Is There A Critical Period For Native-like Pronunciation Of A Second Language?

In a world that grows smaller every day, with instant connection from one side of the globe to the other, it becomes more and more important to be able to communicate with ease. This often means that, to have an advantage, it is necessary to know multiple languages, and while some are fortunate enough to be brought up bilingually, with parents speaking two languages to the child, many need to learn additional languages through education systems. To facilitate this, it is imperative to understand how language, and in particular, how second language acquisition (SLA) works. It was only in the 1940s that SLA became a formal field of research, so the amount of work carried out in the area is relatively small when compared to other topics, such as studies into grammar or writing systems. As such, while there are several commonly held theories and schools of thought within SLA, none have been shown to be entirely and undeniably true.
One idea which gained much popularity from the 1950s to 1970s was the concept that the ability to learn language is hard-wired into us; that we have a “language acquisition device” in our brains that controls our linguistic ability. Following on from this came the notion that there is a critical period (CP) during which a child is sensitive to learning languages, and that after this, it is supposedly exceedingly difficult, or even impossible, to ever truly acquire a new language. In particular, it was thought that “native-like” pronunciation for a second language was impossible for anyone who was older than allowed by the CP.
I intend to examine several aspects of this concept in order to understand why it has been widely accepted, and whether there is enough evidence from SLA to truly support it. This will include looking at the critical period hypothesis itself, what it exactly means, and what hard evidence there is to suggest it might be true; questioning the validity of the idea of native-like pronunciation, and how it is defined; and how other approaches to language acquisition differ from the CP hypothesis. Through these points, I shall attempt to determine the viability of the proposed concept of there being a defined critical period for native-like pronunciation of a second language.

What, precisely, is meant by “critical period”, in relation to linguistics? The term and hypothesis were popularised by Eric Lenneberg (1967), which he explained as being the time during which a language much be acquired. According to him, if a child reaches the end of the CP, at puberty, without having fully mastered the language, they will never be able do so, although it is still possible to learn and become competent in it (Lenneberg, 1967, p. 180). His work took evidence from neurology, and stated that the offset of the CP was coincident with the onset of puberty, which is when the brain assigns each of the left and right hemispheres specialised tasks, one of which is both spoken and written language in the left side of the brain. Based on language acquisition requiring neuroplasticity, it was concluded that “for the purposes of learning languages, the human brain becomes progressively stiff and rigid after the age of nine.” (Penfield & Roberts, 1959, p. 236).
This concept was supported by evidence showing children being able to completely transfer the language functions to the right hemisphere following injuries which left the left damaged. For adults, however, this transfer following injury has a much lower success rate, as their brains have lost the plasticity and ability to adapt to large changes needed to do this. There was also evidence in the cases of feral children, who had not been exposed to language until relatively late in their lives. For example, taking the case of Genie, a thirteen-year-old girl who had been kept isolated and was abused from birth by her father. Upon discovery, she seemed the ideal opportunity for study, and was taken in and educated in language. However, her progress was not strong, and once research interest had waned, the skills she had attained regressed. Compare this to the case of Isabelle, who spent six and a half years without any linguistic stimulus due to being confined to a room with a deaf and mute mother. Eighteen months into her training and she was able to produce complex sentence structures (1998, Tartter, p. 113), and was able to catch up to normal levels for a child of her age quickly. These two cases seem to show good support for the CP hypothesis.
However, these two areas of evidence are not sufficient to conclude that the CP is irrefutable fact. For a start, the studies involved were not done on a large enough scale to say for certain that the findings are true for a larger population, in particular those done on the so-called feral children. In addition to that, it is very likely that there may have been other, possibly psychological, factors involved in any impediment in their learning, and as Lenneberg (1967, p. 142) himself said, “Life in dark closets, wolves’ dens, forest or sadistic parents’ backyards is not conducive to good health or normal development.” As for the neurological studies, it has been shown that even though children can shift the language centre to the right hemisphere of the brain, they are still likely to be left with permanent speech difficulties, and it is also likely that they will suffer as much of an impairment as an adult following a similar brain injury, although it may take some time for the full extent of the injury to reveal itself, due the brain still maturing in a child. There is also the fact that studying children with brain damage may not be entirely relevant as evidence for a CP in normal language development, since they are exceptional cases and not necessarily representative of the overall population.
Another point against the idea of a CP is that it is very difficult to determine when exactly it begins and ends. As Singleton (2005, p. 273) shows, studies between 1959 to 2003 that claim onset between the sixth month of foetal life and two years, and offset anywhere from shortly after birth to the sixteenth year of life, although around puberty is the more common theory. This lack of conclusivity exemplifies the difficulty involved in the task, and also suggests that the hypothesis itself may not but entirely fundamentally sound. As a result, the CP hypothesis is no longer as widely accepted a theory as it once was, although elements of it still seem to stand up today.
In terms of SLA, the CP hypothesis is even less widely accepted in its express form of a window of time after which a language cannot be mastered. Despite the fact that children appear to be more likely to attain a high level of proficiency in their second or additional language (L2) than older learners, they usually progress more slowly in the early stages of learning. Singleton (1995, p. 3) states that “younger = better in the long run”, but does not claim that this is true in all cases, and in fact points out that five per cent of adult bilinguals master their L2 despite beginning learning well after any supposed CP should have ended. From this we can draw that it plausibly not the biological maturity which causes difficulty in older learners. Other factors in individuals’ lives may lead to reduced motivation and less time given to learning, or other mental or psychological elements may have an effect on a person’s ability to learn.
However, the CP hypothesis cannot be written off entirely. According to
Oyama (1976), adult L2 learners may develop perfect grammar, but they almost never entirely mask an identifiable foreign accent, which can cause a speaker to stand out as not having completely mastered the L2 to the “native-like” standard.

How is a person’s level of proficiency in a language judged to be native-like, and is it fair to do so? Scovel (1988, p. 65) said that adults would “never pass themselves off as native speakers phonologically.” He (p. 131) argued that pronunciation was the only aspect of language that had a CP, as it is “the only [one] that has a neuromuscular basis,” and as such would be the sole area affected by the maturing of the brain and reduction in neuroplasticity. However, he (p. 181) did allow for some “superexceptional” cases, saying that perhaps one in every thousand late-learners would not be bound by a CP, and so would be able to achieve high-level fluency in their L2, including accentuation. However, is it fair to judge success in learning or proficiency in a language on whether a speaker has native-like pronunciation? There are several issues with this practice, even ignoring the fact that what qualifies as “native-like” would have to be subjective, based on the native who was judging. To begin, accents, as such, are often under-emphasised when a language is being taught to an L2 learner, as the focus is usually on the grammar, vocabulary and syntax in academic learning places like schools, as the written aspect of a language is easier to judge than spoken. However, this has seen improvement in recent years (for example, the Irish Leaving Certificate Oral exam’s percentage being increased from 25% to 40%), with educators realising there needs to be more attention given to spoken language. Another issue is how accurate this kind of judgement can possibly be. Would a native speaker of a language necessarily pass all the criteria to be judged as having native-like skill in their own language? Taking the United Kingdom as an example, people from the north, in Scotland, have a vastly different accent to those from Inner City London, in England, with much dialectal difference too. Both speak English, but do they both have the fabled native-like pronunciation? On this note, if a person wishes to learn English as an L2, what kind of accent should they be aiming for? Similarly, if their L2 input consists of several varying accents or dialects of the language, the stimulus could become confusing, and as a result of not being able to focus on a single, defined accent, the learner may unconsciously be unable to pick up on how exactly the pronunciation should sound. There are many studies which appear to support the CP hypothesis for pronunciation in SLA, which attempt to show that children can acquire the necessary accent and proper pronunciation for their L2 with relative ease, while older learners struggle or find it impossible to do the same. One such was that of Oyama (1976), who studied sixty male Italian immigrants, all of whom entered the United States between the ages of six and twenty and lived there for time periods ranging from five years to twenty. She examined the degree of American accent they attained, as well as their English listening proficiency, and concluded that age is an important factor to achieve the native-like accent, with the youngest arrivals (those who began learning English before the age of eleven) scoring the highest, and those who arrived after sixteen showing noticeably lower scores (Oyama, 1982, p. 23). However, several other studies have been conducted which show that it is, in fact, possible to achieve native-like pronunciation with appropriate training. Neufeld (1978) trained a group of twenty English-speaking adults over eighteen hours to a sufficient level of Chinese and Japanese that eight and nine of them managed to pass as native-like in their respective languages. This suggests that the CP hypothesis does not hold true in every situation, and that there are exceptions, even sometimes large quantities of them. Looking at these two sets of studies together, it is once again clear that the CP hypothesis cannot be written off entirely. It is obvious that those who start learning an L2 at an earlier stage have some sort of advantage, as it not necessary for them to undergo the same intensive training as older learners, who also require high levels of motivation to attain the proper pronunciation. This does present the question, though, that if we are not to accept the CP hypothesis in its current form, what alternative schools of thought are there for language acquisition?

In the early days of SLA study, during the 1940s and 1950s, the behaviourist theories that were being applied to understand acquisition of a first language (L1) were also applied to SLA. It was thought that those learning an L2 picked it up in much the same way as one did any other skill—through constant repetition, just as a child would learn their L1. Behaviourists also attributed errors produced by the L2 learner to interference from existing knowledge of their L1, arguing that it was “habits” from the L1 that differed in the L2 that caused difficulty. Differences in grammatical structure and phonology between languages were examined in order to predict areas of possible difficulty for language learners in what became formally known as the contrastive analysis hypothesis. Unfortunately, while the theory appeals to the common sense, studies show very little evidence to support it, beyond anecdotal. In fact, very few errors made by L2 learners can be directly linked to attributes of their L1. However, contrastive analysis is not completely irrelevant, as it can be useful in predicting pronunciation problems. It is often possible to tell or make a reasonable guess at speaker’s L1 based on their pronunciation of their L2. In the 1970s, several new views of how SLA happened came forward, one of which was the idea of comprehensible input. This hypothesis, first published by Krashen (1977), attempted to explain language learning as a much more complex interaction of internal and external factors, unlike the series of mechanical habits it had been thought of as previously. The idea of the theory was that language learners are progressing in their acquisition when they can comprehend a level of the L2 that is slightly above their own. The progression was thought to be automatic, in the same way L1 is, as long as the learner was not anxious or harbouring negative emotions towards the L2, its speakers, or anything related to the learning of it. According to Krashen (2003), these feelings create an “affective filter” which can prevent efficient processing of any input the learner is receiving. He also stated that there is a natural order to how learners acquired their L2, and that all languages are acquired in roughly this same order, regardless of the L1. Although this model was criticised for lack of clarity over how to determine the learner’s exact level, it was influential, and later work which built on it was able to more specifically identify learners’ levels. There was also obvious value in them receiving feedback on problems in their speech, which led to the next theory being developed. The interaction hypothesis is based on clarification requests, modified output, and recasts. It suggests that learners can make progress by attempting to communicate with other speakers of their L2, and if they encounter problems, they can resolve them through further interaction with their speaking partner. The teacher can provide corrective feedback based on the issue the learner has. Those who support this theory believe it provides opportunities for learning, but that other sources of learning are necessary as well. However, socioculturalists believe that this kind of interaction is learning in itself. They say that the expert provides scaffolding for the learner to work off, and the opportunity to practise and develop skills within the language, thus enabling them to accomplish more than they could by themselves. In contrast to these approaches, some researchers believe in a universal grammar which guides all language learning. It is largely based on what is thought by supporters of the theory to be the only logical explanation for language acquisition: an innate centre for this kind of learning in the brain, known as a language acquisition device (LAD). It is within the bounds of this theory that the CP hypothesis was developed, arguing that at the offset, the LAD atrophies, and is no longer able to help guide the SLA. While a comprehensive theory which explains every element of SLA has not yet been found, most of these theories have been gradually developed and have often built upon each other, giving us an ever-improving insight into the L2 learning process. Each one has aspects which cannot be ignored or viewed as irrelevant, so it is necessary to take them all into account when studying SLA.

From examining the CP hypothesis itself, it is clear that it does not hold true in every case, as while there is evidence which purports to prove it, there are several studies providing examples of L2 learners achieving both fluency in terms of knowledge of a language and native-like pronunciation. If we accept both of these sets of somewhat contrasting studies as meritable, then we arrive at the conclusion that, while it is not impossible for an adult to fully acquire an L2, it is better to have started learning at a younger age. Taking this in conjunction with the fact that there are several other ideas and theories that also attempt to explain how SLA works, I would argue that it is not a critical period for learning a language, but a sensitive period, after which it becomes more difficult to fully acquire a language, but not impossible. As for how long it lasts, I believe that larger, worldwide studies are needed to properly ascertain the onset and offset, but based on what evidence is already available, I would say that shortly after birth to the beginning of puberty is the most likely timeframe. This sensitive period could explain why there are many cases of older learners not fully mastering a language, but with time, motivation and effort, it can be done.

References:
Krashen, S. (1977). "Some issues relating to the monitor model". In Brown, H.; Yorio, C.; Crymes, R., Teaching and learning English as a Second Language: Trends in Research and Practice: On TESOL '77: Selected Papers from the Eleventh Annual Convention of Teachers of English to Speakers of Other Languages, Miami, Florida, April 26 – May 1, 1977. Washington, DC: Teachers of English to Speakers of Other Languages. pp. 144–158.
Krashen, S. (2003). Explorations in Language Acquisition and Use. Portsmouth: Heinemann.
Lenneberg, E. (1967). Biological Foundations of Language. New York: Wiley.
Neufeld, G. (1978) On the acquisition and articulatory features in adult language learning. The Canadian Modern Language Review 34, 163–174.
Oyama, S. (1976). “A sensitive period for the acquisition of a nonnative phonological system.” Journal of Psycholinguistic Research 5, 261–285.
Oyama, S. (1982). The sensitive period and comprehension of speech. In Krashen, S., Scarcell, R. & Long, M. (Eds.), Issues in second language research. London: Newbury House, 39–51.
Penfield, W. & Roberts, L. (1959). Speech and Brain Mechanisms. Princeton, NJ: Princeton University Press.
Scovel, T. (1988). A time to speak. A psycholinguistic inquiry into the critical period for human speech. Rowley, MA: Newbury House.
Singleton, D. & Lengyel, Z. (1995). The Age Factor in Second Language Acquisition. Clevedon: Multilingual Matters.
Singleton, D. (2005). The Critical Period Hypothesis: A coat of many colours. IRAL 43, 269–285.
Tartter, V. (1998). Language processing in atypical populations. California: SAGE Publications.

References: Krashen, S. (2003). Explorations in Language Acquisition and Use. Portsmouth: Heinemann. Lenneberg, E. (1967). Biological Foundations of Language. New York: Wiley. Neufeld, G. (1978) On the acquisition and articulatory features in adult language learning. The Canadian Modern Language Review 34, 163–174. Oyama, S. (1976). “A sensitive period for the acquisition of a nonnative phonological system.” Journal of Psycholinguistic Research 5, 261–285. Oyama, S. (1982). The sensitive period and comprehension of speech. In Krashen, S., Scarcell, R. & Long, M. (Eds.), Issues in second language research. London: Newbury House, 39–51. Penfield, W. & Roberts, L. (1959). Speech and Brain Mechanisms. Princeton, NJ: Princeton University Press. Scovel, T. (1988). A time to speak. A psycholinguistic inquiry into the critical period for human speech. Rowley, MA: Newbury House. Singleton, D. & Lengyel, Z. (1995). The Age Factor in Second Language Acquisition. Clevedon: Multilingual Matters. Singleton, D. (2005). The Critical Period Hypothesis: A coat of many colours. IRAL 43, 269–285. Tartter, V. (1998). Language processing in atypical populations. California: SAGE Publications.