A concept of 'critical period' for language acquisition
--- Its implication for adult language learning ---
Katsumi NAGAI
Osaka Nougei High School, Evening Course
‚Í‚¶‚ß‚É
`Šú‰pŚę‹łç‚Ě•K—v«‚Ş‹©‚΂ęCŹ¬ŠwŤZ‚ʼnpŚę‹łç‚ŞŽŽŤs‚ł‚ę‚éę•űC'莞§Ť‚ŤZ‚ʼnpŚę‚đŠw‚Ô¬l¶“k‚©‚碂±‚̍΂ŌęŠw‚̕׋‚đ‚µ‚Ä‚ŕ-ł`Ę‚Ĺ‚Í‚Č‚˘‚©Ł‚Ć‚˘‚¤‹^-â‚ŞŹo‚ł‚ę‚éD-{Ź¬_‚Ĺ‚Í‚±‚ę‚ç‚Ě‹c_‚ĚŚ®‚ƂȂ錾ŚęŠwŹK‚Ě—ŐŠEŠú‰Ľŕ‚ɂ‚˘‚Ä‚ĚćŤsŚ¤‹†‚đ'˛Ť¸‚µ‚˝DŚ‹_‚͢—ŐŠEŠú‚Ě`¶ŤÝ‚͉Ľŕ‚Ěć‚đŹo‚Č‚˘Ł‚Ĺ‚ ‚é‚ŞC`ć‚QŚľŚę‚Ě•¶-@Šl“ľ‚Ɖą‰CŠl“ľ‚đŤ¬“Ż‚·‚é‚̂͊댯‚Ȃ悤‚Ĺ‚ ‚éD
1. Teachability and Learnability
Every time adult learners see a child who speaks a foreign language fluently, the learners regret not having started learning the foreign language earlier because the learners' speech necessarily involves a foreign accent. Does their regret have a rationale? Is learning foreign language impossible for adult learners? It is widely believed that the earlier people start learning a second language, the more successful they will be. If there is 'a period during which language can be acquired more easily than at any other time (Richards, Platt and Platt 1992:92),' language learning after that period should be more difficult. Critical Period Hypothesis (CPH henceforth) affects the educational policy-making of foreign language teaching. If the period falls before the age of twelve or thirteen, at which the age Japanese students begin to learn English, the Japanese learners of English are in no way to blame for their foreign accents.
Apart from a question of what the word 'native' means, the CPH provides a good excuse to the people who fail to achieve native-like fluency in foreign languages. The conclusion of this chapter is, unfortunately, that the CPH will remain a hypothesis for ever unless further research clarifies the nature of language acquisition itself. However, the results of experimental studies have two important implications for adult second language learning. One is that children's acquisition of a foreign language is different from that of adults'. The other is that acquisition of pronunciation and grammar is also different because it involves a problem of physiologic aging process. These results do not contradict a widespread belief that adults can learn the grammar of a new language more easily and rapidly than children but that they retain foreign accents.
2. Evidence of the Critical Period Hypothesis
2.1. The Critical Period Hypothesis
Originally the phrase 'critical period' was used for the ethologists' study of species-specific behavior. It is the period when imprinting is observed in certain species such as young birds and rats. For example, geese isolated from their parent birds since the hatching react to and follow the moving object they see first. This kind of behavior takes place only for a short period of time after hatching (Lorenz 1961, quoted in Clark & Clark 1977:520). Another example is a 'crystallization' of chaffinch whose songs are highly intricate. If a young bird does not hear an adult singing within a certain period, the bird will never sing a full song. The CPH posits this kind of a short period for the acquisition of language, which is decided by the learners' behavior.
Flege (1987) lists four characteristics of the imprinting as below.
1. It tends to appear under well-defined developmental conditions.
2. It cannot be forgotten or revised once it has been established.
3. It involves the recognition of species' characteristics rather than individual characteristics.
4. It may be learned long before it is manifested.
He assumes that only the first characteristic above is applicable to the human language acquisition. Is it applicable to second language learners too? A pioneering work by Lenneberg (1967) explores the CPH and says:
Between the ages of two and three years language emerges by an interaction of maturation and self-programmed learning. Between the ages of three and the early teens the possibility for primary language acquisition constitutes to be good; the individual appears to be most sensitive to stimuli at this time and to preserve some innate flexibility for the organization of brain functions to carry out the compex integration of sub-processes necessary for the smooth elaboration of speech and language. After puberty, the ability for self-organization and adjustment to the physiological demands of verbal behavior quickly declines. The brain behaves as if it had become set in its ways and primary, basic skills not acquired by that time usually remain deficient for life (Lenneberg 1967:158).
His book has been widely accepted because of its clear explanation for difficulties in second language learning after puberty. It explains well our experience of seeing small children speak foreign languages quite naturally, and the experience of hearing unnatural foreign accents in the speech of adult learners.
2.2. Studies of a Human Brain and its Lateralization
If a language develops with the increase of the brain cell connections, the development of language and brain should bear some relation to each other. Physiological functions of neuronal cells, circuits, or connections themselves have not been identified clearly so far, but some specified fields of linguistic functions have been reported. Geschwind (1980:62-63) claims:
Although the content, that is, the specific language learned, is completely determined by the environment, the capacity to acquire language is biologically determined. Indeed, one should not speak about a contrast between language and general cognitive ability.
Earlier studies show that the brain seems to have special-purpose computers for limited functions, and there is at present no evidence of any all-purpose computer for any general cognition. If a language is acquired in parallel with the development of the human brain as children grow, it is reasonable to postulate some language function in the brain. Consequently, the critical period for language learning is considered to be the biologically determined period in which the brain keeps its plasticity for acquisition of any language.
At first it was expected that the function of human language acquisition was clarified by exploring a special structure, which all other animal brains lack. The brain of human beings consists of a left hemisphere and a right hemisphere, and different functions are said to develop gradually in different parts of the brain as children grow older. The parts of brain which control a language are placed in the left hemisphere after its lateralization. One of the parts is called the Broca's area and the other part is called the Wernicke's area. The study of aphasia has revealed that the Broca's area controls spoken language. The Wernicke's area is, on the other hand, said to be the center of language understanding.
Lenneberg posits that the development of language is the result of brain maturation. Although both hemispheres of the brain are equal at birth, the function of language gradually settles in the dominant left hemisphere of the brain after biological maturation or the critical period. That is, the critical period for language learning has been considered to agree with the period of the lateralization of brain. Research on people who have suffered brain damage also supplies the evidence for the lateralization of brains. Lenneberg viewed clinical data and claimed that children younger than nine years old had a higher incidence of right hemisphere lesions causing aphasia than adults (Villiers & Villiers 1978:211). Another support of lateralization is an experimental report of speed and accuracy in language acquisition by Kimura (1973). He examined speech sounds heard in each ear. His result is that speech sound through the right ear is processed more quickly than through the left ear because the left hemisphere is connected to a right ear.
A brain is said to lose its plasticity after the lateralization, and some case studies are reported on the impaired brains before and after the critical period. Adults who have suffered brain damage in their left hemisphere fail to recover their language if they don't recover in five months, but children show an ability to recover over a longer period, and have sometimes made a full recovery if they were very young at the time of damage (Crystal 1987:263). Even total removal of the left hemisphere did not preclude children's reacquisition of language. Lenneberg's argument is based on this period of lateralization, of which completion means the end of the critical period.
2.3. Studies of Feral Children
Supports most frequently quoted for the CPH are the case studies of the children who had been isolated from language and who tried to acquire the language before and after their critical period. Unlike the cases of Amala and Kamala, who were said to be reared by wolves and found in 1920, studies of teaching a language are reported in the following three cases.
The first case was a deaf mute child named Isabelle, who was found at the age of six and half. She spent alone in a darkened room before being found, but she succeeded in her language learning because she was at the age of six and half. Brown (1958: 192, cited in Aitchison 1989:85) recorded:
Isabelle passed through the usual stage of linguistic development at a greatly accelerated rate. She covered in two years the learning that ordinarily occupies six years. By the age of eight and one half Isabelle was not easily distinguishable from ordinary children of her age.
It is reasonable to consider that she was able to acquire her language because she started learning before the critical period came to an end.
The second case was Genie, who was found at the age of about fourteen (Curtiss, Fromkin, Krashen, Rigler, and Rigler 1974). Because she started learning a language after the critical period, her progress was slower than other children. For example, her two-word stage, at which every child goes though uttering two words at a time like 'Want milk' and 'Mummy play,' lasted much longer. Genie used this type of primitive form and its negation such as 'No want milk' for a longer period. Her ability to learn vocabulary was superior to other children. However, her grammatical development was much slower and unsuccessful, because her critical period had passed already. Since she started learning a language after she was already pubescent, Genie had to take quite a long time to acquire a language.
The third case was Chelsea, who started to learn language in her early thirties (Curtiss 1988). She showed poor grammatical ability like Genie, but her vocabulary was better. It was recorded that her syntax created sentences such as 'the woman is bus the going' and 'banana the eat.'
All these cases of children reared in isolated environments reveal the difficulties of learning a language after the critical period.
2.4. Criticism of the Critical Period Hypothesis
2.4.1. Critical Period Hypothesis reexamined
As medical science has progressed, some points in Lenneberg (1967) are under criticism. Firstly, his claim that the CPH could be supported by the study of Down's syndrome cases is attacked. According to his argument, the development of children with Down fs syndrome is so slow that they pass their critical period for language learning. However, a recent survey disputes his claim because children with Down's syndrome have a build-in endpoint to their ability (Gleitman 1984).
Secondly, the theory of brain's lateralization at the age of two, with the critical period set by Lenneberg (1967), is also criticized. He claimed that children before their critical period were less severely impaired by brain damage. However, Krashen (1973) reexamined the data used in Lenneberg (1967) and found all the cases of complete recovery from aphasia were under the age of five. Surprisingly, the number of cases of recovery at the age of more than five or over was nearly the same number as the adults'. Kinsbourne (1975) pointed out the difficulty in deciding whether only half of the hemisphere was injured or not. Another piece of counter evidence is seen in MacKain et. al. (1983). Their experiment with babies of six months or less shows that lateralization begins much earlier than two years old. If so, lateralization cannot be the evidence of the critical period. Reports of dichotic listening, experiments in which different stimuli are presented simultaneously to the two ears, also show that language functions are lateralized much earlier than the critical period. Regrettably, no conclusive evidence for right ear advantage based on the lateralization has not been reported.
Thirdly, the case studies of three linguistically isolated children have some problems. Because the early research in Isabelle's case was not written by a specialist in language (Aitchson 1989:86), her speech may have been less sufficient than reported. There is also a report that Genie's left hemisphere was atrophied because of the brain damage. This implies that she used only the right hemisphere of the brain, the part which is said to have little function in language processing. Curtiss, et. al. (1974) writes:
The result of the dichotic tests using environmental sounds also show a left-ear advantage, but only to a degree found in normal subjects. This 'normal' result shows that Genie is not simply one of those rare individuals with reversed dominance, but instead is one in whom all auditory processing currently appears to be taking place in the right hemisphere (Curtiss, Fromkin, Krashen, Rigler, and Rigler 1974:542).
Isabelle was deaf and Chelsea had hearing problems too (Aitchson 1989:87). A difference between deaf and normal children should have been taken into consideration in the case studies. Since no further case of feral children will be reported, the data have been reanalyzed. All of these criticisms of the past studies show how difficult it is to connect brain functions and language acquisition.
Moreover, the different methods for the research on the CPH have also resulted in varieties of inconsistencies. For example, native speakers were asked to rate the performance or naturalness of the utterances of learners in one study (Scovel 1981, cited in Ellis 1984:485), whereas grammatical judgments made by learners of different ages were examined in another study (Coppetiers 1987). Birdsong (1992) designed a grammaticality judgment test for twenty native English speakers who spoke French with native-like fluency in order to examine Long's question (1990:281) of 'whether the very best learners actually have native-like competence.'
2.4.2. The Sensitive Period and the Multiple Periods
The CPH is under criticism, with a large quantity of counter evidence. Nevertheless, no other hypothesis will explain the truth better that language learning by matured students and the brain damaged patients' recovery of language become more difficult with aging.
Because it is difficult to fix the exact span of years during which language learning can take place naturally, some researchers have presented a revised version of the CPH. They use the term 'sensitive period,' rather than 'critical period,' for second language acquisition. The distinction between the CPH and the sensitive period hypothesis is whether acquisition is 'possible only within the definite span of age' or 'easier within the period.' Oyama (1979:88) says that sensitive periods are preceded and followed by less responsive periods. Seliger's proposal (1978) is that there may be multiple critical or sensitive periods for different aspects of language. The period 'during which a native accent is easily acquirable' appears to end earlier than the period governing the acquisition of a native grammar.
2.5. Implications of the CPH for Adult Second Language Learning
2.5.1. The Implication from the Variety of Results
What implications does the argument of the CPH for adult second language learning have? As there are such divergent results on the CPH, the classification by Ellis (1994: 485) is useful to find a point of departure. It is especially so if the priority is given to phonological development, which necessarily involves articulatory motor control.
1. What effect does age have on the rate of L2 learning?
2. What effect does age have on learners' levels of L2 achievement of native-speaker levels of proficiency?
3. What effect does age have on learners' levels of L2 achievement (in those learners who do not reach native-speaker proficiency)?
4. What effect does age have on the processes of L2 learning?
Ellis (1994:493) says, 'if adults substitute inductive cognitive learning strategies for the language acquisition device used by children, differences in the process of acquisition might be expected to occur.' Recent studies of the second language acquisition, however, fail to report consistent results to support his view. Singleton (1989:137) writes:
Concerning the hypothesis that those who begin learning a second language in childhood in the long run generally achieve higher levels of proficiency than those who begin in later life, one can say that there is some good supportive evidence and that there is no actual counter evidence.
Does this claim have the rationale? The following are examples of the studies concerning the CPH.
Harley (1986) shows greater overall control of verb system by older group who studied French in Canada. The result is not supportive of the CPH because, as Ellis (1994:489) writes, formal learning environments do not provide learners with the amount of exposure needed for the emergence of young learners' age advantage. Cummins and Nakajima (1987) conducted a survey of 273 Japanese children in Canada. They found that the older the students were on arrival in Canada, the more they had reading skills and writing skills in English. They claimed that the older learners benefited from prior literacy experience in Japanese. In Neufeld (1978), twenty adult native speakers of English were asked to do an imitation test after they had eighteen hours of intensive instruction in Chinese and Japanese. Their naturalness was judged by native speakers on a five-point scale. The result suggests that adults can achieve native-like ability in pronunciation. His experiment was criticized later on the grounds that the subjects in the study were the elites who produced unnatural rehearsed data (Long 1990). Overall, as in Krashen, Long and Scarcella's argument (1979), adults learn faster, especially so in the case of formal learning of grammar than pronunciation.
The most important reason for the divergent results of the past research is, excepting the differences in the methods themselves, disregard of the natural learning environment. Child learners, playing with other native young children, can enjoy their language interaction. They can speak loudly, they can ask anytime and anywhere, and they do not have to worry about making mistakes. Adult learners, on the other hand, have difficulties in participating in natural language interaction when they learn a foreign language. This affective factor plays an important role in artificial environments. In a classroom, however, adult learners can use mature cognitive skills to compensate for their declining memory and motor control. Adult learners' inductive and explicative processing abilities are in optimal condition (Steinburg 1982:179). This enables them to do better than children in classrooms. Ellis (1994:491) writes:
Adult learners have an initial advantage where rate of learning is concerned, particularly in grammar. They will eventually be overtaken by child learners who receive enough exposure to the L2. This is less likely to happen in instructional than in naturalistic settings because the critical amount of exposure is usually not available in the former.
Research on the CPH suggests that adult learners' developed cognitive skills help them learn a foreign language. This results in an initial advantage over children in the artificial environment such as in a classroom though it does not guarantee their native-level achievement.
While some adult learners of foreign languages can speak foreign languages quite naturally even if they started learning after puberty, most learners fail to reach native-speaker's level. Should adult learners give up their aim at achievement of native-speaker levels? Can they outperform the learners who started learning a second language before puberty? Four studies are considered below.
Coppieters (1987) examined twenty-one adult French speakers who had begun learning French as a second language. When their grammatical performance was compared with native speakers, it was impossible to distinguish the non-native speakers by the number of mistakes and inappropriate wordings. The judgment of grammaticality was, however, different between native and non-native speakers. Coppieters claimed that the divergence between the two groups was more marked in functional distinctions. Birdsong (1992), on the other hand, concludes contrastively. He gave a grammatical test to twenty learners of French, who spoke like native speakers but whose native language was English. In comparison with native speakers of French, no significant difference was found within the theoretical frame of 'Universal Grammar.' His study reassures the existence of competent learners of foreign languages who started learning a second language after puberty. The third example is in Thompson's study (1991). He reported that the learners who arrived in the United States before they became ten years old succeeded in learning more natural English than those who arrived at later age. The two subjects who came from Russia at the age of four failed to achieve native-like pronunciation because of, Thompson claims, their high proficiency in Russian. This hints at the dominance of the first language. The fourth is Scovel's research (1981). He divided the subjects into four groups, i.e. adult native, child native, adult non-native, and child non-native speakers, and asked them to distinguish differences of spoken and written accents. The result showed that even the most advanced non-native speakers could not achieve native-like accuracy (73 percent, compared with 95 percent by the native speakers). Studies by Coppieter and Scovel conclude that even very advanced learners can not perform perfectly.
It is unclear whether the achievement of native-speaker level fluency is possible for adult learners because all research cited above produce mixed results. Children are more likely to achieve higher levels in both pronunciation and grammar than adults, but some adults who started learning their second language after puberty can achieve native-like levels. This inconclusiveness is a warning to teachers and parents of language learners. However early they start teaching and whatever a desirable environment they have, these conditions can not guarantee the native-like performance. It should be also noted that it is impossible to decide whether learners of foreign languages have a real 'native-like' competence without defining the word 'native.'
2.5.2. The Different Acquisition of Pronunciation and Grammar
One important thing obtained from reviewing the past studies above is the fact that the critical period of pronunciation seems to be different from that of grammar. Studies of anarthric children have also revealed that their perception of phonological contrasts is normal. Lenneberg (1967) and Slobin (1971:60) illustrate that anarthric people can acquire a language. For example, they report the case of a speechless child who was unable to articulate speech. The boy's articulation was abnormal, but he was able to understand complete English sounds and syntax. Genie's case (Curtiss, Fromkin, Krashen, Rigler, and Rigler 1974:534) is not simple because she had to learn to control her articulatory organs. The researchers write: Genie can pronounce many sound sequences in imitation which she does not use in spontaneous speech. It is clear that her output is more constrained by her own phonological 'realization' rules than by her inability to articulate the sounds and sound sequences of English.
This shows that, even in an abnormal case of language acquisition, children's phonological competence and their phonetic performance need to be differentiated carefully. One comprehensible example frequently quoted is in Miller (1963). He reports:
Recently a three year old child told me her name was Litha. I answered: