deceptive, as is clear from Tom Wolfe’s The Kingdom of Speech, which attempts to refute Chomsky largely based on his interviews and newspaper articles rather than his technical writings.3 Chomsky does not address the recent criticisms from anthropologists and cognitive psychologists, say, Dan Everett and Vyvyan Evans, now popularized in books like Wolfe’s.4
One thrust of the argument is that we need to explain why the language of human beings differs from animal communication or, as Cicero put it: “Hoc enim uno praestamus vel maxime feris, quod conloquimur inter nos...” (The one thing in which we are especially superior to beasts is that we speak to each other).5
A defining characteristic of human language is the creative aspect of language use: how human language is continually coming up with sentences that have never been said before. You may never have read any sentence in Chomsky’s article before but this does not make them any more difficult to understand than sentences you have already heard. Any theory must then explain not only how we deal with sentences we have encountered before but also those that we might potentially encounter or produce in the future.
A second thrust is what Chomsky has called Plato’s problem: “How do we come to have such rich and specific knowledge, or such intricate systems of belief and understanding, when the evidence available to us is so meager?”6
Children could not have learned some aspects of the language present in their mind from the sentences they encounter. This seems a version of the theological “argument by design” used by William Paley for the existence of a creator: “The marks of design are too strong to be got over. Design must have had a designer. That designer must have been a person. That person is GOD.”7 Or a variation on the joke about the man who asked how to get to Tipperary and was told that he could not get there from here. Children could not derive syntactic principles such as structure dependence from the actual evidence available in the sentences produced by their parents and caretakers. So, as Chomsky says here, “Children know the principles of the internal language without evidence.”
If certain elements of human language are not learnable from evidence, they must be already present in the mind, that is to say, innate. Ideas of what these innate elements may be have reduced over the years to fit the demand for simplicity and are chiefly now concerned with recursion: how one structure may be embedded in another example of the same structure. In this article, however, Chomsky still sees structure dependence—the principle that “all known formal operations in the grammar of English, or of any other language, are structure-dependent”—as a key example, even if the syntactic landscape around it has changed out of all recognition since 1971.8
Accepting these two main thrusts at face value, we can argue that there are two related issues that are overlooked, namely knowledge of written language and bilingualism.
Spoken and Written Language
Chomsky talks of how language “can be externalized through some sensory modality, such as speech,” “signing, or even touching,” but, oddly, not through writing, the most common externalization for most of us today, whether with a screen or keyboard or in more traditional modes. From Aristotle to de Saussure, linguists have assumed that writing is simply a representation of speech, most famously in Bloomfield’s words: “Writing is not language, but merely a way of recording language by means of visible marks.”9 There is no need to spend much time describing an ancillary system.
The grounds for the primacy of speech over writing in adults have, however, mostly been challenged in recent years. Written language has its own characteristics that are not simply reflections of the spoken. Some concern syntax. Chomsky’s example:
- Birds that fly instinctively swim.
is book-ended by the initial capital B and the final full stop. It would be ungrammatical in written language if these were missing:
- birds that fly instinctively swim
or if it lacked the word space used in modern English as a word separator:
- birdsthatflyinstinctivelyswim
Most language experiments involve written, rather than spoken, sentences and words, through grammaticality judgments of written sentences, eye-tracking of reading of written texts, and the like, assuming the written language provides as straightforward an entry to the language in the mind as the spoken.
The only definition of a word that works in English is anchored in the written language: a “sequence of letters without any spaces.”10 Indeed some linguists argue that words are artifacts imposed on speech by linguists who learnt to read in languages where word spaces form part of the script, and that the phoneme, central to much linguistics, is derived from alphabetical scripts that use letter-sound correspondences rather than character-based scripts like Chinese.11 If linguists themselves cannot separate speech from writing in their professional lives, speech and writing clearly have a relationship that is far from a straightforward subordination of writing to speech.
Speech and writing are often seen now as different ways of externalizing language, so far as literate adults are concerned; the acquisition of writing transforms the language knowledge in the mind. Much of our syntax and vocabulary comes from our reading, not our listening. Michael Halliday and Christian Mattheisen claim, for instance, that: “The sound system and the writing system are the two modes of expression by which the lexicogrammar of a language is represented.”12
The routes for acquiring speech and writing obviously differ but a full account of the language in the adult human mind cannot ignore writing any more than it can ignore phonology. The quotation from Galileo on which Chomsky pins his argument after all describes written language; communicating to those “distant by mighty intervals of place and time” and to those “who are not yet born” was only possible through writing in Galileo’s time, even if modern technology now allows speech the same privilege. This vital way of externalizing language in the human mind should form part of the picture, rather than be dismissed implicitly as parasitic on spoken language.13
Two Languages in One Mind
In a 1980s interview with François Grosjean that seems only to exist in samizdat form, Chomsky asked:
Why do chemists study H2O and not the stuff that you get out of the Charles River? … You assume that anything as complicated as what is in the Charles River will only be understandable, if at all, on the basis of discovery of the fundamental principles that determine the nature of all matter, and those you have to learn about by studying pure cases.
In other words, language needs to be studied in an edited idealized form, far removed from the actual language of a particular individual, because science necessarily involves cleaning the data up.
Throughout Chomsky’s article, he deals with the typical human being as having a single consistent language, rather than the agglomeration of dialects, genres, and languages commanded by individuals in the real world. To include people who speak more than one language
would not be “pure” in the relevant sense, because it would not represent a single set of choices among the options permitted by UG [Universal Grammar] but rather would include “contradictory” choices for certain of these options.14
It is not that he denies the existence of bilingualism, but that he insists that an idealization to a monolingual native speaker permits the simplest possible analysis of the language faculty in the human mind: “In most of human history, and in most parts of the world today, children grow up speaking a variety of languages ... That is just a natural state of human beings.”15
Effectively, Chomsky strips away the other language or languages in the mind to reveal a pure single language. Yet arguably most human beings in the world today use more than one language to some extent. The monolingual is far from typical in most countries. Stripping H2O down to its “pure” elements, hydrogen and oxygen, might be scientific, but it would tell you little about water; bilingualism is not adding a second language to a first but a new state of the language in the mind, in which neither the first nor the second language is the same as that of monolingual speakers. Human beings all have the potential to acquire more than one language. The reason why some remain monolinguals is lack of exposure to a second language in childhood, an accident of their environment, not proof that the language faculty consists of only one language.
Chomsky’s specific example of the principle of structure-dependence can be applied to second language (L2) users. In an experiment I carried out, 140 L2 users of English and 35 users of English as a first language (L1) were asked to judge whether sentences like A. Sam is the cat that is black, B. Is Sarah the woman who is early? and C. Was Bill was the man who French? were OK, Not OK or Not sure.16 The C sentences violate structure-dependence as the “wrong” was from the subordinate clause has moved to the front of the sentence rather than the was in the appropriate structural configuration within the main clause. One hundred percent of the L1 users judged the C sentences ungrammatical at least 5 times out of 6, despite most likely never having encountered sentences like this before that break the principle. But ninety-four percent of L2 users also judged them correctly, though they too were unlikely to have met them before. L2 users then had essentially the same feeling for structure dependence as L1 users; they knew it without “relevant evidence,” as Chomsky puts it.
This knowledge might of course have been carried over from the other languages the L2 users knew; they might simply have been transferring structure dependence from their L1 to their L2.
But languages that do not form questions through displacement do not need structure dependence. Some language principles are not needed in some languages. Universality means no language breaches the principle rather than all languages manifest it. Polish, Finnish, and Dutch have such questions, so their speakers could base their knowledge of English on their L1 knowledge. One hundred percent of the L2 users with these L1s indeed spotted the violation. Japanese, Chinese, and Arabic lack such questions, so their speakers would not be able to transfer structure dependence to English questions, yet eighty-seven percent of the L2 users with these L1s nevertheless detected the violation. All of the L2 users were therefore overwhelmingly successful. While there is a difference between the two groups, this may be due to other factors, such as Japanese, Chinese, and Arabic all having different scripts from English while Polish, Finnish, and Dutch do not, written language again playing its role.
The conclusions that are drawn from structure dependence by Chomsky are then equally true for L2 users. As he says in the article: “Structure-dependence follows from principles of universal grammar deeply rooted in the language faculty.”
But this applies as much to second languages as to first. The starting premise of the theory has to be that all human beings may learn more than one language. Whether all the languages in the mind form separate entities or are merged together is one of the questions that second-language acquisition research has been trying to settle for many years. But it is just as crucial to mainstream linguistics; language learning theories are inadequate if they fail to account for the human potential to acquire more than one language.17
The Galilean challenge for me is then how to explain that Galileo wrote in two languages. His words have come down to us because they were written down, as in the quotation; writing was a crucial part of his language. Like any educated person in the Europe of his period, he used two languages, in his case Italian and Latin. The two abilities of literacy and bilingualism are not attributes of geniuses alone but part of the normal condition that we are trying to explain. Idealizing them away from the language in the mind is an unrealistic simplification.
Vivian Cook
Noam Chomsky replies:
Vivian Cook concludes that:
The two abilities of literacy and bilingualism are not attributes of geniuses alone but part of the normal condition that we are trying to explain. Idealizing them away from the language in the mind is an unrealistic simplification.
The first sentence is certainly correct. So are many other abilities of language users.
The second sentence is technically accurate; all idealizations are “unrealistic,” by definition. The question, however, is whether the idealizations are legitimate. It is “unrealistic” to consider a ball rolling down a frictionless plane, but the question is whether by doing so, we learn about laws of motion that hold of more complex phenomena. All of science is based on assuming that there are idealizations that are legitimate in this sense. Every experiment is an idealization.
I do not see anything in Cook’s discussion to suggest that the idealizations mentioned are illegitimate. The examples cited, e.g., concerning L2, seem to support their legitimacy. I am unaware of evidence suggesting that the study of the language faculty (UG) is affected by the fact that some cultures have writing systems.
I do not, then, see any issue beyond the standard problem of determining when idealizations are legitimate—specifically, the idealizations that linguists employ constantly, and inevitably, including the idealizations that enter into study of multilingualism and literacy.
A minor point. I have in fact discussed Everett’s claims. One of the few accurate parts of Wolfe’s strange exercise, which Cook cites, is his quotation of my explanation to him, in a phone interview, of why his entire project collapses. Whether or not Everett’s claims about Pirahã are correct (they apparently are not), they have nothing to do with the sole issue concerning recursion: is it a property of the faculty of language? The sources cited by Everett and Wolfe makes that clear and explicit, as it has been for 60+ years. Pirahã speakers have no problem learning Portuguese. End of discussion. Similarly, if a tribe were found where everyone wears a patch over one eye, it would tell us nothing about the faculty of binocular vision.
A tempest with no teapot.
