The following essay was submitted by Konstantin Andreev as a course paper in Language Acquisition at Dalarna University, Sweden, in 2006.
1. Introduction
If you have ever attempted to learn a foreign language as an adult, chances are you have wondered why something that little children accomplish with apparent ease should require so much rote learning, self-discipline and concentration now that you have grown up – with no guarantee whatsoever of eventual success. It seems sensible – indeed, almost self-evident – to assume that there must be something fundamentally different between how we acquire language in early childhood and how we learn it as adults. But then again, until science proved otherwise, it seemed self-evident that the sun went around the Earth and that a heavy falling object would surely hit the ground sooner than a light one. So what does the science of psycholinguistics have to say on the matter?
That children are naturally superior language learners may sound like a fairly straightforward suggestion, easy to put to a test and eventually prove or reject; however, this is not the case at all. What strikes one about the Critical Period Hypothesis is how many different things it has meant to different people. David Singleton of Trinity College Dublin, at the end of his brief summary of the literature on the subject, likens the idea of a critical period for language acquisition to the “mythical hydra, whose multiplicity of heads and capacity to produce new heads rendered it impossible to deal with” and generally dismisses it as a poorly-defined conjecture that “cannot plausibly be regarded as a scientific hypothesis either in the strict Popperian sense of something which can be falsified or indeed in the rather looser logical positivist sense of something that can be clearly confirmed or supported” (Singleton 2005: 12). The search for a linguistic critical period is thus nothing but a quest for another Holy Grail or, at best, Martian canals – a no doubt fascinating but ultimately unscientific affair. Considering the amount of research that has been invested into the hypothesis, these are strong words, and one would assume that Singleton had good reasons for making such a claim. The question I intend to try and answer for myself in this paper is, therefore, very simple: Does he?
2. How the idea emerged
The concept of “critical periods” in the development of an organism was pioneered in zoology. In the middle of the 20th century, zoologist Konrad Lorenz, who studied greylag geese, found that newly-hatched goslings would develop an attachment to and start following any moving shape that they happened to see during a rigidly limited time period soon after hatching. If that moving shape was Lorenz himself, the goslings would recognise him as their ‘mother’ and ignore the actual mother goose. While the nature of the moving object did not seem to matter much, the time of exposure was crucial: goslings that failed to attach to the mother during the “window of opportunity” seemed unable to do it later. Similar developmental windows of opportunity have since been shown for other animals and traits (Pinker 1994: 293; Bialystok, Hakuta 1994: 56).
Although the Egyptian king Psammettichus is said to have discovered, much to his own chagrin, that children who were not exposed to any language wound up speaking Phrygian, more recent and reliable accounts unequivocally suggest that children who do not hear language will not speak at all. The first person to suggest a critical period for language learning in relatively scientific terms probably was Jean Itard. He was the ambitious French doctor who, in the first half of the 19th century, tried and spectacularly failed to teach language to Victor the Wild Boy of Aveyron, found in a forest at around the age of twelve. In an attempt to explain his failure, Itard speculated that “the apprenticeship of speech”, though very effective in early childhood, must “wane rapidly with age” (Scovel 1988: 131).
In 1959, neurologists Wilder Penfield and Lamar Roberts were prompted by their research into the cortical processes responsible for speech to argue that, after the age of nine, the brain becomes increasingly stiff and rigid, which hinders effective language learning (Singleton 2005: 2). Penfield became convinced that taking up languages in the second decade of life was too late for achieving good results, and he went on to publicly advocate the importance of early foreign language teaching – an idea which has been extremely influential in education and has led, among other things, to the establishment of English-language nursery schools in a number of non-English-speaking countries.
However, the general consensus seems to be that the story of the Critical Period Hypothesis really begins in 1967, when American researcher Eric Lenneberg published his seminal work Biological Foundations of Language. The critical period, as defined by Lenneberg, applies to primary language acquisition as a whole and is limited by “lack of maturation” at the beginning and by “a loss of adaptability and inability for reorganization in the brain” at the end (Lenneberg 1967: 179). The most convincing evidence Lenneberg cited came from young children recovering from acquired aphasia. Unlike adult aphasiacs, whose speech remained severely affected for the rest of their life, child patients were able to fully restore their language if they were less than nine years old before the onset of the disease (Lenneberg 1967: 146).
Arguably, Lenneberg’s definition can be regarded as the classical form of the hypothesis. However, the idea of a critical period for language was bound to fall on fertile ground both in cognitive psychology and in Chomskian linguistics. Conflicting views on the problem began to multiply immediately. Neurological research, case studies of children who had spent the first years of their lives in isolation, late sign-language acquisition in deaf children, studies of the correlation between immigrants’ age of arrival and their phonological and syntactic ability in the language of the host country – these and other sources of evidence, as well as their different interpretations, have led to the emergence of some drastically different ideas about the critical period.
3. What does the critical period affect?
Most fundamentally, it is not clear whether we should look for a critical period only in terms of first language acquisition or if the hypothesis also has to include second language acquisition. Lenneberg assumes that only first language acquisition is crucial since it creates “the matrix for language skills” that can be put to use again later in life (Lenneberg 1967: 176). While it is obvious that adults can and do learn foreign languages, many have argued that native-like competence cannot in principle be achieved by a post-pubescent learner. Thus, whether Lenneberg’s language-learning “matrix” itself is affected by age has been a cause of much disagreement, and Bialystok and Hakuta (1994: 80) sum up the relevant evidence as “at best, confusing”.
That issue aside, Lenneberg originally hypothesised that the critical period imposed a time limit on first-time acquisition of our entire language ability; however, such a broad definition has not been supported by many researchers.
A case has been made for a critical period that applies only to phonology and has no impact on the other linguistic skills. Thomas Scovel (1988: 185) asserts that post-pubescent language learners, while still able to improve their accents, “will never learn to pass themselves off as native speakers phonologically”. He links this to changes in neuroplasticity and suggests a possible evolutionary explanation: for hominids, the ability to tell a member of your own group from a stranger by voice alone might have been a useful asset when visual recognition was limited by darkness, distance or similarities in appearance (Scovel 1988: 80).
Researchers working within a nativist frame of reference have naturally suggested that maturational constraints affect the Language Acquisition Device, or Universal Grammar. According to some nativists, Universal Grammar deteriorates completely after puberty and any post-pubescent language learning has therefore to rely on general problem-solving abilities, which are not as effective in the acquisition of syntax. Others believe that only certain subparts of Universal Grammar, such as the mapping of abstract syntactic features onto their morphological realisations, are subject to age-related degradation. Still others have argued that the innate language device does not shut off at all; rather, it is the non-innate components of our language ability that deteriorate after puberty (Singleton 2005: 6).
Back to non-UG-based explanations, Robert DeKeyser recently suggested that the critical period might end in an irreversible shift from implicit to explicit language learning; the claim is based on evidence that adults beginning to learn a language necessarily use their verbal analytical ability, which child beginners do not employ at all. According to DeKeyser, children are able to acquire a complex abstract system such as language implicitly whereas adults can no longer do that, which results in more effort-consuming and less effective language learning. The interesting aspect of this view is that language is supposed to be just one of the various complex systems whose acquisition is affected by the shift (Singleton 2005: 6, 9).
4. When is the critical period?
Most people would probably venture a guess that a critical period for language – provided there is one after all – must begin at some time soon after birth and end around puberty, but a falsifiable hypothesis is of course expected to contain a more precise schedule than that. An impressive selection of possible onsets and offsets for the critical period has been proposed by different researches. As summarised by Singleton (2005: 5), the earliest suggestion for the onset seems to be the 6th month of foetal life (Ruben 1997; for phonology) while the latest offset proposed so far is 16 years of age (Ruben 1997; for semantic ability). Other suggestions for the offset age include 1 year (Molfese 1977; for phonology), 4 years (Ruben 1997; for syntax), and 15 years (Long 1990; for morphosyntax). Some suggestions (Long 1990; Newport and Johnson 1989) divide the critical period for each aspect of language ability into two distinct phases, each with a separate schedule. On the whole, most researchers have been quite happy to follow Lenneberg’s idea of an offset coinciding with puberty, even though, as Bialystok and Hakuta point out (1994: 79), some of the most convincing data “show a better fit with the notion of a critical period before age five”.
It must be noted that postulating any clearly defined schedule implies a positive answer to a more fundamental question: does the critical period end abruptly at all? If it does, we have to look for a neurological explanation. If, on the other hand, there is a gradual decline in ability stretching over many years, a vast array of other possible explanations can come into play and the very validity of the critical period hypothesis becomes less than obvious.
5. What is the mechanism of the critical period?
Graylag goslings’ window of opportunity for identifying their mother is known to be hardwired into their brains, as is kittens’ critical period for learning to distinguish vertical lines from horizontal ones. It is therefore hardly surprising that neurological explanations for the linguistic critical period in humans have been pursued with more vigour than any alternatives. Moreover, as Bialystok and Hakuta (1994: 53) note, we may be subconsciously “committed to the superiority of organic explanations over psychological and cultural ones, drawing perhaps an analogy from our acceptance of the hypothesis that diseases are caused by organic factors and not by evil spirits”.
As to why any neurological changes affecting language learning should take place at all, the most obvious and compelling explanation is an evolutionary trade-off benefiting a young organism at the expense of an older one. As Steven Pinker (1995: 294) puts it, “learning a language – as opposed to using a language – is perfectly useful as a one-shot skill”. Energy-consuming language-acquisition circuitry is dismantled once it has been used. “The linguistic clumsiness of tourists and students,” Pinker concludes on a somewhat sombre note, “might be the price we pay for the linguistic genius we display as babies, just as the decrepitude of age is the price we pay for the vigour of youth” (1995: 296).
There seems to be no equally compelling suggestion for what exactly happens to the “greedy” language-acquisition circuitry. In their 1959 publication, Penfield and Roberts (Bialystok, Hakuta 1994: 56) referred to general diminution of cerebral plasticity; this idea was later supported, among others, by Scovel (1988). Lenneberg (1967) blamed the effects of lateralisation with regard to the left-hemisphere dominance in language processing. But these are by no means the only neurological explanations available. After lateralisation was shown to be present in the brain from birth, some researchers argued that more specific localisation of language subfunctions within the dominant hemisphere was the culprit (Singleton 2005: 7). Long hypothesised in 1990 that changes in neurotransmitters might affect second language acquisition (Bialystok, Hakuta 1994: 81). In the early nineties, a maturational process called myelination, in which brain cells become insulated with fatty tissue, was suggested as a possible mechanism; however, as the whole purpose of myelination is to facilitate neural communication, it is hard to see why it should impede language learning (Bialystok, Hakuta 1994: 82). In the mid-nineties, the brain-imaging research carried out by an American team revealed that there were differences in the spatial representation of L1 and L2 in Broca’s area of late bilinguals as opposed to early bilinguals (Singleton 2005: 7). While the late bilinguals showed activation in two separate areas, the early bilinguals appeared to have a single activation area for both languages. Even more recently (2001), a team at the Free University of Bozen found similar differences in brain activation between early bilinguals and late multilinguals (Singleton 2005: 7, 8). Nevertheless, just like every other attempt at a neurological explanation, these findings have been challenged. Friederici, Steinhauer and Pfeifer (2002: 1, 6), for example, claim that their brain-imaging study of adults learning an artificial language showed no significant difference between language processing in native speakers and late learners.
Once we are willing to concede that our language-learning ability is not dismantled overnight but rather suffers a gradual decline, there is no shortage of non-neurological explanations to choose from. In 1975 Krashen, an adherent of Jean Piaget’s cognitive approach to language acquisition, suggested that starting from puberty, language learners can no longer rely on ad hoc solutions – they need a theory; hence the fundamental difference between native speakers and late learners (Singleton 2005: 8). In 1981 Felix executed a marriage of Piaget’s and Chomsky’s ideas claiming that problem-solving cognitive structures emerging at the formal operations stage, which is postulated by Piaget, interfere with Chomsky’s Language Acquisition Device (Singleton 2005: 8). On the whole, it is felt that “adopting the cognitive abilities in a language-learning task” results “in less successful learning than found in children” (Gass, Selinker 2001: 342).
Another group of explanations cites social and psychological factors. Different researchers have suggested such constraints on second language acquisition as adults’ reluctance to take on a new language identity, social distance, and attitudes towards the target language (Bialystok, Hakuta 1994). Sagerdahl et al. (2005: 9) closely link first language acquisition to “enculturation”: “Language is so thoroughly intertwined with how we function spontaneously together that it cannot be learned through planned and explicit instruction.” Language as such, they claim, can only be learned in an unplanned manner, “by living together”; this is why adults and children learn languages in fundamentally different ways.
One of the oldest explanations, put forward back in 1939, probably can also be classified as psychological as it takes a Freudian line. Adults’ inherent narcissism, reinforced by their super-ego, it maintains, leads to an acute “sense of shame” when they try to use a new language and thus inhibits learning (Singleton 2005: 9).
6. A methodological nightmare
In critical period research, the number of hypotheses seems to be in inverse proportion to the amount of reliable hard data available. What is more, any hard evidence that is available tends to be interpreted in diametrically opposite ways. To cite just one example, the results of Johnson and Newport’s well-known 1989 study of syntactic judgement in immigrants with different age of arrival were presented by the authors as strong evidence for a critical period for syntax acquisition; Bialystock and Hakuta confidently dismiss this result on methodological grounds (Bialystock, Hakuta 1994). At the same time, their own study of US census data (Hakuta, Bialystok & Wiley 2003), which purports to demonstrate that the data in question contains no indication of a critical period for second-language learning, came under fire from Stevens (2004) for similar reasons.
The now well-established method of studying critical periods in animals is through deprivation. Animal rights issues aside, a scientist is free to take several goslings, make sure they are not exposed to any moving shapes for the entire duration of the suggested critical period, and see how they fare afterwards. For the obvious ethical reasons, this approach is not possible in studies of human linguistic behaviour, but this is only part of the reason why the critical period research has been anything but straightforward.
There is a feeling that even if such language deprivation experiments were possible, they might at best help clarify the issue of the critical period for first language acquisition. When it comes to second language acquisition, the sheer number of variables one has to keep track of and control for is staggering, and for some variables any application of the experimental method is downright impractical. Supposing we wanted to test whether a language could really only be acquired to a native-like level hand in hand with enculturation: how would we go about “ex-culturating” adults? Or how are we to measure the robustness of someone’s language identity and its correlation with the thickness of their foreign accent? As for the interference of cognitive problem-solving abilities with the LAD, it is hard even to conceive of what could be used as an independent variable should it come to testing this proposition.
“Language is far too complex a system to reveal itself through a single skill, a single experience, or a single test. People, too, are complex…” (Bialystok, Hakuta 1994: 80) Age-related factors can impact on language learning in a myriad of ways; and in some cases we cannot even be sure that these factors are age-related in the first place. Neither is it quite clear what it means to be a better language learner (Gass, Selinker 2001: 335): should we take into account initial and medium-term progress, or is it only ultimate attainment that matters? In the former case, adult beginners have been shown time and again to outperform child beginners (Bialystok, Hakuta 1994; Aitchison 1993; Gass, Selinker 2001); in the latter, how exactly do we decide at which point attainment becomes ultimate?
7. Critical period hypotheses
Do this methodological nightmare and the resulting multitude of opinions mean that the Critical Period Hypothesis is indeed a vague unfalsifiable conjecture, a misapplied zoological metaphor? Does the complexity of human behaviour, to which the concept is extended, make it “difficult, if not impossible, to identify the parameters of sensitive periods with appropriate specificity”? (Thompson 2001, quoted in Singleton 2005: 12)
Singleton is certainly right in saying that if we were to reduce the different versions of the CPH to a single summary (“For some reason, the language acquiring capacity, or some aspect or aspects thereof, is operative only for a maturational period which ends some time between perinatality and birth.” – Singleton 2005: 12), the result would be patently imprecise and unsatisfactory. However, I do not see why the whole issue should be reduced to a single hypothesis at all. The impression I have gathered from the literature is that there are at least several largely independent hypotheses which can be formulated with enough precision for their predictions to be unambiguous and falsifiable. Here is a list of the more obvious hypotheses I have been able to identify:
1. There is a neurologically pre-programmed age limit beyond which a human being, previously unexposed to any language cannot fully acquire one. Obviously, one has to specify the age limit and make clear what one means by “full language acquisition” for genuine scientific scrutiny of this hypothesis to be possible. However, once these values have been filled in, looking for relevant evidence is fairly straightforward, if not necessarily easy.
2. There is a fundamental neurological difference between second-language learning that takes place before or after this hypothetical age limit. Brain-imaging studies and studies of possible differences in syntactic intuitions between native speakers and late learners seem to be two promising ways of tackling this problem.
3. There are near-absolute biological constraints on the ultimate acquisition of morphology and syntax to a native-like level which become operative after a certain age. Again, the age value and the definition of native-like morphosyntactic skills have to be specified.
4. There are near-absolute biological constraints on the ultimate acquisition of phonology to a native-like level which become operative after a certain age. See 3 above.
5. There is a gradual age-related decline in a learner’s ability to fully acquire the morphology and syntax of a foreign language. Once this much has been established, the different social and psychological factors which can potentially affect language learning can be considered. As is almost invariably the case, a complex combination of such factors would be most likely to emerge as the cause of the decline.
6. There is a gradual age-related decline in a learner’s ability to fully acquire the phonological system of a foreign language. See 5 above.
On balance, I believe that there is no reason to dismiss the entire field of critical period research and put the problem on the backburner until some more advanced technology allows us to read the workings of the brain as if it were a television set. A number of scientifically valid critical period-related hypotheses can be formulated; in fact, they have largely been formulated by researchers studying particular aspects of the issue. It is obvious that the notion of critical periods as it is understood in zoology cannot be applied to human language without some serious modifications; perhaps, the term itself should be replaced by something less loaded and potentially misleading. It is also obvious that particularly high standards of methodological precision are needed to study any interaction between language and its neurological and social environment. If these requirements are met, however, critical period-related research can help us make important progress towards the ultimate goal of linguistics – complete understanding of how language works.
REFERENCES
Aitchison, J. 1993. The Articulate Mammal: An Introduction to Psycholinguistics. Routledge. London.
Bialystock, E. Hakuta, K. 1994. In Other Words: The Science and Psychology of Second-Language Acquisition. BasicBooks (A Division of HarperCollins Publishers). New York.
Cattell, R. 2000. Children’s Language: Consensus and Controversy. Cassell. London.
Friederici, A. D. Steinhauer, K. Pfeifer, E. 2002. Brain signatures of artificial language processing: Evidence challenging the critical period hypothesis. (Proceedings of the National Academy of Sciences of the United States of America, 8 January 2002, vol. 99, 529-534). The Max Plank Society eDoc Server. http://edoc.mpg.de/18836 (17 October 2006)
Gass, S. M. Selinker, L. 2001. Second Language Acquisition: An Introductory Course. Lawrence Erlbaum Associates, Inc. Mahwah, N.J.
Hakuta, K., Bialystok, E., Wiley, E. 2003. Critical evidence: A test of the critical-period hypothesis for second-language acquisition. (Psychological Science, 14, 31–38.) ). Blackwell Synergy. http://www.blackwell-synergy.com/links/doi/10.1111/1467-9280.01415 (17 October 2006)
Lenneberg, E. H. 1967. Biological Foundations of Language. John Wiley & Sons, Inc. New York.
Obler, L. K. Gjerlow, K. 1999. Language and the Brain. Cambridge University Press. Cambridge.
Pinker, St. 1995. The Language Instinct. Penguin Books. London.
Scovel, Th. 1988. A Time to Speak: A Psycholinguistic Inquiry into the Critical Period for Human Speech. Newbury House Publishers. Cambridge, Mass.
Segerdahl, P. Fields, W. Savage-Rumbaugh, S. 2005. Kanzi’s Primal Language: The Cultural Initiation of Primates into Language. Palgrave Macmillan. Basinstoke.
Singleton, D. 2005. A Coat of Many Colours (International Review of Applied Linguistics 43 (4), 269-285). The directory of personal home pages for people using the Research Computing Facility at the University of Southern California. http://www-rcf.usc.edu/~ionin/SLAgroup/Ling527papers/Singleton%20Critical%20Periods%20iral.2005.43.4.269.pdf (17 October 2006)
Stevens, G. 2004. Using census data to test the critical-period hypothesis for second-language acquisition. (Psychological Science, 15, 215–216). Blackwell Synergy. http://www.blackwell-synergy.com/doi/pdf/10.1111/j.0956-7976.2004.01503012.x (17 October 2006)
