Understanding Second Language Acquisition (2nd ed)

by Rod Ellis

  This historical introduction to SLA has neglected two important areas of enquiry: the role of age in L2 acquisition and the role of individual learner factors such as language aptitude and motivation. We will remedy these lacunae in Chapter 2 and Chapter 3. This chapter has not considered the role of instruction either, an area which has received considerable attention in SLA—not only because investigating how instruction affects learning serves as a means of testing the claims of different theories—but also because SLA has, from the start, been concerned with how it can contribute to effective language pedagogy. The significance of the role of instruction is considered in Chapter 10 and Chapter 11.

  Notes

  1 This definition of L2 acquisition excludes children who are acquiring two languages as their mother tongues. This situation is referred to as ‘simultaneous bilingual acquisition’.

  2 Some researchers prefer the term ‘additional language acquisition’ to second language acquisition’ because it avoids the potentially negative connotation of ‘second’. For some learners—such as those who leave their own country and migrate to a country where the L2 is widely used—the language that they learned ‘second’ can become their ‘first’—i.e. primary—language.

  3 Ortega (2012) argued that the ‘human language faculty is potentially by default bi/multilingual’ and that ‘the possibility of bi/multilingualism remains true all along the life span, from birth across all ages’ (p. 17).

  2

  Age and second language acquisition

  Introduction

  A main aim of SLA is to identify those aspects of L2 acquisition that are universal. This is entirely justified as human beings share a common propensity to learn language so that we can expect them to grapple with the cognitive and social issues involved in acquisition in much the same way. However, learners also differ in a number of ways and these differences impact on the rate of acquisition, ultimate success, and—potentially—on the processes involved in acquiring a language. Thus—before embarking on the detailed study of the various dimensions of acquisition introduced in Chapter 1—in this, and in the following chapter, I will focus on the learner, pointing out some of the ways in which learners differ from each other and how these differences impact on learning.

  One of the most obvious ways in which second language learners differ is in terms of their starting age. Some learners start learning a second language in early childhood while many others have to wait till they go to school (often secondary school). The study of the influence that learners’ starting age has on L2 acquisition is motivated by both theoretical and practical considerations. It is a source of empirical evidence about the nature of human beings’ language learning capability and how this may change as they grow older. It also provides educational policy makers with information that they can use to decide when to introduce the teaching of a foreign language into the school system and also what instructional approach might be best suited to learners of different ages.

  A common assumption is that children are better language learners than adults and that, therefore, learners will be more successful if they start learning a second language when they are young. However—while there is some truth in this assumption—the research evidence shows that the effects of age on L2 acquisition are complex. For a start, it is necessary to distinguish the effect of age on ultimate attainment, the rate of acquisition, and the route of acquisition. I will begin by examining age in relationship to ultimate attainment as this is the most controversial and theoretically interesting issue. I will then consider its effect on the rate and route of acquisition. This section will conclude with a consideration of educational policy regarding the best age to start foreign language instruction.

  Many of the studies I will consider in this chapter are correlational in nature. That is, they obtained measures of two different variables (for example, starting age and L2 proficiency) and then calculated the strength of the relationship between them statistically. A perfect correlation—never attested—is represented arithmetically as 1.0. A strong correlation is one that is close to the perfect correlation (for example, 0.85) while a weak correlation is one that is close to zero (for example, 0.15). Correlations can be positive (i.e. the stronger the learners’ motivation, the stronger their L2 scores) or negative (i.e. the older the learners’ starting age, the weaker their L2 scores).

  Age and ultimate attainment

  To address the relationship between starting age and ultimate attainment, two separate questions need to be investigated:

  Can adult learners of a second language acquire native-like proficiency in a second language?

  Do learners who start learning a second language in childhood acquire higher levels of L2 proficiency than learners who start as adults?

  To answer these questions, it is necessary to define the variables involved. What is meant by an ‘adult learner’ and by a ‘child’ learner? What is ‘language proficiency’? In fact, these constructs are not easy to define. Some researchers (for example, Snow and Hoefnagel-Höhle 1978) have distinguished prepuberty learners, adolescents, and adults (i.e. post-puberty learners). However, this constitutes a somewhat crude way of characterizing ‘age’; there is a big difference, for example, between a four-year-old and a ten-year-old child and perhaps an even bigger one between a 17-year-old and a 50-year-old adult learner. Researchers have investigated ‘proficiency’ primarily in terms of pronunciation (for example, the learner’s accent) and grammar, but there are other important aspects of proficiency that need to be considered—for example, lexical knowledge (including formulaic sequences) and the pragmatic ability needed to use language in sociolinguistically appropriate ways—which have been little investigated. A further issue is whether proficiency is to be conceptualized as implicit knowledge (i.e. the kind of knowledge needed to engage in fluent, spontaneous language use) or explicit knowledge (i.e. the kind of knowledge that many traditional language tests tap).

  Critical Period Hypothesis

  The first of the above questions is the most important one from a theoretical standpoint. It concerns the Critical Period Hypothesis (CPH), first proposed by Penfield and Roberts (1959). The hypothesis states that there is a period (typically defined as the period up to the onset of puberty) during which learners can acquire a second language easily and implicitly and achieve native-speaker competence, but after which L2 acquisition becomes more difficult and is rarely entirely successful.

  Evidence in support of the hypothesis initially came from outside SLA. Lenneberg (1967) reported research which showed that when children suffered injuries to the right hemisphere they experienced language problems, but adults did not. Conversely, when adults underwent surgery to the left hemisphere almost total loss of language occurred whereas this did not occur with children. Adults showed permanent language impairment after such operations, but children rapidly recovered total language control. Lenneberg concluded that the biological basis of language in children and adults differs.

  Further evidence for this conclusion came from cases of unfortunate children who were deprived of opportunities to hear and speak a language during childhood. Curtiss (1977) reported a study of Genie, who was kept in virtual isolation for most of her life. When discovered at the age of 13, she had no language. Genie was subsequently successful in learning English to some extent, especially vocabulary, but failed to achieve full grammatical competence. She also had problems in engaging in normal social interaction. Curtiss suggested that her limited grammatical development was because she had passed the critical period for language acquisition. Rymer (1993) pointed out, however, that the root cause may have been the emotional disturbance that Genie had experienced as a child and continued to manifest as an adult. This could explain her problems with social interaction which in turn may have limited the development of her grammatical competence.

  There is no clear consensus on when the critical period for language learning ends. Singleton (2005) in a survey of the literature that
addressed this issue reported claims ranging from near birth to late adolescence. Also, it has become clear that—if there is a critical period—this varies depending on the aspect of language under examination. Granena and Long (2012), for example, provided evidence to suggest that the window of opportunity closes first for L2 phonology (perhaps as early as four-years-old), then for lexis and collocation, and finally (in the mid-teens) for grammar.

  Theoretical importance of the CPH

  The theoretical importance of the CPH lies in the fact that it supports Chomsky’s (1965) view of language. Chomsky argued that children are equipped with a language acquisition device—an innate, biologically-given capacity for learning language that is distinct from other cognitive abilities. This device—according to Chomsky—contains knowledge of the linguistic universals that underlie the grammatical rules of every language and—because children have access to these universals—they are able to master the grammar of their mother tongue. This nativist account of language learning, which downplays the role of the linguistic environment, was prominent in the early period of SLA when behaviourist learning theory with the emphasis on habit-formation was called into question (see Chapter 6). It provided a psycholinguistic rationale for the CPH. Children acquire full grammatical competence because they have access to the language acquisition device. Adults, however, lose access to it and therefore are forced to rely on general cognitive abilities such as those associated with the formal operational stage of development (Piaget 1973), which begins at the onset of puberty. At this point, people develop the capacity for logical thought, deductive reasoning, and systematic planning. Such abilities suffice to enable people to learn a language to a certain extent but do not totally compensate for the loss of the language acquisition device. As a result, older learners fail to acquire all of the grammatical features of the L2.

  Investigating the CPH

  The CPH has been subjected to empirical study in two main ways. One way involves comparing groups of learners who commenced learning as children with other groups who started post puberty. The second way involves investigating whether learners who started learning post puberty were successful in achieving full competence in the second language (i.e. had become totally native-like). These two approaches reflect somewhat different versions of the hypothesis. One version is that the end of the critical period signals the point at which decline in the ability to learn a second language begins. This can be considered the weak version of the hypothesis. The strong version of the hypothesis is that once past the critical age, natural acquisition is blocked irrespective of whether learners are just past it or many years past it. What both conceptualizations have in common, however, is the notion of a discontinuity in learning; that is, after a certain age, the pattern of learning changes.

  One of the most commonly cited studies is Johnson and Newport (1989). They studied 46 native Koreans and Chinese who had arrived in the United States between the ages of three and 39, half before the age of 15 and half after 17. The learners were asked to judge the grammaticality of 276 spoken sentences, about half of which were grammatical. Overall the correlation between age at arrival and correct judgement scores was –0.77 (i.e. the older the learners were at arrival, the lower their scores). In contrast, neither the number of years of exposure to English beyond five nor the amount of classroom instruction they had received was related to their grammaticality judgement scores. Also, although an effect for ‘identification with American culture’ was found, this was much weaker than that for age. Johnson and Newport interpreted the results of their study as evidence for the strong version of the CPH as they argued that there was a sharp discontinuity in the effects evident for age after the critical period. In the case of the early starters, there was a gradual decline in performance according to age. However, in the case of the late starters, the relationship between age and performance was essentially random. However, this claim was subsequently disputed. Bialystok and Hakuta (1999) reanalysed Johnson and Newport’s data and concluded that there was no evidence of a clear discontinuityNOTE 1.

  Turning now to the second approach to investigating the CPH, I will focus first on two studies that, like Johnson and Newport, measured achievement by means of grammaticality judgement tests (i.e. tests that present learners with a set of sentences and ask them to judge whether they are grammatical or not). I will then consider a number of other studies of post-puberty L2 learners that examined attainment in ways that some researchers (including me) consider more valid.

  Coppetiers (1987) compared the performance of 20 native speakers and 21 highly proficient learners of French on a grammaticality judgement test. All the learners had begun learning as adults and they all appeared nativelike in their spoken French. The results showed clear differences between the two groups. Coppetiers concluded that the grammatical competence of the L2 learners differed from that of native speakers. However, Birdsong (1992) replicated this study and reported very different results. He administered a grammaticality judgement test to 20 English-speaking learners of L2 French who were native-like in their oral ability and to 20 native speakers of French. Birdsong found no evidence of any dramatic differences in the judgements of the two groups. A number of the non-native speakers performed in the same range as the native speakers. This study, then, suggests that at least some learners who start learning a second language after puberty can achieve a level of competence indistinguishable from that of native speakers.

  Other studies carried out in-depth investigations of individual learners. Table 2.1 summarizes two such studies. Ioup, Boustagui, El Tigi, and Moselle (1994) studied a highly successful learner (Julie) who did not start learning her L2 (Arabic) until she was 21 years. Lardiere (2007) studied a learner (Patty) who had had almost no contact with the L2 (English) until she was 18 years old but subsequently—like Julie—was immersed in the L2 environment. Both learners had been learning their L2 for more than 20 years and both—especially Julie—demonstrated a high level of grammatical proficiency. However, arguably neither of them achieved totally native-like ability. Julie, for example, did make some mistakes in a translation test and did not perform exactly like native speakers in a grammaticality judgement test. Patty continued to make morphological errors in features such as subject-verb agreement and plural -sNOTE 2. These studies, then, lend support to the CPH.

  Ioup et al. (1994) Lardiere (2007)

  Learner Julie was an L1 speaker of British English. When she married an Egyptian she moved to Cairo at the age of 21. She reported that Arabic became her dominant language after her third year of residence in Cairo. She had been living in Egypt for 26 years at the time of data collection. She was entirely a naturalistic learner (i.e. she never received any formal instruction in Arabic). Patty was Chinese. She was born in Indonesia in 1953 but left for China in 1969 and lived there for two years before moving to Hong Kong. After finishing high school, she worked in an import-export company in Hong Kong, rarely speaking English. She arrived in the United States in 1976 at the age of 22. She lived with her Vietnamese fiancé’s family and began college-level study and, a little later, took a waitressing job. In 1985, she separated from her husband and in 1989 married a native English speaker.

  Target language Arabic English

  Data collection Data collected from both native speakers of Egyptian Arabic, Julie and other L2 learners: Oral recordings of their explanations of their favourite recipes.

  A test of their ability to distinguish different Arabic accents.

  Oral recording (1) was made in 1986 when Patty had been living in the United States for about 10 years. Recordings (2) and (3) were made two months apart in 1995. Written samples, mainly from email messages, were also collected.

  Measures of proficiency Native-speaker ratings of all the participants’ accents; Julie’s ability to distinguish different Arabic accents; a translation task (English into Arabic); a grammaticality judgement test (GJT); a test of ability to process complex syntactic structure (anaphora).
Lardiere investigated a number of grammatical features in Patty’s English speech and writing, including tense, aspect, and agreement, English question formation, and possessive pronouns and plural marking.

  Results Seven of the 13 judges rated Julie’s accent as native. Julie was able to successfully distinguish Arabic accents. She made very few mistakes in the translation test. Her judgements differed from those of native speakers on only five out of 37 sentences. Her ability to process complex syntactic structure was comparable to the native speakers’. Patty had not achieved a native-like grammar by the end of the study. She was able to employ English syntactical constructions correctly, but continued to make errors in morphological features (for example, frequent omission or overuse of inflectional markers on verbs and nouns). Patty’s written English was more accurate than her spoken English.

  Table 2.1 Summary of two studies of advanced adult L2 learners

  However, other studies suggest that L2 learners who started learning post-puberty were able to achieve native proficiency. Bongaerts (1999), for example, used native-speaker ratings to investigate whether nine post-adolescent Dutch learners of L2 French had attained a native-level accent. Recordings of their speech were mixed in with recordings of 18 lower-level Dutch learners of French and nine native speakers. Three of the advanced learners passed for native speakers. Bongaerts argued that high motivation combined with pronunciation training enabled these talented learners to achieve native level.