Reply to Longa

 

Geoffrey Sampson

University of Sussex

 

 

 

In response to an invitation by the editors of Linguistics, I should like to comment on Víctor Longa’s review (Longa 1999) of my Educating Eve (Sampson 1997).

 

Educating Eve was a response to Steven Pinker’s best-selling book The Language Instinct (Pinker 1994), which very successfully popularized the doctrine of linguistic nativism – the idea that human beings are born with innate knowledge of language structure.  (Longa makes a point of preferring the term “innatism”, a word which is not found in standard dictionaries; I shall use the normal English term for the doctrine of innate knowledge.)  Educating Eve identified each of the many separate arguments used by Pinker, by his predecessor Noam Chomsky, and by some other writers who have contributed to the wide current awareness of linguistic nativism; and it set out to show that each one of these arguments was logically fallacious, or founded on false premisses (or, in some cases, both).

 

Longa does not believe that the book made its case.  It would be excessive to consider all his disagreements with me, so I shall sample his criticisms here by examining the points made in the first two pages of his MS after the introduction (i.e. the first seven paragraphs of Longa’s §2).

 

Longa’s §2 is structured as a set of subsections responding in turn to the sections of chapter 2 of Educating Eve, which take issue successively with various arguments of Noam Chomsky’s for linguistic nativism:  the argument from speed of first-language acquisition, the argument from age-dependence, the argument from poverty of data, and several others.  The first section of my chapter pointed out that the argument from speed of acquisition is vacuous.  To argue for linguistic nativism on the ground that the child’s acquisition of his first language is remarkably rapid is to claim that the two or three years the process takes is much shorter than the time, T, that it would be predicted to take if the child began with no innate knowledge.  But Chomsky, Pinker, and other linguistic nativists never identify any specific finite period, T, which would be required in the absence of innate knowledge.  They give us no basis for estimating T even in order-of-magnitude terms.  Would language acquisition without innate knowledge take five years? Twenty?  Five hundred?  Since no basis for any particular prediction is forthcoming, we are entitled to ask (in words Longa quotes from Educating Eve) “Why is it appropriate to regard a learning period of two years or so as ‘remarkably fast’ rather than ‘remarkably slow’?” – or indeed as “about what one would expect”.

 

Longa concedes that on this point “S[ampson] is correct”; but then he blurs the concession, by saying that “if contextualized in the overall acquisition process, one realizes that the child is able to acquire a highly complex grammar system without explicit instruction, … ”, and that in view of this and similar considerations my argument “loses its strength”.  This confuses separate arguments.  The fact that first-language acquisition proceeds without explicit instruction is one of the premisses of the “poverty of data” argument for linguistic nativism.  Educating Eve refutes that argument in detail, too, in the section devoted to it; but the section which Longa is addressing at this point is concerned with the argument from speed of acquisition.  Longa says nothing to cast doubt on my conclusion there, that the argument from speed of acquisition is entirely vacuous.

 

Because the linguistic nativists use many separate arguments to support their beliefs about innate linguistic knowledge, when they find themselves pressed hard about the weakness or emptiness of any one of their arguments it is always open to them to urge that the criticism fails because there are so many other arguments all tending in the same direction.  In this situation, what a sceptic has to do is to identify each separate strand of argumentation used by believers, and patiently, one after another, demonstrate in each case that the argument fails.  That is what Educating Eve does.

 

However, Longa also claims that I contradict myself with respect to the speed of acquisition.  I deny that language acquisition in general can meaningfully be described as remarkably fast; then later, in a discussion of Derek Bickerton’s hypothesis of discontinuity between young children’s “protolanguage” and adults’ full language, I agree that when children begin to acquire grammar their grammatical structures seem to increase in complexity rather abruptly.  This is not a contradiction.  It is a predictable consequence of the nature of grammatical recursion, the concept under discussion in the passage cited by Longa.

 

A phrase-structure grammar is recursive if, for at least one of its nonterminal symbols, there is a valid derivation in which that symbol occurs more than once on a path from leaf to root node.  (Note that recursivity is a property of a set of rules:  a grammar can be recursive, without any individual rule having the same symbol on both left- and right-hand sides.)  A nonrecursive grammar permits only a finite range of structures, with finite maximum complexity; any recursive grammar permits infinitely many distinct structures, with no finite limit to the complexity of individual structures.  If a set of grammar rules are learned one by one, the set may be large, and the time taken to learn them all may be long, but (assuming that the set as a whole forms a recursive grammar) there must be some particular point in the sequence where the addition of a single rule changes the set learned so far from nonrecursive to recursive.  Thus there is no tension between denying that language acquisition is remarkably rapid overall, and accepting that the move from telegraphic speech to complex grammatical output may be abrupt.  Such matters should be familiar to an enthusiast for generative grammar theory.

 

Longa finds it “extravagant” of me to offer, as one explanation for the success of the logically-vacuous argument from speed of language acquisition, the fact that people have a natural tendency to be favourably disposed to small children, so that they find it emotionally easy to accept ideas which imply that children’s achievements are impressive.  I do not know whether Longa means that my explanation may be correct but ought not to be mentioned in a scientific context, or whether he thinks it implausible that such emotional considerations could have played any part in the success of linguistic nativism at making converts.  If the latter, I find it naïve to imagine that the hundreds of thousands of people worldwide who have accepted the idea of a human “language instinct”, often on the basis of no more than a few hours’ reading of popular books and/or exposure to one or two radio or television programmes, have all assented to this idea exclusively on the basis of judicious logical assessment of the rival arguments.  Even ideas that are better-founded than linguistic nativism do not spread only in such an ideal manner.  If Longa means the former, I see no good reason to impose such censorship.  It is not helpful to readers to demonstrate that an argument which has won many converts is vacuous, and then stop.  The readers are bound to want to know how the argument could nevertheless have been successful.  If the true explanation is psychological or sociological, it needs to be given, so that readers can understand how rhetorical success was compatible with logical vacuity.

 

In some passages, Longa constructs his case against Educating Eve by choosing to ignore large parts of what the book says on a particular topic.  He refutes my suggestion that differential success in L2 acquisition is explicable partly in terms of different degrees of motivation by saying “it should be supposed that immigrants … will perceive the task as worthwhile”, and then pointing out that adult immigrants vary greatly in degree of L2 mastery.  As I pointed out at some length in Educating Eve, citing individual examples (pp. 89-90), there are immigrants and immigrants.  It is unreasonable to suppose that the degree of psychological identification with a new culture and its language will be comparable as between, say, a Jew who has escaped the Nazi Holocaust in which all his close relatives perished, and a citizen of a peaceful, successful modern State who finds it convenient to take a lucrative overseas job but maintains contacts with his home country.  Does Longa see no relevant difference between such cases?  (It would be open to him to argue against Educating Eve by trying to show that degree of success at L2 acquisition is not correlated with such differences; but Longa does nothing like that.)

 

One crucial idea in the debate about linguistic nativism is the so-called “critical period” or “critical age” hypothesis – the concept, first widely publicized by Eric Lenneberg (1967), that children are biologically programmed to acquire language during a fixed period that ends at about the onset of puberty.  In connexion with this hypothesis, Educating Eve discusses the famous case of “Genie”, classically documented by Susan Curtiss (1977).

 

According to Longa,

S[ampson] only considers the case of Genie … to surprisingly point out that “Susan Curtiss herself regarded Genie as refuting the strong version of Lenneberg’s claim”. 

Longa is not the only critic of Educating Eve who has taken issue with the quoted sentence.  Feargal Murphy, in an electronic book review (LINGUIST List 9.734, 18.5.1998), similarly writes:

The author … states that “Susan Curtiss herself regarded Genie as refuting the strong version of Lenneberg’s claim, that natural language acquisition cannot occur after puberty” (p. 37).  Susan Curtiss might want to respond to that one herself. 

The only interpretation I can place on Longa’s “surprisingly”, and Murphy’s “might want to respond to that one”, is that these critics are insinuating that Educating Eve misrepresents Curtiss’s book, without explicitly saying so.  I am sure this is the inference that most readers would take from Longa’s and Murphy’s words, and I cannot imagine what other force they can be intended to have.  Indeed, Longa later describes the view I attribute to Susan Curtiss as “quite surprising and erroneous” – the last word quoted perhaps makes the charge of misrepresentation explicit.

 

Susan Curtiss’s actual words were: 

There are two versions of Lenneberg’s critical age hypothesis that can be extracted from his writings:  (1) a strong version … and (2) a weak version …  The “strong” version, that natural language acquisition cannot occur after puberty, can be dismissed in this case.  As demonstrated in Part II, Genie is acquiring language from “mere exposure.” (Curtiss 1977: 208-9) 

My report of Curtiss’s view could scarcely have been more exact.  I included a page reference to Curtiss’s remark, so, assuming that Longa and Murphy were not deliberately aiming to mislead their readers, their standards of scholarship here leave something to be desired.

 

(It may of course be no coincidence that both reviewers make the identical mistake.  Longa may have been simply echoing Murphy’s earlier comment, rather than making an independent judgement.  But that too would be unimpressive scholarship, if it is what happened.)

 

It is true that Susan Curtiss later appeared to change her mind fairly radically about linguistic nativism.  In Educating Eve I discuss, for instance, how she subsequently took a much more Chomskyan line in connexion with another case, “Chelsea”, for whom the published data were so extremely scanty as to be compatible with virtually any theory about language acquisition.  Peter Jones (1995) has documented in detail the way in which Curtiss’s discussions of Genie in later publications contradicted what she wrote in her 1977 book, although no fresh evidence was available to her and no explanations for the contradictions were given.  But Curtiss’s 1977 book was a careful, scholarly study, in which (contrary to Longa’s suggestion) she said precisely what I reported her as saying.

 

We see, then, that in two MS pages Longa’s tally of points against me amounts to one confusion of the issue, one claim that a non-contradictory statement is contradictory, one arbitrary suggestion that a relevant point is too “extravagant” to count, one failure to notice that my book addressed a point which he raises as if new, and a repeated false imputation of misrepresenting a third party.

 

When a fundamental doctrine is almost universally held by members of a social group, they sometimes see criticism of the doctrine as not merely intellectually mistaken but as something that ought not to be uttered.  There are hints of this attitude in Longa’s review, for instance when he describes the tone of Educating Eve in his introduction as sometimes “irritating”.  I trust that there still exists a community of readers who see the doctrine of linguistic nativism as a point of view open to debate like any other, rather than as an untouchable axiom.  If Educating Eve encounters no criticisms more telling than the sample of Longa’s remarks analysed above, I feel confident that such readers will agree that linguistic nativism is refuted.

 

References

 

Curtiss, Susan  1977  Genie: a Psycholinguistic Study of a Modern-Day “Wild Child”.  New York:  Academic Press.

Jones, Peter  1995  “Contradictions and unanswered questions in the Genie case: a fresh look at the linguistic evidence”.  Language and Communication 15.261-80.

Lenneberg, Eric  1967  Biological Foundations of Language.  New York: Wiley.

Longa, V.M.  1999  Review article on Sampson (1997).  Linguistics 37.325–44.

Pinker, Steven  1994  The Language Instinct.  New York:  Morrow.

Sampson, Geoffrey  1997  Educating Eve.  London: Cassell.