[ In the end, having commissioned this the editors of this encyclopaedia refused to print it. I think they were shocked that anyone dared to disagree with the sainted Noam! ]
Empiricism is a philosophical principle which emphasizes the role of experience of the external world in the growth of knowledge. (The word derives from Greek empeiría, experience.) The principle has two facets. On one hand, it holds that anyone, as a child beginning to learn while still too young to understand verbal teaching, does as a matter of fact gain knowledge mainly by distilling it from sense-data, rather than by extracting knowledge from mental mechanisms that are built-in at birth (perhaps coded in the child’s DNA). On the other hand, the empiricist principle urges that those whose social role is to generate new knowledge (e.g. scholars, scientists) ought to make their discourses accountable chiefly to interpersonally-observable data, rather than treating great names of the past as authoritative or relying on aprioristic argument and speculation.
These two aspects of empiricism, which we might respectively call descriptive empiricism and normative empiricism, are clearly interrelated; but they are in principle quite separable. For instance, whether or not it is likely in practice, logically it is entirely conceivable that psychologists, studying the behaviour of infants using standard scientific techniques of observation, hypothesis-testing, etc., might discover that they gain significant swathes of factual knowledge without any exposure to relevant experience (or teaching), purely as a consequence of the operation of innate cognitive equipment. If that were to happen, empiricism in the normative sense would have undermined descriptive empiricism.
Before we examine the relevance of these ideas to linguistics, we should briefly look at the alternatives to them. To many people, empiricism (whether normative or descriptive) sounds like “plain common sense”, so that it is not obvious why it needs a long name ending in –ism. And, at least in the English-speaking world at most times after the Middle Ages, empiricism has indeed been the dominant intellectual attitude. But it is not the only possible attitude, and at other times and places it has not always held sway. We shall see that, thanks in part to linguistics, empiricism has recently lost some of its prestige even within the English-speaking world.
Arguments for and against empiricism, like many philosophical controversies, go back to the ancient Greeks. But, in the modern period, descriptive empiricism was contradicted most notably by the French philosopher René Descartes (1596–1650), who held that although a small child cannot in a straightforward sense be said to “know” many things that an experienced adult knows, nevertheless knowledge is already present within the infant much as it is present in the adult while he is asleep. (Differences between men and women have no relevance to anything in this entry, so rather than repeating the cumbersome “he or she” I shall adopt the traditional solution of making “he” stand for hypothetical individuals of either sex, without implying thereby any disrespect for females.) What we think of as “learning” by the child does not involve acquiring novel concepts, but merely (according to Descartes) discovering the truth or falsity of propositions built out of ideas that were implanted in the child’s mind before birth, though they were initially dormant.
Descartes illustrates the concept of “innate ideas” using highly abstract examples, such as the idea of God or the geometrical idea of a triangle – it is relatively plausible to suggest that these cases are independent of physical experience. But he claims that his principle applies much more widely: “all [propositions] are innate in us” (letter to Mersenne, 22 July 1641). This rejection of descriptive empiricism is called either rationalism or nativism. (The logic of these names is that knowledge stems not from experience but from the faculty of reason, Latin ratio, and that knowledge is “native”, i.e. inborn, in us rather than acquired after birth.)
The foundational text of modern empiricism, An Essay Concerning Human Understanding by the English philosopher John Locke (1632–1704), was probably consciously intended as a reply to Descartes.
In the case of normative empiricism one cannot point to an influential individual, comparable to Descartes, who has systematically opposed it in modern times. Rather, the hallmark of the intellectual life of the modern period, from the seventeenth century onwards, has largely been a general acceptance of normative empiricism. In the Middle Ages, although some thinkers operated empirically, many others were more interested in deducing theories about matters of fact from aprioristic principles, or in “arguments from authority” – deciding how the world must be by working out what would be most compatible with pronouncements by some great name of antiquity.
Until the 1960s, linguists took empiricism for granted. Infants were seen as acquiring their first language by listening to the speech of their elders and imitating it (descriptive empiricism), and linguists were expected to pursue their researches by observing speakers’ behaviour and using it to test the validity of linguistic theories (normative empiricism). However, the American linguist Noam Chomsky (q.v.) disagreed with empiricist assumptions, and succeeded for many years in carrying much of the discipline with him. Chomsky’s anti-empirical linguistic stance is better called “nativist” than “rationalist”: he stresses the claim that much of language structure is innate in the human mind, without linking that claim to the particular mental faculty of reason. Chomsky’s linguistic nativism was briefly adumbrated in his 1964 book Current Issues in Linguistic Theory, and was developed at length in later writings.
Chomsky’s linguistics contradicts descriptive empiricism by claiming that the child’s ability to master its mother tongue depends chiefly on much of the structure of language being encoded in our genetic inheritance, as our bodily structure uncontroversially is – this “innate knowledge of language” needs only the trigger of exposure to speech in order to be aroused and applied by the growing child, and the only specific features of the mother tongue which have to be learned from experience are those details which differentiate one language from another. (Chomsky, perhaps remarkably, holds that these are fairly trivial as compared to the properties which are common to all human languages.)
Chomsky also rejects normative empiricism as a recipe for doing linguistic research. There is obviously a difference between linguistics and sciences such as, say, meteorology, or marine biology: a student of the latter is considering things which lie entirely outside himself, but an English-speaking linguist describing some aspect of the English language is discussing, among other things, features of his own cognitive functioning. This is uncontroversial; but it leads Chomsky to hold that grammatical description need not be based on objective evidence, because a native speaker can look inside his own mind and extract reliable “intuitions” about what he can and cannot say.
Chomsky’s rejection of descriptive empiricism was revived and reinforced a generation later by the Canadian linguist Steven Pinker in his book The Language Instinct (1994), which made linguistic nativism accessible to an audience far wider than Chomsky’s readership. Although both Chomsky and Pinker base their arguments for innate knowledge mainly on facts about language and language-acquisition, for both of them language is merely an aspect of cognition where the evidence for innateness happens to be specially clear. They believe that in other areas of cognition also, the contents of our minds are largely what our genetic endowment makes them.
This turning away from descriptive empiricism became noticeable in a variety of human sciences towards the end of the twentieth century, and was recognized (e.g. by Donald Broadbent, 1973: 189) as stemming from the influence of Chomsky’s linguistics. (Many strands of late-twentieth-century academic discourse turned away from normative empiricism also, though there the influence of linguistics was probably not significant.)
A point worth stressing before we look further at empiricism in linguistics is that the general philosophical issue is a less-or-more rather than all-or-nothing question. Thinking of descriptive empiricism, it is obvious that the newborn child brings something important of its own to the process of gaining knowledge: a stick or stone will never know anything, no matter what external stimuli impinge on it. Conversely, not all knowledge can possibly be innate – I could not know when I was born in 1944 that the prime minister in 2009 would be Gordon Brown. Similarly for normative empiricism, while the modern scientific method expects hypotheses to be tested against objective experience, it accepts that as candidates for testing they emerge in a fairly mysterious way from the scientist’s mind. And conversely, even a highly aprioristic mediaeval thinker would undoubtedly have conceded some relevance of experience to knowledge. So, when we describe someone as an empiricist (or as opposed to empiricism), what we mean is that he sees the role of experience as in some respects greater (or less) than it is deemed to be by the current consensus, or by some particular comparison group.
In those terms, Chomsky unquestionably ranks as an anti-empiricist. Although standard scientific method accepts that hypotheses depend on individual scientists’ intuition, it vehemently opposes the idea that they can be relied on as veridical (as Chomsky believes in the case of native speakers’ intuitions about their language). In the case of descriptive empiricism, probably no-one for centuries before Chomsky believed that children inherit knowledge of detailed language structure in the way he claims.
Since Chomsky was the first to argue against empiricism in linguistics, and other linguistic nativists have essentially developed Chomsky’s thought rather than taking opposition to empiricism in additional directions, it is inevitable that an account of empiricism in recent linguistics must be about reactions to Chomsky’s ideas.
There have been many such reactions. On the normative issue, it was argued from an early stage that introspective judgements are not a reliable guide to the way a language is spoken. People’s intuitions about a given construction are often vague or contradictory, and sometimes a speaker is plain wrong about his own usage. These points were demonstrated persuasively by William Labov (1975), who for instance discussed the “positive anymore” construction heard in the Philadelphia area. Speakers of this dialect say things like “John is smoking a lot anymore”, to mean “John is smoking a lot now, regrettably”; but many Philadelphians claim in all sincerity to be unfamiliar with this construction and not to understand it, and guess wrong when asked what it might mean – and yet are themselves heard to use it.
Turning to descriptive empiricism, the nativists’ arguments against this are many and various (so that space does not permit a comprehensive discussion). But, for instance, we have seen that belief in innate knowledge of language is based in part on a claim that all human languages share extensive common features – language universals – which can only be explained as the consequence of a shared genetic inheritance. The strength of that argument clearly depends on how extensive the language universals are, and empiricist linguists have complained that we are given few specifics to support the claim that universals exist. Geoffrey Pullum and Barbara Scholz (2002) combed through the nativist literature and found just four examples of language universals quoted (each of which they regarded as questionable). Steven Pinker (1998) himself conceded that “U[niversal] G[rammar] has been poorly defended and documented in the linguistics literature”. From a wide-ranging investigation of diversity among languages Nicholas Evans and Stephen Levinson (2009) conclude that language universals are a “myth”.
Many applied linguists have been impressed by accounts of the Genie case. “Genie” was a girl born to insane parents in California, who reared her as if she were an animal, with no exposure to speech or most other normal human experiences, until at age 13 the authorities discovered and released her, and attempted to rehabilitate her. In the years immediately following her release, Genie did not become a normally-competent speaker of English, though she acquired some language ability; nativists have claimed that this lack of success was because Genie first encountered language only after the biologically-fixed age at which a child’s innate knowledge is available to be aroused. However, the linguist who originally investigated and documented this harrowing case, Susan Curtiss, at the time saw Genie as evidence against, not for, linguistic nativism (Curtiss 1977: 208–9). In later years Curtiss appears to have become converted to the nativist position, and took to writing about Genie as if the case contradicted descriptive empiricism; but, as Peter Jones (1995) points out, it is not clear what justified that volte-face – it cannot have been fresh evidence about Genie’s development, because there was no new evidence. (To safeguard Genie’s welfare the authorities forbade further scientific study.)
One reason why attacks on descriptive empiricism seemed persuasive to some in the late twentieth century was that certain empiricists had developed the principle in an extreme and strange direction. The behaviourist school of psychology founded by J. B. Watson (1878–1958) argued, in essence, that because scientific method required psychological theory to be based on observations of stimuli impinging on people and on their behaviour, rather than on subjective introspection about what might be going on within the mind, therefore patterns of linkage between stimuli and behavioural responses are all there is – “mind” is a layman’s fallacy. This conflates normative with descriptive empiricism, and seems no more sensible than saying, because I can only see the keystrokes that users enter into a computer and the output appearing on screen or from the printer, that therefore the computer has no internal workings – concepts like “compiler” or “half-adder” are naive fictions. One of the earliest writings that brought Chomsky to public attention was a slashing review in 1959 of a behaviourist account of language (Skinner 1957). If empiricism were identified with behaviourism it would certainly be natural for linguists to reject empiricism. However, behaviourism is a crude distortion of empiricism rather than a typical version of it. Locke would have given behaviourism short shrift.
The computer has made a large concrete difference to the outlook for empiricist linguistics: the availability of computers has enlarged the range of issues which can be empirically checked. In the 1950s, linguistics focused mainly on phonology, which typically involves a few dozen phonemes combining in limited ways, so one could expect to encounter all possibilities within a reasonable time. When attention shifted in the 1960s to syntax, where many thousands of words can be assembled into effectively innumerable combinations, it was less practical to check whether particular sequences occur simply by listening out for them. As a result, basing syntactic theory on native-speaker introspection was appealing. However, electronic corpora – large samples of real-life language usage – are now commonplace, and these can quickly be searched to answer questions which would take too long to resolve by manual methods.
One consequence is that nativist beliefs which for years were accepted because they sounded plausible are now being objectively tested, and often refuted. For instance, the nativist doctrine of poverty of the stimulus asserts that children regularly acquire grammar rules for which the speech they hear includes no evidence. Chomsky’s standard example relates to the rule for forming questions in English; in order to establish empirically that this rule is “structure-dependent” (as it in fact is), a child would need to hear a particular type of question uttered, and according to Chomsky that question-type “rarely arise[s] … you can go over a vast amount of data of experience without ever finding such a case” (Piattelli-Palmarini 1980: 114–15). One may wonder at the confident tone of this assertion, but when it was made (in 1975) it would probably have been difficult or impossible for a sceptic to test. However, in 1995 the British National Corpus was published, containing among other material some four million transcribed words of casual, spontaneous speech by a cross-section of the British population. Using this, Geoffrey Sampson (2002) calculated that an average rate at which one will hear the forms at issue must be at least once every few days. In this and other ways, empirical linguistic research is suggesting that the language data available to a child are much less “impoverished” than nativists have supposed.
In the early 21st century, normative empiricism is re-establishing itself within linguistics; linguists are routinely challenged to back up speculative-sounding claims with objective evidence. There remain many linguistic nativists who reject descriptive empiricism. But nativist theories are now under pressure within the discipline, in a way which forty years ago they were not.
References
Broadbent, D. E. (1973). In defence of empirical psychology. London: Methuen.
Chomsky, N. (1959). Review of Skinner’s Verbal Behavior. Language, 35(1), 26–58; reprinted with new preface in L. A. Jakobovits & M. S. Miron, Readings in the psychology of language (pp. 142–71), Englewood Cliffs, New Jersey: Prentice-Hall, 1967.
Chomsky, N. (1964). Current issues in linguistic theory. The Hague: Mouton.
Curtiss, Susan (1977). Genie: A psycholinguistic study of a modern-day “wild child”. New York & London: Academic Press.
Evans, N., & Levinson, S. C. (2009). The myth of language universals: language diversity and its importance for cognitive science. Behavioral and Brain Sciences, 32, 429–92.
Jones, P. (1995). Contradictions and unanswered questions in the Genie case: A fresh look at the linguistic evidence. Language and Communication, 15, 261–80.
Labov, W. (1975). Empirical foundations of linguistic theory. In R. Austerlitz (Ed.), The scope of American linguistics (pp. 77–133). Lisse: Peter De Ridder Press.
Locke, J. (1690). An essay concerning human understanding.
Piattelli-Palmarini, M., Ed. (1980). Language and learning: The debate between Jean Piaget and Noam Chomsky. London: Routledge.
Pinker, S. (1994). The language instinct: The new science of language and mind. New York: William Morrow.
Pinker, S. (1998). Posting 9.1209, 1 Sep 1998, on the LINGUIST List (linguistlist.org/issues/9/9-1209.html).
Pullum, G. K., & Scholz, Barbara C. (2002). Empirical assessment of stimulus poverty arguments. The Linguistic Review, 19(1–2), 9–50.
Sampson, G. R. (2002). Exploring the richness of the stimulus. The Linguistic Review, 19(1–2), 73–104.
Skinner, B. F. (1957). Verbal behavior. New York: Appleton-Century-Crofts.
Suggested readings
Tipton, I. C. (Ed.) (1977). Locke on human understanding. Oxford: Oxford University Press.
Sampson, G. R. (2005). The “language instinct” debate (revised ed.) London & New York: Continuum.
