Neural Plasticity and the Issue of Mimicry Tasks in L2 Pronunciation Studies

Yvonne F. Stapp
Gaikokugo Center
University of Tsukuba, Japan
<stapp@sakura.cc.tsukuba.ac.jp>

Abstract

In second language research, the ability to mimic foreign words is frequently cited as evidence for neural plasticity. However, if the type of neural plasticity related to language acquisition is not manifested in mimicry, the assumed connection is questionable. In an investigation of the relationship between mimicry skill and neural plasticity, 28 monolingual Japanese subjects age 4-17 repeated a list of simple English words containing /r/ and /l/. Analyses were made of individual and age-group scores, and the consistency of individuals' pronunciation across word tokens. In the aggregate, the adolescents proved superior to the children. However, only one adolescent actually scored high enough to qualify as a good mimic. The results here suggest that mimicry ability is not related to age, but is really a talent available to particular individuals throughout life. This is different from the neural plasticity which gives young children a long-term advantage in L2 pronunciation, whether or not they are good mimics at the outset.

1. Introduction

The neural circuitry of a phonological system is established over a lengthy process in both first and second language. In first language, where phonological development overlaps with the development of associated cognitive and motor systems, the neural circuitry is generally not complete until at least age three or four (Gazzaniga, 1994; Kosslyn and Koenig,1992; Wode, 1989). By contrast, in early second language acquisition (L2) beginning at between approximately age five or six, children often achieve native-like pronunciation and syntax within a relatively short period of a year or two. The ability of young children to achieve native-like proficiency in a foreign language in a rather short time is a reflection of a type of neural plasticity, and it appears to be related to the distinct characteristics of the young brain. However, it must be stressed that, even with the advantage of greater neural plasticity, the formation of the L2 phonological system in the brain still requires time. [-1-]

In view of the time factor, the ability to imitate unfamiliar foreign words on a single presentation as in mimicry tasks would seem to be unrelated to neural plasticity. However, research in second language often assumes this relationship. Numerous studies based on tasks in which subjects of different ages mimic words and/or phrases in a foreign language have described an immediate advantage in L2 pronunciation for young children over older children and adults (e.g., Cochrane,1980; Cochrane and Sachs, 1979; Fathman, 1975; Yamada, Takatsuka, Kotabe, & Kuruse, 1980; Tahta, Wood and Lowenthal 1981b). The implication of such studies is that the superior performance in mimicry exhibited by young children is due to the relative neural plasticity in children which promotes native-like L2 pronunciation in early acquisition. The following comment by Long (1990) makes this assumption clear:

[The] sharp drop in imitation abilities. . . [after age six] can be accounted for by positing that maturational constraints begin to set in as early as 6 for suprasegmental phonology in some learners and soon after that for segmental phonology. . . .Phonological attainment is strongly conditioned by learner age. (p. 266)

This study attempts to clarify the distinction between mimicry skill and neural plasticity. First, neural plasticity is described (Section 2). Then the general long-range advantage of young children in the acquisition of second language (L2) accent, which reflects neural plasticity, is reviewed (Section 3). The claims and implication(s) of L2 mimicry studies are then evaluated, and studies favoring young children are compared with those favoring older learners (Section 4). A study of the mimicry skills of 28 Japanese children and adolescents, age 4-17, is then reported (Section 5). In this experiment, adolescents proved superior to children in the aggregate, but true mimicry skill proved to be an individual talent when the data were carefully analyzed. The findings indicate that the ability to mimic foreign words is unrelated both to age and to neural plasticity. The implications of these findings are considered in the conclusion (Section 6). [-2-]

2. Neural plasticity

Brain functions are based on multiple integrated neuronal circuits that respond to diverse stimuli and process information. The circuits consist of huge numbers of individual neurons, which individually and collectively exhibit plasticity in the sense that they are able to respond to new information and experience. The new information is instantiated in some type of physical modification of the respective circuits. In sensory systems, such as sight or touch, and in cognitive systems such as language, this ability to reorganize makes it possible to recognize new objects or learn a new language. In this respect, neural plasticity is essentially a permanent feature of the brain, ensuring continued adaptation and learning throughout life. However, the same degree of plasticity is not available for all types of learning over a lifetime. While we tend to "learn" new faces in much the same manner throughout life, some types of information and skills (like a foreign language) are as easier to learn in childhood than in maturity. This is an indication that the availability plasticity differs across brain functions. Conditions governing some types of plasticity clearly differ from one stage of maturity to another (Greenough, 1986; Greenough, Black and Wallace, 1996; Lauder and Krebs, 1986; Shepherd, 1988). Distinctions obtain, for instance, between the neural conditions which govern critical periods in the development of motor, sensory, and cognitive systems (including first language), and the neural conditions which support various types of learning after fundamental neural systems are in place (Humes-Bartlo, 1993; Trevarthan, 1987). Because of critical periods, certain types of learning and experience are compromised by age. The evidence for this is well established for the development of sensory and motor systems. The types of experiments used for investigating the development of these systems and cognitive systems the brain cannot be performed on human subjects. However, it is possible to extrapolate from the available evidence that there are time constraints for a number of brain systems, including cognitive systems such as language.

2.1 Critical periods

Critical periods in the development of fundamental neural systems (e.g., visual, auditory, cognitive) are "critical" in the sense that appropriate stimuli must intersect with neurochemical and other neurobiological sequences in a specific time frame. These phases are marked by neural characteristics that do not reoccur after the critical period. Such features include an "anticipatory" condition in brain cells preceding the actual onset of stimuli and a heightened response to relevant stimuli at the appropriate time, representing a genetic predisposition to specific stimuli which will trigger development of the system (Black, 1994; Lauder & Krebs, 1986). Another feature of a critical period is the unique manner in which the system takes shape in ontogenesis. In the infant brain [-3-] there is an overabundance of neural connections, and these compete for a limited amount of neurochemical factors which serve development and stability of neural circuits. During the developmental process, projected neural systems are carved from the most viable of the excessive connections, leaving many cells to atrophy (Black, 1994; Brown, Hopkins and Keynes, 1991; Greenough, 1986). These two features, the anticipatory condition which stimulates formative connections (but in overabundance) and the pruning condition which selects the system's most viable connections, are peculiar to the critical period. Responses to stimuli after the formative phase have different characteristics.

A critical period for first language development is suggested by several types of evidence.[1] For example, anticipatory brain lateralization is apparent in the fetal stage, where the future language areas in the left hemisphere are already larger than homologous areas in the right hemisphere (Springer and Deutsch, 1985; Trevarthen,1987). In addition, human infants are extraordinarily sensitive to language sounds very early in life. Even at the age of 2-3 months, well before language onset, infants are able to discriminate phonemes from any number of languages (Eimas, 1975; Kuhl, 1980, 1981,1985; Werker and Tees, 1984). [2] This capability gradually declines even in childhood, and by adolescence most individuals experience difficulty discriminating unfamiliar phonemes in foreign languages.

Other indications of a critical period for language come from the substantial documentation of language deficiency or impairment as a result of late language development. Incomplete L1 acquisition has been described for cases of brain damage (Ager, Ernhart, Martier, Greene, and Sokol, 1990; Lenneburg, 1967; Vargha-Khadem, Carr, Isaacs, Brett, Adams, and Mishkin, 1997), delayed exposure to sign language for congenitally deaf children (Newport, 1984), and social isolation and neglect in early childhood (Candland, 1993; Curtiss,1977). [3]

2.2 Post-critical period neural conditions

Toward the end of a critical period, important changes occur which alter the conditions of neural plasticity and the response to stimuli thereafter. For example, cytoskeletal components involved in the stabilization of cells and circuits peak in different brain regions and at different times in development (Cronly-Dillon and Nona, 1988). In addition, certain of the neurochemical factors operating in the developmental stages of a neural system undergo a functional shift (Black, 1994; Parnavelas, Papadopoulos, Cavanagh, 1988). Although these same factors may continue to operate within the mature system, they take on different functions, and there is no reversion to their respective ontogenetic roles. The neurochemical conditions of a critical period cannot be revived after the period has lapsed (Black, 1994; Brown, et al., 1991; Greenough, 1986; [-4-] Lauder & Krebs, 1986). This functional alteration at the chemical level is one reason why some fundamental neural systems (visual, auditory, cognitive) develop incompletely if they are somehow blocked during the critical period. Such systems generally remain attenuated even with subsequent compensatory stimuli, once the optimal period for development lapses.

For language and other cognitive systems, changes in the neural condition also occur as a result of the differential maturation schedules of participating cognitive components. Language acquisition, for example, requires several types of memory, which themselves develop in relation to other brain systems. For instance, one type of memory depends upon the development of the limbic circuitry, which develops quite slowly (Black, 1994; Cotman & Lynch, 1990). Cortical substrates that participate in various types of learning are also subject to different types of maturational schedules and shifts. Research on monkeys (Cotman & Lynch, 1990; Goldman-Rakic, 1982,1992) and humans (Greenough, 1986; Shatz, 1992; Trevarthen, 1987) indicates that a learning task associated with one cortical region very early in life may be assigned to another cortical substrate if the same task is learned later on, with the implication that the task is learned somewhat differently depending on when it is learned. Language might be an example. While first language (including simultaneous bilinguality) is densely represented in the language areas of the left hemisphere, a second language acquired later is typically more distributed in the brain (Ojeman and Whitaker, 1978). [4] Recent experiments using functional magnetic resonance imaging (fMRI), have provided new evidence about such neurofunctional patterns of distribution. Investigations by Kim, Relkin, Lee, and Hirsch (1997) confirm that second languages acquired in very early childhood are spatially close to first language in the frontal lobe, while languages acquired later are well separated in this region.

Another type of change which characterizes the post-critical period neural state is that in place of the anticipatory condition of the critical period and the subsequent pruning which creates the neural system, learning and experience are represented in massive dendritic expansions (Greenough, 1986; Greenough, et al., 1996). This latter mechanism is the brain's continued response to stimuli over the lifetime. Studies on rats have demonstrated that even mature animals respond by dendritic elaboration to enriched environments (Diamond, Rosenzweig, Bennett, Lindner, and Lyon, 1972; Greenough, 1986;), and such elaboration reflects one or more types of permanent neural plasticity. [5] However, this fact should not encourage either the idea that the neural chemistry of critical periods and post-critical periods are the same, or the optimistic view that all types of learning have the same potential over the life span. Although the rats in these famous experiments were not learning language, such research has influenced the view that second language acquisition is not subject to maturational constraints." Advocates [-5-] of this view tend to argue that a near-native level of L2 acquisition is potentially feasible at any time in life because neural plasticity is a permanent feature of the human brain (e.g., Bongaerts, van Summeren, Planken, and Schils, 1998; Flege, 1987; Jacobs, 1988; Klein, 1986; Neufeld, 1980). A related version of this notion is that older children and adults actually have an advantage in L2 acquisition over younger children because of superior cognitive skills which are deployed in the initial stages of L2 acquisition (e.g., Ellis, 1985; Genesee, 1988; Walsh and Diller, 1981). In view of the fact that ultimate proficiency is the real issue in second language acquisition, it is important to understand the nature of neural plasticity serving different types of learning. While some type of advantage may exist at some stage for older language learners, their ultimate proficiency is not statistically equal to that of young learners.

Although delayed acquisition of first language at age five or six is problematic for complete development, the same age is ideal for attainment of native-level second language acquisition (Johnson and Newport, 1989; Long, 1990; Patkowski, 1990; Scovel, 1988; and Seliger, 1978). The type of neural plasticity that at this age facilitates second language seems at least partially associated with skills involving motor coordination. Phonology and syntax share with certain types of musical, dance and sports training the requirement of early instruction for optimal potential development. [6] The plasticity associated with finely coordinated motor skills diminishes with maturity. It is likely that this plasticity constraint is responsible for the common experience of foreign accent and the incomplete mastery of the syntax of a second language when learning begins in adolescence or adulthood. [7]

3. Children's long-range advantage in L2 pronunciation

4. Mimicry and the issue of immediate advantage in L2 pronunciation

Research findings on long-range acquisition are consistent with the principles of neural plasticity. However, numerous studies have reported an immediate L2 pronunciation advantage for young children based on their success on imitation tasks. Cochrane and Sachs (1979), for instance, in a comparison of 7-year-olds and adults on mimicry tasks in Spanish, found the children were superior at outset. Yamada, Takatsuka, Kotabe, and Kuruse (1980) described similar results in a comparison of younger and older children. Cochrane (1980) compared the ability of Japanese children 8-13 and adults 21-45 to mimic the difficult English /r/ and /l/ in nonsense words, short English phrases, and spontaneous descriptions. She found that the children demonstrated an impressive advantage over the adults even at the earliest stages. Fifty percent of the children -but no adults-- were rated as "native."

One of the largest mimicry studies was an investigation of 231 English speaking children ages 5-15 by Tahta, Wood, and Lowenthal (1981b). Subjects were required to imitate, without rehearsal, audiotaped French and Armenian words and phrases. The authors found that the ability to mimic single words and intonation patterns was best in the age 5-7 group, and that the ability declined over the older groups (ages 8-11, 12-15). The authors concluded that if L2 acquisition begins by age 6, there is no transfer of L1 accent; but after that age the degree of transfer in the L2 accent increases steadily.

The putative link between mimicry skill and neural plasticity would perhaps be more convincing if the mimicry skill were observed only in young children. However, in some studies older children or adults have performed better on imitation tasks than young subjects. In a partial replication of the experiment by Tahta et al.(1981b) described above, which had demonstrated an initial advantage for young children, Lowenthal & Bull (1984) found an advantage instead for older subjects. The authors used stimuli similar to that in the Tahta et al. (1981b) study but altered the presentation strategy. Their 39 (English L1) subjects ages 7-15, like those in the Tahta study, mimicked pre-recorded syllables, words and phrases in Armenian. But in the earlier experiment by Tahta et al.(1981b), the presentation of target items to the youngest children was handled by testers who actively encouraged the children. Lowenthal & Bull provided no special conditions. The instructions and modeling of Armenian tokens were presented on audiotape and all subjects were required to repeat what they heard. The results favored the older subjects. Lowenthal & Bull attributed the superior performance of the older children to the experimental conditions themselves, and concluded that manipulation of testing conditions can substantially affect the performance of young children. Studies claiming equal or superior mimicry ability in older subjects have more typically allowed practice sessions. In part of Cochrane's (1980) study, [-7-] subjects were a permitted a small amount of intensive practice. Cochrane found that such training affected the adults but not the young children. In a study by Snow and Hoefnagel-Hohle (1977) practice sessions for imitation of target nonsense words also proved advantageous to the older learners.

The rehearsal strategy has been exploited particularly by Neufeld (1978; 1979), who has promoted a popular view among language instructors especially that adults can achieve a native accent in a second language. In a rather unusual experimental protocol Neufeld selected talented bilingual adults whose L2 pronunciation was already very good. These subjects were not compared with younger subjects, nor with mature individuals previously unexposed to the L2. The select groups were allowed to practice reading aloud and recording short passages as many times as they liked before being judged. Despite these unusual conditions, Neufeld made two remarkable claims. First, he hypothesized that all adults could acquire native-like pronunciation in an L2 provided they are given ample practice. Second, Neufeld argued that his experiments demonstrated the permanence of neural plasticity for second language acquisition and that the notion of a sensitive period -i.e., early childhood--for second language is false. This view has received enthusiastic endorsement in various L2 research sectors, as the following example from Klein (1986) indicates:

[The] biological explanation [for difficulty in L2 acquisition after puberty] can be replaced, or supplemented, by arguments of a social nature. It may well be, for example, that the adult is much less willing to give up his well-established social identity. Even in the case of phonology-including intonation-where adult second language learners often seem to encounter special difficulties, investigations by Neufeld (1979) have shown that suitably motivated adults are capable of mastering to perfection the pronunciation of the (for them) most exotic languages, as revealed by the fact that native speakers could not recognize any "foreign accent" in their speech. This shows that ideal second language acquisition is biolo- gically feasible even after the age of puberty (p. 10).

Flege (1987) has advanced a similar claim in a review of work by Neufeld and other studies in which older children or adults were initially superior to young children. On the basis of such evidence, Flege rejects the notion of critical periods, and argues that neural plasticity affecting second language acquisition is not subject to time constraints; full L2 acquisition is potentially possible at any time. [9] A more sophisticated, but related, argument by Jacobs (1988) advances the notion of native-like potential in mature L2 learners because "different neural structures can give rise to equivalent behavior" (Jacobs, 1988: 326). [-8-]

More than a decade ago Walsh and Diller (1981) argued for the superiority of older L2 learners based on evidence that neural circuits are modifiable. They maintained that, with the exception of pronunciation, older L2 learners were more efficient than children in acquiring grammar and lexicon, and were superior in receptive skills (e.g., reading). Notably, they did not address the problem of fossilization in syntax, the other significant difference between child and adult L2 acquisition. But knowledge of grammar rules is very different from the ability to accurately deploy the rules in speech and writing. The former is an intellectual skill, and the latter is a productive skill. Productive skills are particularly compromised in older learners, and these skills are related to neural plasticity. That neural circuits are modifiable for learning is not generally in question; that neural circuits for the productive skills (including the use of syntactic rules) are modifiable enough to ensure native-like attainment is simply not demonstrated.

The confusion about neural plasticity is because scientific information in this area is quite limited and technical. There is presently very little work devoted to second language. However, regarding pronunciation, it is clear that neural plasticity cannot be invoked as the cause for superior mimicry in both young children and older subjects until it can be demonstrated that older, average L2 learners really do compare with younger, average L2 learners in ultimate proficiency. For the reasons discussed in Section 2 above, the type of neural plasticity involved in language development and other (motor-related) skills is not available over the life span. Therefore, if skill in mimicry tasks does reflect neural plasticity, a serious revision of the plasticity concept itself is necessary. The ability to mimic foreign words or phrases at any age is probably due to factors other than neural plasticity. The likely candidate is simply innate skill in mimicry. However, some research based on talented L2 learners has argued for personality factors such as motivation, or external factors such as learning environment variables (e.g., Bongaerts, van Summeren, Planken, and Schils, 1998; and White and Genesee, 1996).

5. Experiment: Comparison of Japanese children and adolescents in mimicry tasks

This experiment was motivated by conflicting L2 research that in some cases has presented evidence of superior mimicry performance by children, and in other cases, superior mimicry by older subjects. In both cases, mimicry skill has been attributed to neural plasticity. In the present study, 28 Japanese children and adolescents were compared across groups and individuals for their mimicry ability. The subjects were grouped as follows: children, age 4-11, with zero exposure to English; junior high school, age 13- 15, with first exposure to English lessons; and high school adolescents, age16-17, with 2-3 years English instruction. The [-9-] 4-year old children had acquired the basic syntax and phonological system of their native language. The number of subjects for each group depended partly on availability for two sessions.

age 4 (preschool): 4 subjects
age 5-6 (kindergarten): 5 subjects
age 8-11 (elementary school): 2 subjects
age 13-15 (junior high school): 9 subjects
age 16-17 (high school): 8 subjects

5.1 Method.

Subjects were required to mimic a short list of English words that contained /l/ or /r/. The task was made simple in order to accommodate the limitations of the youngest children. The data were analyzed across several dimensions, as described below. Two tokens of each word-item were obtained in two sessions separated by approximately one week. This strategy guaranteed at least one useable token per word from each subject, a major concern in the case of the young children. This was much less a problem with the adolescents. The two-session approach also made it possible to compare subjects' consistency in imitation skill for the same word at two different times and the consistency of mimicry skill from one word to another. For example, the utterance of a word like bear could be compared in session #1 and session #2, and the word bear could be compared in the scoring with the word chair in two different sessions.

All subjects were familiarized with the task in advance by means of a brief practice using different words from those on the target list. The target words were then modeled one at a time by the investigator and the subjects repeated the words into the recorder microphone.

The 4-year-olds presented a measure of difficulty because of their unfamiliarity with both foreigners and the nature of the task. To make them more comfortable, a Japanese mother who knew all of the [-10-] youngest subjects sat with each child while the investigator selected a toy and gave its English name (e.g., duck). The mother then explained to the child in Japanese that the investigator would name these items with different words, and that the child was to repeat the words into the tape recorder microphone. During the warm-up session, each child recorded a word that was not on the target list (e.g., duck), and then listened to his/her own voice on playback. This made the unfamiliar situation a kind of entertainment, thereby facilitating the elicitation procedure. Once each 4-year-old was comfortable with the situation, the investigator presented a toy representing a target word (e.g., doll, bear, chair) and said the word in English very slowly and clearly. The assisting Japanese mother then encouraged the child to repeat the word.

For the kindergarten and elementary groups, line drawings of the target word-items were substituted for objects. In each group, a Japanese teacher explained the task to the children and then sat nearby during the procedure to reassure and encourage them. Again, a short practice was provided to familiarize the children with the task and the taping equipment. For the adolescents, an explanation of the task by a Japanese teacher was adequate, and no warm-up session was necessary. Line drawings with printed labels of the word-items were presented, and the students were permitted to read each word as the investigator pronounced and they repeated.

The speech tokens from all the subjects were randomized on an audiotape. The tape contained the first useable token for each word from each subject plus a randomly selected second token of one word per subject. In addition, fifteen exact duplicates of words were randomized with the other tokens in order to provide some indication of consistency of judgment. The evaluation sheet contained a list of numbered tokens, with no identification of the randomized voices. A linguistically untrained native-speaker judge ranked the tokens on a scale of 1-5 as follows:

1- very bad accent, very difficult to understand word
2- poor accent, but possible to understand word
3- average or fair quality non-native (very distinct accent)
4- very high quality non-native speaker
5- indistinguishable from a native speaker

5.2. Results

The scores were analyzed across several dimensions: the relationship between age and mimicry skill; individual high scorers; subjects' score variability across different tokens; and score variability across identical word-item tokens. [-11-]

5.2.1. Age and mimicry skill.

An analysis of individual scores showed that the children ranked significantly lower than the adolescents, and that all the children were below the "average" score of 3.0. The adolescents, while varying in scores, were closer to the 3.0 range overall, as indicated in Table 1.

Table 1: Comparison of individual scores: Low to high


   Children (age 4-11)  Adolescents (age 13-17)


     (age  4)  1.37        (age 17)  2.42
     (age  5)  1.71        (age 13)  2.57
     (age  4)  1.75        (age 13)  2.71
     (age  4)  1.75        (age 13)  2.71
     (age  6)  1.87        (age 15)  2.75
     (age 11)  2.00        (age 14)  2.85
     (age  8)  2.00        (age 17)  2.85
     (age  5)  2.12        (age 16)  3.00
     (age  5)  2.14        (age 17)  3.00
     (age  4)  2.25        (age 17)  3.00
     (age  5)  2.37        (age 17)  3.00
                           (age 14)  3.12
                           (age 15)  3.12
                           (age 17)  3.25
                           (age 16)  3.28
                           (age 15)  3.42
                           (age 15)  3.57

Table 2: Mean Scores by Age Group (low to high)


     Age 4-11      (N = 11)    1.93
     Age 13-15     (N = 09)    2.98
     Age 16-17     (N = 08)    2.97

5.2.2 Individual mimicry skill.

Scores considered in the aggregate give the impression that all or many adolescents were good mimics, with the possible implication that the adolescents would be more likely to become native-like in L2 pronunciation than the children. However, the aggregate scores are misleading in this respect. When the scores of the adolescent [-12-] subjects were evaluated individually, only one adolescent subject appeared to have a distinct skill in mimicry. Table 3 ranks the children, junior high school (JHS) and high school (HS) subjects for number of tokens receiving a 4.0 score. Only one subject (age 15) received scores of 4.0, and his high scores were not consistent across all tokens. None of the children age 4-11 received a score of 4.0 on any token. For the adolescents, scores of 4.0 were awarded to less than 50% of the individual tokens, with the single exception of the high scorer.

What emerges from these data is that mimicry skill is not generalized in the adolescent group--that is, age is not the determining factor. Rather, mimicry skill appears to be an individual talent, and in any group of subjects is likely to be dispersed somewhat unpredictably. For this reason, where scores are considered only in the aggregate, one or two high scorers in a group may skew the collective data, such that an association might be mistakenly made between scores and age, when in fact the aggregate data mask important individual differences.

Table 3: Individual Subjects Scoring 4.0 ("good mimic")


    subject   age         score of 4
    children  age  4-11    0
    JHS-#1     14          1 of 8 tokens
    JHS-#5     15          1 of 8 tokens
    JHS-#8     14          1 of 7 tokens
    HS- #3     17          2 of 8 tokens
    HS-#10     17          2 of 7 tokens
    HS-#15     16          3 of 7 tokens
    JHS-#4     15          3 of 7 tokens
    JHS-#6     15          4 of 7 tokens

5.2.3 Score variability across different word tokens.

To determine the level of consistency in mimicry ability from one simple word to another (containing l/r), an analysis was made of each subject's score variation between word-items (e.g., bear vs. chair). A 2-point difference between individual words was interpreted as an indication that a subject's mimicry skill was not uniform. Table 4 shows that this differential between word-tokens was much higher for the children than for the adolescents. The adolescents' mimicry skill, therefore, at whatever the individual level, was collectively more stable across words. The children, by contrast, were much more subject to articulatory inconsistencies. This finding obtained for the entire range of age 4-11, and cannot be attributed to incomplete acquisition of the native phonological system. [-13-]

Table 4: Variability across different word-tokens


     two-point difference
     (n = 11)  age 4-11       9  (.818)
     (n = 09)  age 13-15      2  (.222)
     (n = 08)  age 16-17      1  (.125)

5.2.4. Variability across identical word-tokens.

An individual's ability to mimic was not even necessarily consistent for the same word in different sessions (e.g., a 2 for bear from one session, a 3 for bear another session. The difference was not related to whether a token was from the first or second session (i.e., practice was not a factor). However, these variations were overwhelmingly in the one-point range, as shown in Table 5, and were not considered significant, except that the variability predominated in the children, just as the 2-point differential for multiple same- word tokens had. For 7 of the 11 children, there was a 1-point difference for two tokens of the same word. That tendency declined with age in the adolescent group: 5 subjects of 9 in the age 13-15 group, and 2 subjects of the 8 in age 16-17 group had a one-point difference between identical tokens from different sessions.

6. Conclusion

In this study, although the adolescent subjects proved superior to the children in mimicry, the advantage for the older subjects was not universal. Only one subject (age 15) of the 17 adolescents received better than 50 percent for the 4.0 scores on word tokens. This suggests that age is not well correlated with mimicry skill. Rather, mimicry skill appears to be a distinct talent that is distributed unpredictably across any given age population. This conclusion is reinforced by the fact that the children in this study exhibited no mimicry skill and their imitations of words were much more inconsistent across tokens than the adolescents. If the mimicry skills of the children here were an indicator of long-range potential in L2 accent, one would have to predict a substantial "foreign" accent in L2 for all the children, even with optimal exposure to the L2 at an early age. Yet the statistical evidence from long-range L2 studies (Section 3), would predict that these same children would be more likely than the adolescents to attain native-like L2 pronunciation under normal learning conditions.

The major concern of this experiment was the relationship between mimicry skills and neural plasticity. Since children do have the type of neural plasticity that makes it possible to acquire native- like L2 accent and syntax in the long run, the absence of mimicry skill among the children strongly suggests that there is no relationship between mimicry skill and the neural plasticity. Instead, individuals exhibiting mimicry skill are likely to have that ability throughout the lifetime, but it is very doubtful that [-14-] there is any connection between such skill and either neural plasticity or a vestigial language acquisition device. In late L2 learners, exceptional ability in L2 pronunciation does not guarantee the same ability for L2 syntax (or vice versa). It is much more common for exceptional older L2 learners to exhibit mastery of either L2 pronunciation or the syntax, but not both. By contrast, most children acquire both pronunciation and syntax, and probably because of a qualitative difference in neural plasticity in the case of children.

It has been proposed that the conditions of the experiment can affect children's mimicry skills or L2 acquisition. Lowenthal & Bull (1984), for instance, suggested that the performance of children is boosted significantly by an encouraging environment in the experimental situation. The authors compared the performance of the children in their study, in which the children repeated words modeled on an audiotape and did poorly, with an earlier study by Tahta et. al (1981b), in which the children repeated words after a prompt from an encouraging tester, in which case they outperformed the older subjects. It is true, as Lowenthal and Bull point out, that the mimicry task is quite alien to the youngest children. With this in mind, in the experiment described here a considerable amount of time was devoted to making the children comfortable with the task, and the foreign words were demonstrated with familiar objects (or drawings, in the case of the older children). Despite that strategy, the performance of the children age 4-11, individually and as a group, was significantly lower than the performance of the adolescents. This suggests that supportive conditions may simply be necessary for obtaining adequate data from children, but do not in themselves improve the children's pronunciation of foreign words. If mimicry skill is an individual talent, then it is unlikely that testing conditions will significantly heighten or diminish that ability.

If mimicry skill is not related to neural plasticity in children, it is probably not possible to invoke neural plasticity as a basis for superior performance by adolescents and/or adults. Claims of neural plasticity as a basis for successful mimicry in older L2 learners (Flege, 1987; Klein,1986; Neufeld, 1978,1979) should therefore perhaps be reconsidered. In addition, if typical adolescent or adult L2 learners can perform mimicry tasks better than children, then it should be the case that their performance will be consistently superior from study to study. Where such claims are based either on rehearsed mimicry tasks or talented mimics, it is difficult to establish a relationship with ultimate L2 pronunciation skill. Regarding rehearsed tasks, as Seliger (1978) has pointed out, rehearsed phonological drills largely result in improved performance of isolated target sounds and words, but do not contribute to the integration of the practiced sounds in normal L2 production. Regarding the use of talented mimics, older learners with such talent will naturally compare favorably with children who [-15-] lack this ability, but not because of neural plasticity. Therefore, the argument frequently promoted in such studies that all or most language learners can achieve native-like pronunciation in L2 (because of neural plasticity) is misleading, both scientifically and pedagogically. Talented older L2 learners appear to have a cognitive profile that differs from the typical learner (Novoa, Obler and Fein, 1988; and Obler, 1982). If typical older learners are compared with typical child learners over the long term, the plasticity issue becomes much more obvious.

The possibility of a more general underlying neural plasticity serving a number of skills and mental faculties deserves consideration in second language research. There may be a relationship between the few late L2 learners who achieve native- like L2 pronunciation or syntax and late learners who achieve high skill levels in the types of musical instruments, sports, dance and other skills which require very subtle motor coordination. In such cases, successful learners who begin learning the skill after the optimal learning period in childhood succeed because they are endowed with a distinct talent, and are therefore able to compensate for delayed training (although, with the late start, they do not necessarily reach their original potential). As brain studies advance, making it possible to investigate complex cognitive faculties such as language, we will reach a far better understanding about different types of learning and neural plasticity. At present, however, little is known about neural plasticity in human cognitive functions. The best evidence for ultimate proficiency in second language pronunciation is therefore the long-range comparison of typical children and older learners. That evidence clearly indicates that the typical child has the advantage over the typical older learner in the ultimate acquisition of L2 pronunciation --with or without good mimicry skills at the outset.

NOTES

[1] The critical period for first language development is thought to be between age 8 months and 6 years (Gazzaniga, 1992), but this range assumes early exposure with development continuing to age six. Primary exposure delayed to even age 3-4 may handicap acquisition. Newport (1984), for example, found an asymptotic effect in congenitally deaf children over age four whose exposure to sign language was delayed.

[2] This is not necessarily an indication of a specific language acquisition device. Certain animals such as chinchillas can also discriminate phonemes (see Edelman, 1987; and Kuhl, 1981). Edelman suggests that humans have evolved an exceptionally sensitive auditory system that is exploited for language. [-16-]

[3] In a case study of late language onset described by Vargha-Khadem et al., (1997), late acquisition was characterized in part by a phonological impairment similar to "foreigner speech": vash for wash, etc. Interestingly, the "foreign accent" does not occur in normal first language (including simultaneous bilingual development), but it does, of course, occur in regular second language acquisition, even by very young children. This type of phonemic distortion in late L1 was also reported in Curtiss's (1977) study of Genie. In both the Vargha-Khadem et al., (1997) and Curtiss (1977) studies, the children did not go through the usual phonological stages, and both children were right hemisphere-based for L1. This suggests a different learning mechanism from that associated with normal first language acquisition, perhaps in line with the type of learning differential described by Goldman-Rakic (1992).

[4] These are necessary generalizations. Obviously there are individual differences in the distribution of first language in the brain, but most differences are essentially trivial. For a discussion, see Segalowitz and Bryden (1983).

[5] See Greenough's (1986) description on the effects of environmental enrichment on the rat brain. See also Black (1994) for a description of neurochemical changes that affect different types of memory.

[6] The central relationship of the motor system in language development is elaborated in J. Brown (1982). For a discussion of relevant memory systems, such as motor memory, see Kosslyn and Koenig (1992).

[7] There is some speculation by Brown et al. (1991) as to whether the neural plasticity associated with first language learning remains in attenuated form for exploitation in early second language acquisition. Of course, if the two neural conditions are the same, the question arises as to why delayed acquisition of first language at this same age produces very poor results, while second language acquisition is very successful.

[8] The long-range acquisition advantage has been demonstrated in syntax as well, of course (see Fathman, 1975; Seliger, Krashen & Ladefoged 1975; Johnson and Newport, 1989). There is dissociation between L2 acquisition of syntax and L2 phonological acquisition, although children have the long range advantage in both cases. Evidence for the dissociation can be seen in the many cases where adolescents or adults achieve native-like L2 syntax but not pronunciation, and vice versa, with the tendency toward the former rather than the latter. Joseph Conrad and Henry Kissinger are exemplars of superior skill in L2 syntax, with fossilization in L2 pronunciation. [-17-]

[9] Flege rejects the notions of a critical period for (first) language and a sensitive period for second language, and bases his argument on the fact that much critical period research is based on animal studies. Although it is true that the need for direct, intrusive experiments limits research to animal studies, it is entirely legitimate to extrapolate from such studies to various types of human development. The fundamental ontogenetic mechanisms of neural systems are quite similar across species. On the issue of lateralization, Patkowski (1990) has correctly pointed out that lateralization is not directly relevant to the notion of the critical period hypothesis (CPH) for language; rather the concern in CPH is with neuronal plasticity.

References

Ager, J., Ernhart, C.B., Martier, S., Greene. T., and Sokol, R. (1990). Prenatal alcohol exposure and language development. Alcoholism: Clinical and Experimental Research, 14, 937-945.

Asher, J., and Garcia, R. (1969). The optimal age to learn a foreign language. The Modern Language Journal, 553, 334-41.

Black, I.R. (1994, paper edition.) Information in the brain.Cambridge, MA: MIT Press

Bongaerts, T., van Summeren, C., Planken, B., & Schils, E. (1998). Age and ultimate attainment in the pronunciation of a foreign language. Studies in Second Language Acquisition, 19, 447- 465.

Brown, J.W. (1982). Hierarchy and revolution in neurolinguistics. In M.A. Arbib, D. Caplan, and J.C. Marshall (Eds.), Neural models of language processes (pp. 447-467). New York: Academic Press.

Brown, M.C., Hopkins, W.G., and Keynes, R.J. (1991). Essentials of neuronal development. Cambridge: Cambridge University Press.

Candland, D.H. (1993). Feral children and clever animals. New York: Oxford University Press.

Cochrane, R. M. (1980). The acquisition of /r/ and /l/ by Japanese children and adults learning English as a second language. Journal of Multilingual and Multicultural Development, 1 (4), 331-360.

Cochrane, R. M., and Sachs, J. (1979). Phonological learning by children and adults in a laboratory setting (Part 2). Language and Speech, 22, 145-149. [-18-]

Cotman, C.W., and Lynch, G.S. (1990). The neurobiology of learning and memory. In P.E. Eimas & A.M. Galaburda (Eds.), The neurobiology of learning and cognition (pp. 201-241). Cambridge, MA: MIT Press (A Bradford Book).

Cronly-Dillon, J., and Nona, S.N. (1988). Changes in cytoskeletal elements during postnatal development of cerebral cortex. In A. Peters & E.G. Jones (Eds.), Cerebral cortex, Vol. 7: Development and maturation of cerebral cortex (pp 273-308). New York: Plenum Press.

Curtiss, S. (1977). Genie: A psycholinguistic study of a modern- day "wild child". New York: Academic Press.

Diamond, M.C., Rosenzweig, M.R., Bennett, E.L., Lindner, B, & Lyon, L. (1972). Effects of environmental enrichment and impoverishment on rat cerebral cortex. Journal of Neurobiology, 3, 47-64.

Edelman, G. M. (1987). Neural Darwinism. New York: Basic Books.

Ekstrand, L. (1976). Age and length of residence as variables related to the adjustment of migrant children, with special reference to second language learning. In G. Nickel (Ed.), Proceedings of the Fourth International Congress of Applied Linguistics, Vol. 3 (pp. 179-198). Stuttgart: Hochschulverlag.

Eimas, P. (1975). Auditory and phonetic coding of the cues for speech discrimination of the r-l distinction by young infants. Perception and Psychophysics,18, 341-347.

Ellis, R. (1985). Understanding second language acquisition. Oxford: Oxford University Press.

Fathman, A. (1975). The relationship between age and second language productive ability. Language Learning, 25 (2), 245-253.

Flege, J.E. (1987). A critical period for learning to pronounce foreign languages? Applied Linguistics, 8 (2), 162-177.

Gazzaniga, M.S. (1994, paper edition). Nature's mind. Hammondsworth, Middlesex: Penguin Books.

Gazzaniga, M.S. (1989). Organization of the human brain. Science, 245: 947-952.

Genesse, F. (1988). Neuropsychology and second language acquisition. In L. Beebe (Ed.), Issues in second language [-19-] acquisition: Multiple perspectives (pp. 81-112). Cambridge, MA: Newbury House.

Goldman-Rakic, P.S. (1992). Working memory and the mind. Scientific American, 267 (3) (September), 110-117.

Goldman-Rakic, P.S. (1982). Organization of frontal association cortex in normal and experimentally brain-injured primates. In M. A. Arbib, D. Caplan, & J.C. Marshall (Eds.), Neural models of language processes (pp. 469-483). New York: Academic Press.

Greenough, W.T. (1986). What's special about development? Thoughts on the bases of experience-sensitive syaptic plasticity. In W.T. Greenough & J.M. Juraska (Eds.), Developmental neuropsychobiology (pp. 387-407). New York: Academic Press.

Greenough, W.T., Black, J.E., and Wallace, C.S. 1996 (1993). Experience and brain development. In M.H. Johnson (Ed.), Brain development and cognition (pp.290-319). Cambridge, MA: Blackwell Publishers, Ltd.

Humes-Bartlo, M. (1993). Variation in children's ability to learn second languages. In K. Hyltenstam and L. K. Obler (Eds.), Bilingualism across the lifespan (pp. 41-54). NY: Cambridge University Press.

Jacobs, B. (1988). Neurobiological differentiation of primary and secondary language acquisition. Studies in Second Language Acquisition, 10, 303-337.

Johnson, J.S., and Newport, E.L. (1989). Critical period effects in second language learning: The influence of maturational state on the acquisition of English as a second language. Cognitive Psychology, 21, 60-99.

Kim, K.H.S., Relkin, N.R., Lee, K-M, and Hirsch, J. (1997). Distinct cortical areas associated with native and second languages. Nature, 388 (July 10), 171-174.

Klein, W. (1986). Second language acquisition. Cambridge, England; New York; Melbourne, Australia: Cambridge University Press.

Kosslyn, S.M., & Koenig, O. (1992). Wet mind: The new cognitive neuroscience. New York: The Free Press.

Kuhl, P.K. (1980). Perceptual constancy for speech-sound categories in early infants. In G.H. Yeni-Komshian, J.F. Kavanagh, & C.A. Ferguson (Eds.), Child phonology, vol.2: Perception (pp. 41-66). New York: Academic Press. [-20-]

Kuhl, P.K. (1981). Discrimination of speech by nonhuman animals: Basic auditory sensitivities conducive to the perception of speech-sound categories. Journal of the Acoustical Society of America, 70, 340-349.

Kuhl, P.K. (1985). Categorization of speech by infants. In J. Mehler & R. Fox (Eds.), Neonate cognition: Beyond the blooming, buzzing confusion (pp. 231-262). Hillsdale, New Jersey: Erlbaum.

Lauder, J.M., & Krebs, H. (1986). Do neurotransmitters, neurohumors, and hormones specify critical periods? In W.T. Greenough & J.M. Juraska (Eds.), Developmental neurobiology (pp. 119-174). New York: Academic Press.

Lenneburg, E. (1967). Biological foundations of language. New York: Wiley.

Loewenthal, K., and Bull, D. (1984). Imitation of foreign sounds: What is the effect of age? Language and Speech, 27 (1), 95-98.

Long, M.H. (1990). Maturational constraints on language development. Studies in Second Language Acquisition, 12 (3), 251-285.

Newport, E.L. (1984). Constraints on learning: Studies in the acquisition of American Sign Language. Papers and Reports on Child Language Development, 23, 1-22.

Neufeld, G. (1978). On the acquisition of prosodic and articulatory features in adult language learning. Canadian Modern Language Review, 34, 163-174.

Neufeld, G. (1979). Towards a theory of language learning ability. Language Learning, 29, 227-241.

Neufeld, G. (1980). On the adult's ability to acquire phonology. TESOL Quarterly, 14, 285-98.

Novoa, L., Obler, L., and Fein, D.A. (1988). A neuropsychological approach to talented second language acquisition: a case study. In L.K. Obler & D. Fein (Eds.), The exceptional brain: The neuropsychology of talent and special abilities (pp. 294-302). New York: Guilford Press.

Obler, L. (1982). Exceptional second language learners. In L. Obler & L. Menn (Eds.) Exceptional language and linguistics. New York: Academic Press. [-21-]

Ojeman, G., and Whitaker, H. (1978). The Bilingual Brain. Archives of Neurology, 35, 409-412.

Oyama., S. (1978). A sensitive period for the acquisition of a phonological system. Journal of Psycholinguistic Research, 5, 261-283.

Parnavelas, J.G., Papadopoulos, G.C., & Cavanagh, M. (1988). Changes in neurotransmitters during development. In A. Peters & E.G. Jones (Eds.), Cerebral cortex, Vol. 7: Development and maturation of cerebral cortex (pp. 177-209). New York: Plenum Press.

Patkowski, M.S. (1982). The sensitive period for the acquisition of syntax in a second language. In S.D. Krashen, R.C. Scarcella, & M.H. Long, (Eds.), Child-adult differences in second language acquisition (pp. 52-63). Rowley, MA: Newbury House.

Patkowski, M.S. (1990). Age and accent in a second language: A reply to James Emil Flege. Applied Linguistics, 11 (1), 73-89.

Schmid, E. (1986). A comparative study of children's and adults' acquisition of tone accents in English. Language Learning, 36, 185-210.

Scovel, T. (1969). Foreign accents, language acquisition, and cerebral dominance. Language Learning, 19, 245-253.

Scovel, T. (1988). A time to speak: A psycholinguistic inquiry into the critical period for human speech. Cambridge, MA: Newbury House.

Segalowitz, S.J., and Bryden, M.P. (1983). Individual differences in hemispheric representation of language. In S.J. Segalowitz (Ed.), Language functions and brain organization (pp. 341- 372). New York:Academic Press.

Seliger, H. (1978). Implications of a multiple critical periods hypothesis for second language learning. In W.C. Ritchie (Ed.), Second language acquisition research (pp.11-19). New York: Academic Press,.

Seliger, H., Krashen, S., and Ladefoged, P. (1975). Maturational constraints in the acquisition of second language accent. Language Sciences, 36, 20-22

Shatz, C. J. (1992). The developing brain. Scientific American, 267 (3)(September), 61-67. [-22-]

Shepherd, G.W. (1988, 2nd ed.). Neurobiology. New York: Oxford University Press.

Snow, C., and Hoefnagel-Hohle, M. (1977). Age differences and the pronunciation of foreign sounds. Language and Speech, 20, 357-365.

Springer, S.P., & Deutsch, G. (1985). Left brain, right brain. New York: W.H. Freeman and Company.

Tahta, S., Wood, M., & Lowenthal. (1981a). Foreign accents: factors relating to the transfer of accent from the first to the second language. Language and Speech, 24(4), 265-272.

Tahta, S., Wood, M., & Lowenthal. (1981b). Age changes in the ability to replicate foreign pronunciation and intonation. Language and Speech, 24 (3), 363-372.

Thompson, I. (1991). Foreign accents revisited: The English pronunciation of Russian immigrants. Language Learning, 41 (2), 177-204.

Trevarthen, C. (1987). Brain development. In R.L. Gregory (Ed.), Oxford companion to the mind (pp.101-110). New York: Oxford University Press.

Vargha-Khadem, F., Carr, L.J., Isaacs, E., Brett, E., Adams, C., & Mishkin, (1997). Onset of speech after left hemispherectomy in a nine-year old boy. Brain, 120, 159-182.

Walsh, T.M., & Diller, K.C. (1981). Neurolinguistic considerations on the optimum age for second language learning. In K.C. Diller (Ed.), Individual differences in language learning aptitude (pp. 3-21). Rowley, MA: Newbury House.

Werker, J.F., and Tees, R.C. (1984). Cross-language speech perception: Evidence for perceptual reorganization during the first year of life. Infant Behavior and Development, 7, 49-63.

White, L., & Genesee, F. (1996). How native is near-native? The issue of ultimate attainment in adult second language acquisition. Second Language Research, 12, 233-265.

Wode, H. (1989). Maturational changes of language acquisitional abilities. In C. Gass, D. Madden, S. Preston, & L. Selinker (Eds.), Variation in second language acquisition, Vol. 2: Psycholinguistic Issues (pp. 176-185). Clevedon, Avon: Multilingual Matters.[-23-]

Yamada, J., Takatsukua, S. Kotabe, N., & Kuruse, J. (1980). On the optimum age for teaching foreign vocabulary to children. International Review of Applied Linguistics, 28, 245-247.

About the Author