AGE EFFECTS IN SECOND LANGUAGE ACQUISITION: OVERVIEW
INTRODUCTION OF THE PROBLEM
Age has been regarded as an important factor in the ways in which language learners differ, and a vast amount of research has been conducted regarding age effects on second language acquisition（e.g. Birdsong, 1992; DeKeyser, 2000; Oyama, 1976; Patkowski, 1980）. It is generally believed that children enjoy an advantage over adults in learning languages because of their ‘plasticity’. According to Chomsky（1957）, humans are equipped with a ‘language acquisition device’, which enables them to acquire the language in a way that goes beyond simple habitual formation. The ‘universal grammar’ proposed by Chomsky later on（1966） is thought to be an innate system of language acquisition, the socalled language acquisition device. Although Chomsky has not mentioned the possibility of applying this theoretical device in the brain to the acquisition of second languages, ‘grammaticality judgment tests’, the purpose of which is to measure learners’ universal grammar, have been widely used for second language acquisition research（e.g. Johnson & Newport, 1989. These ‘grammaticality judgment tests’ consist of morphosyntactic items, implying that the ‘universal grammar’ is really about how learners organize the target language’s morphosyntactic system.
On the other hand, Lenneberg（1967）hypothesized that humans’ latent language structure, i.e. the cognitive structure for automatic language acquisition, might stop functioning when the human brain matures, or at the time of lateralization of the human brain, which possibly occurs around puberty. He established the critical period hypothesis, which was originally proposed by Penfield & Roberts（1959, and explained the difficulty of acquiring our ﬁrst language after puberty, based on neuropsychological factors（Lenneberg, 1967. In the area of second language acquisition research, the critical period hypothesis has been taken into consideration in age-related studies. There is believed to be a period up to a certain age during which learners can acquire a second language easily and achieve native-speaker-like competence. The sensitive period hypothesis, which is used by Patkowski（1980, has been sometimes used as an alternative term to refer to the critical period hypothesis in second language acquisition, and has often been freely substituted in second language research literature. However, the critical period hypothesis has been predominantly used in first language acquisition, whilst the sensitive period hypothesis has been generally restricted to second language acquisition.
Does age effects the learning second language process?
We have observed that learners who start early in life to expose themselves to their second language are more likely to attain a native or native-like accent than older starters. Oyama （1976）examined 60 male learners who had immigrated to the United States. Their ages ranged from 6 to 20 years old and they had lived there for between 5 and 18 years. Two adult native speakers judged the ‘native-ness’ of the learners’ accents during a reading-aloud task and during free speech. The results showed a signiﬁcant negative correlation in ‘age of arrival and acquisition’, which meant that the younger their age of arrival was, the more authentic the accent they acquired. For instance, the youngest arrivals were rated the same as native speakers. However, no signiﬁcant relationship was found between the length of stay and their accent. Other studies that examined the effect s of age on pronunciation（e.g. Tahta, Wood & Loewenthal, 1981）also indicated that an earlier age of arrival or acquisition leads to better pronunciation.
Similar results have been provided from studies in morphosyntax/grammar, but in their studies the cutting-off age for the critical/sensitive period is later or older than the studies on pronunciation. Patkowski（1980）invest igated 67 immigrants to the United States, ﬁnding that learners who had entered the United States before the age of 15 were rated as more proﬁcient in grammar than learners who had entered after the age of 15. There was also a significant difference in the distribution rate of scores based on a ﬁve-point scale for the two groups. The range of adult group scores was smaller than the range of child group scores. In addition, Patkowski examined the effects of the length of the stay in the United States, the amount of informal exposure to English and the amount of formal instruction. Neither the length of the stay nor the amount of formal instruction provided a signiﬁcant effect but the amount of informal exposure did have a significant effect, though this was much less signiﬁcant than the age factor.
In a similar line to Patkowski（1980）, Johnson & Newport （1989）investigated 46 native Koreans and Chinese who had immigrated in the United States between the ages of 3 and 39, using an aural grammaticality judgment test. Half of them arrived there before the age of 15 and the other half arrived after the age of 17. The participants were asked to judge the grammaticality of 276 spoken sentences. The results indicated a negative correlation between age at arrival and judgment scores, which was – 0.77, meaning that the later the learner arrived, the lower the score they got. However, one difference from Patkowski’s study was that the scores of the younger group varied less than those of the adult group. Also, neither the number of years of exposure to English nor the amount of classroom instruction was related to the grammaticality judgment scores.
Johnson（1992）followed up on the study by Johnson & Newport（1989）by using the same participants in the earlier study a year later with written tests, working on the belief that written test materials eliminated extragrammatical properties that were present in the auditory materials. The results showed a negative correlation（r = – 0.54）between age of arrival and performance, and suggested that the grammatical knowledge of young learners is native or near-native whereas that attained by older learners is ill-formed or incomplete. Thus, the critical period effects could be found in a test of grammar with a minimum number of extragrammatical properties. This shows the robustness of critical period effects in second language acquisition
Whether critical/sensitive period hypothesis exists, age clearly should be regarded as an important factor that influences the possibility of attaining native-like proficiency in a second language, though there are some differences in the learning difficulties involved because of the similarities and differences between the ﬁrst languages and second languages and because of given contexts such as whether or not the learners reside in the countries where their second languages are spoken（Birdsong, 2007）. However, the empirical studies about age as an important factor in second language acquisition, described in the previous section, have provided different positions towards the critical/sensitive period hypothesis. The first five studies（Oyama, 1976; Tahta, et al., 1981; Patkowski, 1980; Johnson & Newport, 1989; Johnson, 1992） support the hypothesis, that is, second language learners will not be able to attain a level of native-like proﬁciency if the age of arrival or acquisition is after the critical/sensitive period. The first two studies used data about phonology/pronunciation and the other three studies were based on the results of grammaticality judgment tests measuring the level of morphosyntax/grammar. Thus, the adult participants in these studies may have been in the fossilized phase of development in the areas of phonology and grammar. Regarding the critical ages for acquisition, according to several researchers（e.g. Ellis, 1994; Long, 1990）acquiring native-like pronunciation is possible until the age of 6 – the ﬁnal age for arrival and acquisition. On the other hand, native-like grammatical/morphosyntactical competence should be possible up to the age of 15（e.g. Patkowski, 1980）. As Selinger（1978） proposes, there may be multiple critical/sensitive periods for different aspects of language. The period during which a native accent is easily acquirable appears to end sooner than the period governing the acquisition of a language’s grammar. In other words, the biological sensory acuity for attaining native-like pronunciation terminates much earlier than the cognitive plasticity that manages grammaticality judgment. Pinker（1994） makes the following note. Acquisition of a normal language（phonology）is guaranteed for children up to the age of six, is steadily compromised from then until shortly after puberty, and is rare thereafter. Maturational changes in the brain, such as the decline in metabolic rate and the number of neurons during early school-age years, and the bottoming out of the number of synapses and metabolic rate around puberty, are plausible causes.（p. 293）
On the other hand, the most recent neurocognitive evidence has indicated the mechanism that manages language in the brain’s system. Ullman (2007) argues as follows. In first language, lexical knowledge depends on the declarative memory brain system, which underlies semantic and episodic knowledge, and is rooted in temporal-lobe structures. Grammar in ﬁrst language relies rather on the procedural memory system, which subserves motor and cognitive skills, and is rooted in frontal/basalganglia circuits. In contrast, evidence suggests that in later-learned second language, learners initially depend largely on declarative memory, not only for lexical knowledge, but also for the use of complex forms. However, with increasing experience second language learners show procedural learning of grammatical rules, becoming first language-like. Importantly, because the behavioral, computational, anatomical and physiological bases of the two memory systems are reasonably well-understood, including the nature of forgetting of knowledge and skills in these systems, we can make relatively speciﬁc predictions about language, including with respect to language attrition.（p. 9）
Thus, second language learners are unable to acquire the target language as long as they use the declarative brain memory system for its grammatical rules. As Ullman（2007）points out, through experience, second language learners come to make use of the procedural memory system. Neurocognitive researchers have presented these findings as reliable through the use of advanced technology, which makes them persuasive. Given that ﬁrst language grammar is dealt with in this procedural memory system, the so-called universal grammar (morphosyntax in Practice) or language acquisition device presumably may refer to the process of using the procedural memory system for grammar or language rules. If so, with the possible exception of getting a native-like accent, even adult learners could attain native-like proﬁciency in their target language if they practice it enough to make the language behavior their automatic routine – like riding a bicycle, which also uses the procedural memory system – and to make the procedural memory system active in utilizing the second language’s mophosyntax/grammar. The maxim that practice makes perfect may hold true for acquiring a second language. In the case of child learners, or learners before the age of 15, the procedural memory system rather than the declarative memory system is more likely to be used for second language grammar. Possibly a lack of plasticity in the brain’s system may lead to difficulty in acquiring second languages when we are older. Regarding the subtle distinction between a ‘critical’ and a ‘sensitive’ period, the question is whether completely successful acquisition is deemed to be only possible within a given span of a learner’s life（critical）, or whether acquisition is just easier within this period (sensitive). Therefore, the sensitive period hypothesis seems to be more, appropriate for second language acquisition, though the ‘critical period hypothesis’ has been predominantly used.
The second category of studies with mixed results consists of Burstall（1975）, Harley（1986）and Riney（1990）. Burstall （1975）showed that the late starters excelled in writing, reading, and speaking while the early starters were better only in listening. As Ellis（1994）posits earlier in this article, younger learners are better in sensory acuity, which led to the better listening skills described by Burstall learners were better in all other areas – writing, reading and speaking – may refute the critical/sensitive period hypothesis. It holds true, indeed, that the older learners, secondary school students in this study, outperformed the younger learners because of their advanced cognition and more mature social positioning. However, in the previous literature（e.g. Ellis, 1994）, the younger group was said to overtake the older group.
In Burstall’s study（1975）, at the age of 16, the older group still outperformed the younger one. As one explanation, it was assumed that 16 years old was not yet the end of the best period for acquisition. The other explanation was that the older group of learners could have practised until they reached the stage of using the procedural memory system, which enabled them to use their second language automatically like their ﬁrst language（c.f. Ullman, 2007）. Also, various individual differences, not limited to age factors, played a stronger role in their performances in second language learning. His study shows that age is less important and that the more sophisticated cognitive or possibly academic skills they had in their first language played a more meaningful role in their second language acquisition, except in the area of listening, which may be biological and less inﬂuenced by external factors. Harley’s study（1986）is a robust example showing that the younger learners were able to perform better in the long run, which endorses the validity of the critical/ sensitive period hypothesis. Riney’s investigation（1990） showed condi t ional resul ts control led by phonological environments. Epenthesis, the insertion of schwa sounds, is a well-known phenomenon among Japanese learners of English as well as an example of negative transfer. Possible fossilization caused by having less ﬂexible physiological natures may have rendered the adult learners unable to fix the epenthesis, even after they had exposed themselves to the correct language environment.
The last category of studies consisted of six studies refuting or at least not complying with the critical/sensitive period hypothesis, though the last study by Harley & Hart（1997） suggested different cognitive processes among early and late learners. The ﬁrst three example studies（Bailey, et al., 1974; Fathman, 1975; Harley, 1986）dealt with morphology, a part of so-called grammar; however, they did not present the features of the critical/sensitive period hypothesis, which may question the validity of the hypothesis. The fourth study, by Cummins & Nakajima（1987）, gave clear results counter to the critical/ sensitive period hypothesis because the older learners provided better results in writing and reading tests. The interdependent principle（Cummins, 1981）, which emphasizes the importance of academic skills in first languages, may well support the starting age for learning English in Japan. The reality, however, is that both reading and writing are not directly related to oral communication. In other words, linguistic behaviors such as writing and reading can be classified only as a school subject, not as a part of language acquisition. Probably that is why they failed to comply with the critical/sensitive period hypothesis. On the other hand, the fifth study by Ioup, et al.（1994） investigated an exceptional older learner who succeeded in acquiring a second language and its ﬁndings should encourage adult language learners to try to gain native-like proﬁciency. Various kinds of interpretations have been provided to account for the existence of a critical/sensitive period. For instance, Muhlhauser（1986）proposes that ‘adults and children appear to behave very much in the same manner’, which indicates that ‘activation of certain linguistic developments is dependent on the presence of specific environmental factors, rather than on different cognitive abilities of children and adults’ after an extensive study of the developmental stages of pidgin languages and their similarities to language acquisition（1986, p. 265-266）. Long（1990）, on the other hand, concludes that a neurological explanation is best and proposes a ‘mental muscle model’, where the language-speciﬁc faculty remains intact throughout our lives, but access to it is impeded to varying degrees and impeded progressively with age, unless the faculty is used and so kept plastic. Such a view is compatible with studies of exceptional language learners, which demonstrate that some adult learners are capable of achieving native-speaker levels of competence, as seen in the study by Ioup, et al.（1994）. As Birdsong（1992） points out, the critical/sensitive period hypothesis may have to be reexamined if many such learners are found.
The criteria for the most appropriate age to acquire a language seem to be based on phonology（pronunciation）and morphosyntax（grammar）. Previous age-related studies have claimed that the process of acquiring a second language grammar（morphosyntax）is not substantially affected by age, but that of acquiring pronunciation（phonology）may be. The critical period hypothesis that originated from first language acquisition (Lenneberg, 1967; Penfield & Roberts, 1959) is based on neuropsychological factors, and the most important of these is brain maturation. It is widely known that the cognitive structures that allow for automatic language acquisition in a child deteriorate as the human brain matures. In second language acquisition, if a critical/sensitive period hypothesis does exist, adult learners or learners starting to learn after a certain age（puberty for instance）may experience fossilization in phonology and/or morphosyntax regardless of their efforts, due to neurological/physiological factors. All those who possess a ﬁrst language are certainly capable of acquiring some degree of a second language; however, second language acquisition in a mature human is not as successful as ﬁrst language acquisition in many cases. Although some researchers (e.g. Bley-Vroman, 1988) have argued that older learners no longer have access to their innate language acquisition device, consisting of the principles of universal grammar（Chomsky, 1981）and language-specific learning procedures, it has been found to be possible for adult learners to activate such a device by using the procedural memory system (Ullman, 2007) instead of using the declarative memory system, by following the innate grammatical structure while using the language, and by thorough practice until the structure is internalized in the learners’ minds and becomes automatic in their behavior. Ullman（2001）suggests that ‘an increasing amount of experience（i.e. practice）with a second language should lead to better learning of grammatical rules in procedural memory, which in turn should result in higher proficiency in the language’（p.118）. Even in adult language learning, which has usually been achieved through ﬁrst language knowledge, so-called universal grammar may be accessible to adult second language learners, but their second languages are eventually acquired only if they are encouraged to use the procedural memory system instead of the declarative memory system.
Bailey, N., Madden, C., & Krashen, S. (1974). Is there a ‘natural sequence’ in adult second language learning? Language Learning, 21, 235-243.
Bialystok, F. & Hakuta, K.(1999). In other words: The science and psychology of second-language acquisition. NewYork: Basic Books.
Birdsong, D. (1992). Ultimate attainment in second language acquisition. Language Learning, 68, 706-755.
Birdsong, D. (2006). Age and second language acquisition and processing: A select ive overview. Language Learning, 56, 9-49