1
Birth Language Attrition/Retention and
Relearning: A Longitudinal Study on
Perception of Mandarin Chinese Tones
among School Age Chinese Adoptees in the
Netherlands
Wen Cui Zhou (Selina)
S1939246
MA in Applied Linguistics
Faculty of Liberal Arts
University of Groningen
Supervisors:
Dr. Monika S. Schmid
Dr. Wander Lowie
2
Contents
Abstract……….3
Introduction…….………..4-6 Background.………..………...7-18 1. Language development from a DST perspective ………7
1.1 Language development is highly context dependent……….…………...8
1.2 Language development is significantly based on input and usage ………9
1.2.1 Language development depends on input………9
1.2.2 Language development relies on real life use……….………..9
1.3 Conclusion………...10
2. First language attrition………10-11 2.1 Birth language attrition research in international adoptees……….11
2.2 The rationale of the present study………...12
2.2.1 Young age groups of adoptees should be tested……….12
2.2.2 Relearning investigation should be conducted together with attrition……..12-13 2.2.3 More different language systems should be investigated………13
2.3 Conclusion………..14
3. Child’s first language development in phonology……….14-15 3.1 Categorical perception………15-16 3.2 Birth language attrition in phonology………16-17 3.3 Mandarin Chinese tonal system……….17
3.3.1 Acoustic characteristics of Mandarin Chinese tones……….17-18 3.4 Conclusion………...18
Statement of purpose………..19
Research questions………..19
3
ABSTRACT
A recent proposal for the theoretical interpretation of first language (L1) and second language (L2) interaction that results from the sudden change of language environment in international adoptees claims that such children experience a total loss of their birth language (Pallier et al., 2003; Ventureyra et al., 2004). The current study challenges such theoretical proposal by examining perception of Mandarin tones among school age Chinese adoptees, young Dutch learners of the language, and Chinese children who were born and are raised in Chinese families in the Netherlands. The present study involves data collection sessions spaced three and a half months apart, namely an attrition investigation (pre-test) and a relearning investigation (post-test). According to the data from the pre-test, the adopted Chinese children did show better perception of Mandarin tones compared to the young Dutch learners. After two months’ (six hours in total) training on recognizing and distinguishing the Mandarin tones, most participants showed different extents of improvement in their scores, but no such improvement was found for the Chinese adoptees. However, an advantage in perceiving and producing the Mandarin tones in the Chinese adoptees has been observed during the training process. Therefore, the writer contends that an account which includes a more maturational and comprehensive approach can better account for language replacement data, and is therefore preferable for future research.
4
INTRODUCTION
International adoptees are considered as unique language acquirers. They are similar to L1 learners because they also acquire the new language from child-directed speech in their new homes but differ from other L1 learners because they have early childhood exposure to another language, namely their birth language (Hyltenstam et al., 2009:122). They are similar to L2 learners because they also acquire the new language after one language has been partially established but differ from other L2 learners because they start learning the new language on the basis of the ends of their birth language acquisition and use (Hyltenstam et al., 2009:122). In this paper, to avoid confusion, the term birth language will be used instead of first language where international adoptees are concerned. Chinese adoptees in the Netherlands are typically adopted at a very young age (between a few months and three years old1) before their birth language has been fully acquired. In this case, their birth language is eventually completely replaced by their new language, Dutch. The adoptees speak the new language fluently and apparently natively; but forget or lose their birth language to a very dramatic extent. Therefore, birth language attrition/retention best suits the situation of language deterioration in international adoptees.
Concerning birth language development in international adoptees, two theoretical questions have been raised. The first is whether these international adoptees have totally forgotten or lost their birth language due to long-term non-exposure to the language, or alternatively might they maintain the knowledge and skills but be unable to access it (Ventureyra et al., 2004). The second question is whether relearning may help recover an attrited/forgotten language; for instance, will adoptees have a relearning advantage over learners with no previous experience? These two questions have received a great deal of attention over the past decade.
As for the first question, there are some studies which have shown a complete loss of a first language. But most of these cases have to do with children of immigrants and school-age children from a minority language context who learn a majority language and have very reduced input from the minority L1 (Pan & Gleason, 1986). As far as international adoptees are concerned, a number of studies have also claimed to have found a total loss of their birth language. Glennen & Masters (2002) reported that such populations quickly become unable to use their L1 within a short period of time of non-exposure to the language in the new linguistic environment. Although the length of this period depends on the age at adoption and corresponding L1 development of the individual child, it is important to know that some studies also reported that even children who were adopted at the age of 10 are unable to use their L1 after a relatively short period of time (Gardell, 1979; Hene, 1987, 1993). Moreover, another two recent studies reported similar
1
5
findings. Pallier et al. (2003) used functional Magnetic Resonance Imaging (fMRI) technique and studied 8 adult Koreans adopted during childhood (between 3 and 8 years old) by French families. They found no evidence of remnants of the birth language in the adoptees. Ventureyra et al. (2004) studied 18 Korean adoptees in France and also found that the childhood experience with the birth language (even up until age 9) did not give any benefit in perceiving the attrited language’s phonology later in life. However, the theoretical proposal of a total loss of birth language still needs to be further investigated because there is a possibility that the long-term non-exposure to the language merely leads to an extreme difficulty to access the language.
This implies that an important question to address is whether a specific type of input, namely training aimed at relearning, will reactivate the childhood memory of the birth language. Although present knowledge about the effects of relearning a language that has become inaccessible due to long-term non-exposure is limited, various accounts indicate that a once “forgotten” language maybe “reawakened”, even after a long period of time of non-use (Yukawa, 1997; de Bot & Stoessel, 2000). Additional evidence comes from research on people who overheard or spoke a language during childhood and abandoned it later on. Such studies found that the early exposure gives benefit when relearning certain phonological features (Au et al., 2002; Oh et al., 2003). Findings of this kind lend support to the assumption that remnants of a lost language can be expected to be found primarily at the phonological level. A recent study with international adoptees further confirmed this assumption. Hyltenstam et al. (2009) investigated relearning of the birth language phonetics among adult Korean adoptees in Sweden and found clear advantages in the birth language phonetic discrimination in the Korean adoptees over Swedish learners with no early exposure to Korean. These positive findings give us a reason to believe that formal instruction on the birth language may help reactivate the attrited language. Therefore, the current study intends to contribute a perspective on the two above-mentioned questions by combining an attrition/retention investigation with a relearning investigation among school age Chinese adoptees in the Netherlands.
Since China started the adoption cooperation with the Dutch government in 1992, a growing number of Chinese orphans2 have found new homes in the Netherlands. According to the 2005 annual CBS3 report, almost 60 percent of adopted children in the Netherlands are from China, currently comprising around 6,000 Chinese adoptees residents all over the Netherlands. In spite of the high number of Chinese adoptees little attention has been given to their language development, particularly their birth language development. Although a few studies have addressed this issue (e.g., Nicoladis & Grabois, 2002; Xing Tan & Yang, 2005;Snedeker, Geren & Shafto, 2007), they mainly focused on their new language development. As for their birth
2
Please refer to fiure 1 from the Netherlands' Centraal Bureau voor de Statistiek in the appendix 3
6
language development, namely Mandarin Chinese in the new environment, I found only one presentation (Eng 2006)4, which looks at birth language attrition in Chinese adoptees in the U.S. Therefore, the current study aims to fill in this gap.
In summary, research up till now has shown that birth language attrition/retention in international adoptees is still a new line of thought. It has not been replicated into different groups of population with different language backgrounds. As unique bilinguals, this special group of language learners deserves more attention for their language development, particularly in the birth language. The primary objective of the present pilot study is to establish an initiative stage of birth language attrition/retention in Chinese adoptees in a new linguistic environment. Any findings are expected to serve as both theoretical and methodological foundations for future research in this aspect. The first section discusses three aspects of background literature that this study is based on; namely, language development from a Dynamic Systems Theory perspective, first language attrition and child’s first language development in phonology. The second section describes the design of this study. The third section introduces the results from this study, and the fourth section contains a discussion and conclusion, which explains the possible reasons for the results and summarizes both advantages and pitfalls of this study. The last sections consist of acknowledgements, references and appendix.
4
7
BACKGROUND
Language development is a complicated and fluctuating process. Every language learner will experience a different developmental trajectory and various changes during the processes of acquiring their L1, L2 and other foreign languages. Some may develop their L1 and L2 smoothly, but some may experience great difficulties, such as learning L2 in a monolingual L1 environment, or acquiring L1 in an L2 environment where little exposure to L1 is available. During the process of acquiring different languages, they may lose or forget either one of the languages (L1 or L2) under certain conditions. The present study focuses on L1 development in an L2 environment where the L1 is not accessible anymore for the studied participants: Chinese adoptees in the Netherlands. The purpose of this study is to investigate whether the Chinese adoptees have completely lost their birth language after long term non-exposure to the language, or alternatively whether the adopted Chinese children might retain their birth language knowledge yet be unable to access it. Moreover, it will be investigated whether relearning can help with recovering a forgotten or attrited language; that is, whether the Chinese adoptees have a relearning advantage over learners with no prior experience to the Chinese language. In the following, the above mentioned concerns will be discussed from three theoretical perspectives; language development from a dynamic systems theory (DST) perspective, first language attrition and, child’s first language development in phonology.
1. Language development from a DST perspective
8
system. However, as a dynamic system, language does not develop without its necessary resources, such as its ambient linguistic context, meaningful input and real life use. Detailed discussions are presented in the following sections.
1.1 Language development is highly context dependent
From a dynamic systems point of view, language development is never smooth; it is full of changes. Among the most important changes in language attrition are language-related major life events; such as the change of linguistic environment, namely immigration and adoption and so forth. When an individual moves or is adopted to another country, his language environment is suddenly changed because he lives in an immersed new language (L2) environment where there is little or no exposure and input for their first language available anymore. Both his first language and second language will change in different ways. As de Bot (2007: 58) noted “first language attrition is basically the interaction between a decline of the availability of one language and the development of another language”. Migration to an environment in which another language is spoken and accordingly the amount of use and contact with the L1 is reduced is the most obvious major life event for L1 attrition; such as Chinese adoptees were immigrated into the Netherlands in their early years.
9
1.2 Language development is significantly based on input and usage
De Bot, Lowie, & Verspoor (2005:29) claimed that “the cross-linguistic interaction between different languages is dynamic and related to input and use”. Evidence from an early cross-sectional study conducted by Hart & Risley (1995) has shown that children’s linguistic development was strongly associated with the amount of the input and use in their daily life. It suggests that language input and use are crucial for language development, which will be further discussed in the following sections.
1.2.1 Language development depends on input
Linguists such as Krashen (1985, 1994) and Long (1996) stressed that meaningful input is one of the most important resources for language development. Without continuous input, acquisition and retention of any language knowledge and skills are impossible (Verspoor, Lowie, & de Bot, 2008: 62). Studies from Tomasello (2000) and Diesel & Tomasello (2000) in Child L1 development have proved that there is a strong relation between a child’s input and its language development, particularly in early L1 acquisition. Moreover, from a DST perspective, language input is not static. It interacts with the language system itself and other factors, such as linguistic context all the time and it changes over time. Just as Verspoor, Lowie, & de Bot (2008: 63) claimed “the same input will not be processed in the same way by the same learner at different times because the learning process itself is constantly changing in a learning/attrition continuum due to a complex, dynamic interaction between input and all other factors affecting language development”. Finally as a dynamic system, language co-evolves together with its environment over time. It highly relies on resources; with sufficient conditions, it adapts, interacts, and self-organizes continually (Verspoor, Lowie, & de Bot, 2008). Therefore, input is necessary as a resource for both maintenance and growth. But input is also necessary to keep the system going (Van Geert, 1991); “like any other natural system, language has a natural tendency to decline when not used and people may even forget their first language if it is not used” (Verspoor, Lowie, & de Bot, 2008:71). Therefore, besides continuous input, language use is also crucial for language development.
1.2.2 Language development relies on real life use
10
trend of language development is that “non-use leads to loss”, language skills become progressively stable with more use; otherwise, information that is not used regularly will become more difficult to access and eventually disappear (de Bot, Lowie, & Verspoor, 2005). However, “attrition does not happen like an on/off phenomenon” (de Bot, Lowie, & Verspoor 2005:31). Non-use results in gradual difficulties in recovering the language knowledge and skills acquired in the past; after some time, the difficulties in retrieving the language will be greater, language attrition or loss may occur. However, when the language is used more often again, retrieval will be easier and faster again too; therefore, any language system stays in a continuous state of fluctuation, which relies on the amount of exposure to the language (de Bot, Lowie, & Verspoor, 2005:31).
Language knowledge and skills are not only gained through input and daily use but are also practised and consolidated through daily use. Therefore, language input and use are the two indispensable factors for linguistic development across the lifespan.
1.3 Conclusion
To conclude, language development is a dynamic process. It involves a large amount of fluctuating components and is resource dependent. It self-organizes and develops through the interactions among the variables both within and across the linguistic systems over time. However, the interactions and development do not take place without continued input and real life use in the related linguistic context. Without these vital conditions, language knowledge and skills will develop into the negative direction, namely decline or attrition, throughout the lifespan of language development.
2. First language attrition
11
especially starting with the remarkable work done by Seliger & Vago (1991), who organized the first collection of papers in language attrition carrying the term “attrition” instead of “forgetting” or “loss” and narrowed down language attrition research to L1 attrition (Köpke & Schmid, 2004). Since then, L1 attrition was separated from L2 attrition and became an independent subfield in language attrition research.
2.1 Birth language attrition research in international adoptees
A large number of significant studies in first language attrition have been published since the 1990s. These studies have involved different populations in different language environments. As far as the current study is concerned, birth language attrition research in international adoptees was recently brought onto the discussion board. This new line of thought has aroused great interest among a number of language attrition researchers. Generally speaking, individuals who have experienced birth language attrition under such conditions are believed to have completely lost their birth language (Maury, 1999; Pallier et al., 2003). However, “any trace of birth language in childhood attriters can be expected to be rather subtle, maybe even unconscious” (Ventureyra et al., 2004: 2). Several recent studies in this aspect have tried to give us a clear picture on how birth language develops in the adopted language environment.
For instance, Glennen & Masters (2002) studied 130 infants and toddlers adopted from Eastern Europe. They followed the participants from their age at adoption up till age 36 to 40 months and investigated their expressive vocabulary growth and development of four bound grammatical morphemes of their adopted language, namely English. They found most of these children acquired English using the same developmental trajectories as non adopted peers. In the meantime they also found these children’s first language disappear quickly after adoption. Nicoladis & Grabois (2002) looked at aCantonese girl who was adopted at the age of 17 months by an English speaking family in Canada and reported a simultaneous acquisition of English and a rapid loss in both production and comprehension of her birth language as well. Although these studies mainly focused on L2 development of the adopted children, their findings seem to suggest that birth language loss is evident and drastic in international adoptees. However, it still needs to be further investigated whether these children have completely lost their birth language or, alternatively they just have difficulty to access to it due to long-time non re-exposure.
12
adoptees. However, some other studies that included instruction on relearning the birth language found clear advantages, particularly in the phonemes perception among the Korean adoptees compared to foreign learners of the language. Quite a number of efforts have been made, but very controversial findings have been elicited. Therefore, more new attempts on different areas of language have to be made in order to give a thorough picture on birth language attrition/retention in international adoptees.
2.2 The rationale of the present study
Considering the controversial findings from the previous studies, the following three aspects may explain us the reasons: (1) only adult adoptees have been tested, (2) very little research on relearning effects has been conducted, and (3) very few languages have been investigated. In the following sections, these three concerns will be discussed respectively.
2.2.1 Younger age groups of adoptees should be tested
Due to the differences between child and adult language development, the controversial findings from the previous studies cannot be extrapolated among varying adoptee age groups. Since most research found no trace of native language retention in adult adoptees, this may chiefly be due to the long time period of L1 severance. Köpke & Schmid (2004) have suggested that L1 attrition in children is more common and severe compared to adults because attrition more likely reflects difficulty in access to knowledge rather than a true ‘disappearance of knowledge’ from memory. Moreover, evidence from other studies (Ammerlaan, 1996;Yukawa, 1997, 1998; Pelc, 2001) also suggests that age of attrition is positively related to the degree of L1 retention, the younger the children are when their language environments change, the faster and deeper they will attrite (Köpke & Schmid, 2004:10). Finally, as early as the 1990s, de Bot & Clyne (1994:27) have suggested that five to ten years after immigration can be a decisive period of attrition; however, this suggestion still has to be further investigated. Therefore, this paper intends to help fill this gap by examining birth language attrition in six-to-eleven-year old Chinese adoptees who were adopted around age one to two and have been living in the L2 environment for at least five years.
2.2.2 Relearning investigation should be conducted together with attrition
13
provides an advantage, particularly in phonetic perception (Au et al., 2002; Oh et al., 2003; Tees & Werker, 1984). Although the study from Ventureyra et al. (2004:11) failed to find any advantage of early experience with the childhood language (even up until age nine) in availability of the attrited language’s phonology later in life, they did make the interesting observation that one particular participant who had taken a Korean language class for 3 months prior to the experiment performed better on all tests than the remainder of the adoptees. This special case led Ventureyra and her colleagues to suggest that extensive re-exposure can be an important factor in the recovery of phonetic knowledge (Ventureyra et al., 2004:10), “which may be present but difficult to retrieve”.
It is possible that “extensive training, such as that administered by Logan, Lively, & Pisoni (1991) to Japanese learners of English on the /r/ and /l/ contrast, may be able to reactivate the dormant L1 in the adoptees” (Ventureyra et al., 2004:10). Under this concern, Ventureyra and her colleagues are administrating a training program with Korean phonemes to the same Korean adoptees and the results are forthcoming. Recently more studies on birth language attrition/retention have included relearning investigation; such as the studies from Oh et al. (2003) and Hyltenstam et al. (2009). They all applied formal instruction on relearning the birth language, namely Korean, and reported a clear advantage in phonetic perception from the Korean adoptees compared to foreign learners of the language Most recently, Broersma5 (forthcoming) has also included relearning investigation in her research on perception of Korean prosodic patterns and consonant voicing in Korean adoptees in the Netherlands. These recent studies suggest the importance of relearning investigation for birth language attrition/retention research.
2.2.3 More different language systems should be investigated
Up till now research on birth language retention in international adoptees is still quite new and has not been replicated with bigger and different populations. Most available studies at present have focused on European adoptees (Glennen & Masters, 2002; Glennen, S. L., 2007) and especially Korean adoptees (Oh et al., 2003; Pallier et al., 2003; Ventureyra et al., 2004; Hyltenstam et al., 2009; Broersma et al., (in process)). To better analyze birth language retention in international adoptees, more populations should be examined, particularly those who are adopted from a very different language environments. An ideal population for wider investigations is Chinese adoptees because of its big number of population particularly in the Netherlands mentioned in the introduction section and the unique language system they were from, which will be further discussed in section 3.1.
5
14
2.3 Conclusion
As a particular population, international adoptees have been the focus of research in many areas; such as cognition, behavior and educational effects (see, for example, Juffer & Van Ijzendoorn, 2005; Beverly, McGuinness & Blanton, 2008). However, as for their language development, especially their birth language development, such research is still limited. Moreover, international adoptees are normally adopted at a very young age. Take Chinese adoptees in the Netherlands for example, the average age of adoption was 2.5 years old in 20046. At such a young age, these children have been through dramatically changeable life experiences compared to other children raised in normal families. The dramatic change in their early life may have influenced all aspects of their development, but most especially language. Moreover, as a dynamic system, language development of international adoptees may experience more severe changes and variations. Such as the sudden cut-off of birth language input, context and use, and along with an abrupt increase of L2 input, context and use. Finally, due to the existing differences among different language systems, these children must have experienced very different developmental trajectories of their language systems. Therefore, such research in birth language attrition/retention in different adoption populations is highly valued and needed. However, due to the young age at adoption, what birth language constructions have been acquired and can be investigated have to be carefully considered. Such concern will be discussed in the following section.
3. Child’s first language development in phonology
In child language acquisition, phonology/phonetics is among the first linguistic systems to be acquired. The critical developmental period of L1 phonology/phonetics starts around the age of two to twelve months 7(Werker & Tees, 1984; Kuhl et al., 1992, 2004). Prelinguistic studies in infants have showed that they can discriminate many phones used contrastively in human languages (see, e.g., Jusezyk, 1979; Kuhl, 1987), but with their growth, “this ability may be attenuated near the end of the first year of life for certain phonetic contrasts”, especially the ability to discriminate non-native phones (Flege, 1989:1684). This lost ability may result in comprehensive difficulty in foreign languages in their adult life (Flege, 1989; Johansson, 1978). With regard to the current study, the young Dutch learners of Mandarin Chinese may experience more difficulties in distinguishing Mandarin tones as compared to the native Mandarin children and childhood over-hearers of the language. Tees & Werker (1984) studied English-monolingual adults leaning Hindi who used to have regular exposure to Hindi during their first two years of life and found they have better perception of Hindi sounds than others who had no such prior experience with the language. Beside previously mentioned notions, this finding further
6
http://www.cbs.nl/en-GB/menu/themas/bevolking/publicaties/artikelen/archief/2005/2005-1722-wm.htm 7
15
confirmed that early language experience is crucial for phonological development, even if this experience with a language is limited to just the early years, and it indeed appears to have long-lasting benefits for language development later in life8.
3.1 Categorical perception
Regarding to the lost ability to discriminate contrasting phones, a possible explanation is the notion of categorical perception of phoneme boundaries and contrasts. As Wode (1993) pointed out infants are naturally able to perceive auditory sounds categorically and continuously. These two modes of perception interact with each other over time and make it possible for neonates to develop their perceptual categories of the surrounding languages. However, these original perceptual abilities will be constrained with growth and become difficult to access during later stages in life (Borden et al., 1983: 500). This also explains why it is difficult to perceive and produce unfamiliar phones for foreign language learners; such as Japanese speakers have difficulty in discriminating English /r/'s and /l/'s (Miyawaki et al., 1975). Moreover, Flege & Munro’s (1992) study on perception of both adjacent and non-adjacent vowel categories in native English and Spanish speakers showed that categorical status of the vowels in a pair plays an important effect on perceiving their dissimilarities. These preliminary findings have produced good evidence for categorical phoneme perception.
As far as the current study is concerned, the Mandarin Chinese tones are an acoustic feature that has clear categorical boundaries and contrasts. Previous studies have proved that Mandarin Chinese tones are perceived and categorized by native speakers as compared to speakers of non-tonal languages at the contrastive phonological level. As early as the 1970s, Chinese linguists claimed that perception of Mandarin Chinese tones is clearly categorical (Chan, Chuang, & Wang, 1975; Wang, 1976). This finding was later supported by Stagray & Downs (1993). They reported the Mandarin natives “are less sensitive than English listeners to small F0 contour variations in a
same–different discrimination task” (Hallé et al., 2004:6). Although this finding seems counterintuitive, it is compatible with the notion that “Mandarin natives ignore irrelevant tonal variations (i.e., within-category variations) in order to efficiently categorize F0 contours into tones
but English listeners do not experience this process” (Hallé et al., 2004:6). Under this notion, Hallé et al. (2004) conducted a large empirical study on the identification and discrimination of Mandarin Chinese tones by Mandarin listeners and French listeners. They also found that native Mandarin listeners perceive and process tones as contrastive linguistic categories, displaying certain degrees of categorical perception, whereas French listeners perceive tones in a more psychophysical way. With regard to the difference in perception of Mandarin Chinese tones
8
16
between native Mandarin Chinese speakers and other learners of the language, the current paper will pay special attention to the categorical features of the tones, particularly when designing relearning materials and teaching methods for the relearning investigation section.
3.2 Birth language attrition in phonology
However, what may happen to the birth language phonology when the child is removed from this language environment during the critical period? Will perception of the birth language phonology attrite? As Köpke & Schmid (2004) suggested attrition of native language is expected when exposure to the language is reduced; however, it is not known whether the child’s birth language will be completely lost or alternatively, whether the child might retain some hidden knowledge yet be unable to access it; and whether extensive re-exposure or relearning will help in recovering their memory of their attrited language (Ventureyra et al., 2004).
With respect to the present paper, the participants involved are Chinese adoptees in the Netherlands. These children are normally adopted at a very young age, namely around one to three years old as reported by the Netherlands' Centraal Bureau voor de Statistiek (CBS)9. At such a young age, the most possible linguistic system they have been exposed to and might have required should be Chinese phonology, namely Mandarin tones. As far as Chinese adoptees are concerned, they all live at the Child Welfare Institutes (CWIs) of their birth places before adoption. As required by the local and central Civil Affairs Bureaus in China, all the caretakers in the Child Welfare Institutes must obtain the certificate for Mandarin Chinese Level Test and speak Mandarin as their working language at the CWIs; therefore, these adoptees can be assumed to have been exposed to Mandarin Chinese before adoption10. However, due to the poor conditions in China that there were only few caretakers to look after a big number of orphans at each CWI; it is assumed that these adoptees may have received very little language input and exposure through their caretakers. Although some CWIs did arrange foster care for a small group of children, especially in the period of time immediately preceeding adoption to help the children receive more language input (Miller & Hendrie, 2000); their language exposure is still limited. The only linguistic feature that is expected to be familiar to them is the Mandarin Chinese tones. More detailed discussion on the characteristics and perception of Mandarin tones will be presented in the next sections.
9
Most international orphans who are adopted into the Netherlands are at a very young age, the average age in 2004 was 2.5 years3 (Source: CBS). Please refer to Figure 2 from the Netherlands' Centraal Bureau voor de Statistiek in the appendix.
17
3.3 Mandarin Chinese tonal system
The Mandarin Chinese language has a fairly uncommon tonal system. Strictly speaking, there are altogether five different tones, including the neutral tone, as shown in Table 1. Mandarin Chinese tonal system is syllable (Pinyin) and tone combined, and carries rich semantic features. Every Pinyin syllable may have different tones. Every Pinyin syllable with different tones may carry different semantic information. In other words, the same Pinyin syllable with different tones has different meanings and even the same Pinyin syllable with the same tones may have different meanings as well. Table 1 illustrates these differences. These unique features of Mandarin tones make it a very interesting case for investigating birth language attrition/retention and relearning.
Table 1 Mandarin Chinese tones with descriptions, examples and meaning(s). Tone Mark Description Example(s) Meaning(s) 1st macron (¯) High-level pitch mā 妈 mother 2nd acute accent (ˊ) High-rising pitch má 麻/蟆 hemp/toad 3rd caron/háček (ˇ) Low-dipping pitch mǎ 马/码 horse/code 4th grave accent (ˋ) High-falling pitch mà 骂 to scold
Neutral Flat, with no
emphasis. ma 吗
(used at the end of a sentence) to indicate a question
3.3.1 Acoustic characteristics of Mandarin Chinese tones
Previous studies on the acoustic characteristics of Mandarin Chinese tones have shown that differences between tones can be determined physically by different fundamental frequency (F0)
values (Liu, 1924), with F0 height and F0 contour as the main acoustic factors describing
Mandarin Chinese tones (Howie, 1976). Please refer to Figure 4 in the appendix from Wang et al. (2002) for detailed descriptions. Moreover, based on the proposal of a 5-point pitch scale for Mandarin tones from Chao (1948), Xu (1997) illustrated a more detailed descriptive schema, namely Figure 5 in the appendix with an example monosyllable “ma” and defined pitch values (0% (onset), 25%, 50 % , and 100% (offset)). Finally, previous evidence from acoustic and functional analysis on tonal languages showed that the existence of the unique functional connection between F0 characteristics and the supra-segmental features makes it very difficult for
18
3.4 Conclusion
19
STATEMENT OF PURPOSE
The purpose of the current study is three-fold. First, as most previous studies only focused on Korean adult adoptees, the present study intends to enrich the limited literature in birth language attrition/retention among international adoptees by examining perception of Mandarin tones among school age Chinese adoptees in the Netherlands. Second, according to the previous theoretical proposal that international adoptees experience total loss of their birth language (Pallier et al. 2003; Ventureyra, et al. 2004), the present study expects to bring more insight into the theoretical interpretation of L1 and L2 interaction that results from the sudden change of language environment in international adoptees by combining both attrition and relearning investigations. Last but not least, this study aims to inform parents and educational parties who want to “maintain a continuous connection with an adopted child’s birth language and culture by seeking out contexts where the child can achieve that connection as soon as possible after the child’s arrival in the adopted country” (Nicoladis & Grabois, 2002).
RESEARCH QUESTIONS
When investigating birth language attrition/retention in international adoptees, the most intriguing questions are whether this special group of population has completely lost their birth language after long term non-exposure to the language, or alternatively whether they might maintain their birth language knowledge and skills but are difficult to access it due to the long time non-exposure. Finally, whether relearning can help with recovering a forgotten or attrited language; in other words, whether the adoptees’ childhood experience with their birth language may give any benefit when relearning the language compared to learners with no prior experience to the language. Under these above-mentioned concerns, the research questions of the current study have been formulated as the following:
1) Have Chinese adoptees completely lost their childhood memory of Mandarin Chinese tones after a moderate-period (at least 5 years) of non-exposure to the language?
2) Can Chinese adoptees perform as well as Chinese children who have been raised in Chinese families after relearning Mandarin Chinese tones?
20
METHODOLOGY
Design
The present research is a longitudinal study on birth language attrition/retention in school age Chinese adoptees in the Netherlands. In order to answer the research questions properly, altogether 23 children and teenagers of three different groups, namely Chinese adoptees in Dutch families, young Dutch learners of Mandarin Chinese and Chinese children who were born and are raised in Chinese families in the Netherlands have been examined. Two sections of data collection have been conducted, namely pre-test (attrition/retention investigation) and post-test (relearning investigation). The testing and relearning materials mainly focused on perception of the four Mandarin tones. The whole process of data collection took place over a period of three and half months from the middle of February till the beginning of June 2010. Detailed descriptions of the design of this study are presented in the following sessions.
Participants
23 children and teenagers participated in the current study voluntarily. They are residents from Groningen and Leiden in the Netherlands. They have three different nationalities; Dutch, Chinese and Turkish. They are between 4 and 16 years old. None of them have significant hearing impairment, speech problems or other language impairment. They are divided into two groups; the experimental group and control group.
21
combination of the local dialect and Mandarin Chinese. Detailed information of these adoptees is presented in Table 2 in the appendix.
The control group is further divided into two subgroups, namely control group A and control group B, according to the participants’ different nationalities. Control group A involves 6 Dutch children and teenagers and one Turkish girl. They age between 6 and 16 years old. Among the 6 Dutch participants, only one six-year-old girl is monolingual, all other five can speak both Dutch and English fluently. Moreover, two of the Dutch participants had never been exposed to Mandarin Chinese before the tests while the other five have learnt the language for 0.5 years, 1.5 years and 2.5 years, respectively. As for the Turkish girl, she is able to speak Turkish, Dutch and English fluently; but had never been exposed to Mandarin Chinese before the tests. Detailed information about control group A is presented in Table 3 in the appendix.
Control group B consists of 7 Chinese children who were born and have been raised in Chinese families in the Netherlands. They age between 4 and 11 years old. They speak Mandarin Chinese to their parents or babysitters at home. The majority of them have attended the Chinese school of Groningen from 0.5 years to 3.5 years before the tests except the youngest girl, Liya, who is only four years old and does not meet the age requirement for attending the Chinese school in Groningen. She does, however, speak fluent Mandarin Chinese with her parents, elder sister and babysitter. Therefore, they are considered to have been living in a comparatively natural Mandarin language environment. The writer of this paper has been teaching these Children Mandarin Chinese for five months. She has observed the Mandarin Chinese proficiency of these children and considered it as native-like. Detailed information about control group B is presented in Table 4 in the appendix.
Materials
22
Second, with regard to the language abilities test, altogether three types of perception tests on Mandarin Chinese tones11 were used in both attrition and relearning investigations: (1) an AX tone discrimination test (AX), (2) an AXB tone discrimination test (AXB), and (3) a tone recognition test (TRT). In particular, AX and AXB tests have been used in the initial attrition investigation while all three types of tests, AX AXB and TRT have been used in the relearning investigation. Besides a new set of AX, AXB and TRT tests in the relearning section, the same pre-AX test was used too. The purpose is to obtain a clear picture of each participant’s improvement after receiving the instruction on Mandarin tones. Detailed information about the tests has been presented in section IV, tests materials, in the appendix.
(1) AX tone discrimination test: The stimuli in the AX test are four Mandarin Chinese tones (tone 1, tone 2, tone 3 and tone 4). Sixteen pairs of Mandarin Chinese Pinyin syllables under four tones were designed and randomly arranged (see Table 5, appendix). The participants listened to the stimuli twice and determined whether the second Pinyin syllable (dependent variable X) has the same tone as the first one (independent variable A) or not. The subject might, for example, be presented with the sequence “bā, bí”. The second Pinyin syllable X, namely “bí” has the second tone, but the first Pinyin syllable A, namely “bā” has the first tone, so that the correct response would be to indicate that they are different.
(2) AXB tone discrimination test: The stimuli in the AXB test are also the four Mandarin tones. Four combinations (ABB, AAB, BAA, and BBA) of three Pinyin syllables (ma, mi and ma) with six A-B pairs were designed. Altogether twenty-four combinations of the stimuli were randomly arranged (see Table 6, appendix). The participants listened to the stimuli twice and determined whether the middle Pinyin syllable (dependent variable X) had same tone as the first Pinyin syllable (independent variable A or B) or the third Pinyin syllable (independent variable B or A). The subject might, for example, be presented with the sequence “mā, mí, má (ABB)”. The middle Pinyin syllable X, namely “mí” has the second tone, the first Pinyin syllable A, namely “mā” has the first tone, and the third Pinyin syllable B, namely “má” has the second tone, so that the correct response would be to indicate that the middle Pinyin syllable X has the same tone as the third Pinyin syllable B.
(3) Tone recognition test (TRT): The TRT test, consisting of thirty Mandarin Pinyin syllables with four different tones (see Table 7, appendix), was administrated only in the relearning investigation. All four tones were randomly allocated to different Pinyin syllables. The participants heard every stimulus twice and determined which tone the stimulus they just heard has; such as, jiāng, which has the first Mandarin tone.
11
23
Recordings
All the stimuli for the three types of tests were recorded using a solid state recorder in a quiet room. Two Mandarin Chinese speakers, one male and one female, read all the stimuli. They are both from the north of China, speak standard Mandarin Chinese, have gained the certificate of Standard Mandarin Chinese in China and work as Mandarin Chinese teachers at the Chinese school of Groningen, the Netherlands. All the stimuli were recorded twice and played twice for the participants. There was five seconds’ silence between every stimulus. The stimulus was recorded at a sampling rate of 44100Hz. Moreover, considering the young age of the most participants, the stimulus was read at a comparatively slower speed than normal. Furthermore, some stimuli were repeated by the tester when required by the participants12. Finally the whole testing process was also recorded with a solid state recorder and conducted in a quiet room where only the tester (the writer herself), the participant and the translator were present.
Procedures
The present study involves two sections of data collection: birth language attrition/retention (pre-test) and relearning (post-(pre-test).
(1) Pre-test. This section was completed by all 23 participants, namely 9 Chinese adoptees, 7 young Dutch learners and 7 Chinese children. Both AX and AXB tests had been conducted consecutively on the same day with each participant in a quiet room. Before each test, about ten minutes were spent on detailed explanations and instructions on how to answer the questions, together with five practice items with feedback for each participant. The two tests were conducted consecutively. After finishing the five practice items, the participants started the experimental trials. For each test, they were asked to repeat the Mandarin Chinese sounds they hear first, and then answer the questions as required for each test. All the stimuli were presented on an laptop computer. The participants listened to the stimuli over the laptop speaker13 at a comfortable hearing level. Each participant’s responses were recorded with a solid state recorder.
(2) Post-test. This section was only completed by half of the participants. They are 2Chinese adoptees, 3 young Dutch learners and 6 Chinese children. All these participants received instruction on Mandarin tones from the tester and took all four tests in the post-test section. The post-test was conducted after six hours’ extensive training on distinguishing and recognizing Mandarin Chinese tones over a period of two and a half months (6 hours in total) with each participant. During the training session, Mandarin Chinese Pinyin and tones were formally
12 As commonly acknowledged that children have very short concentration. During the tests, they may be distracted by any surrounding factors. Therefore repetition of stimuli by the tester when required is necessary in order to get an answer to each question.
13
24
introduced and taught. The main focus of the training session is to enable the participants to produce, distinguish and recognize different Pinyin syllables with different tones. Moreover, some simple daily communicative Mandarin Chinese expressions, short sentences and phrases were also included in order to help them better understand how Mandarin tones function in real life language context. Detailed training materials are given in the appendix. Once the training session was finished; four tests, namely same pre-AX, new post-AX and AXB, and TRT tests, were consecutively carried out with each participant individually on the same day14. Finally all four tests were delivered under the same requirements and conditions as that in the attrition investigation.
Operationalisations
For all tests, accuracy of performance was scored. The scores of each test for each participant are interval variables and have been used for quantitative analysis. The alpha value has been set at 0.05. In this study, the resulting data has been compared in the following three ways; (1) comparison among groups in the attrition investigation, (2) comparison of improvement between pre-test and post-test within groups, and (3) comparison among groups in the relearning investigation. Detailed information has been presented in the results sections.
25
RESULTS
The results15 are analyzed and presented in the following three ways as described in the operationalisations section.
I. Comparison among groups in the attrition investigation. The results on the pre-AX test
are presented in Figure 25. As expected, the 7 native Chinese controls scored almost at ceiling, M=13.71 (out of 16). The 9 Chinese adoptees scored higher than the 7 young Dutch learners, M= 12 and 10.57, respectively. But the difference among groups was found to be non-significant by a one way ANOVA test, F (2, 20) =2.357, P=0.120 > 0.05. A Turkey’s Post-hoc analysis revealed that the three groups did not perform significantly differently from each other. But the variation among the native Chinese controls (SD=1.60, range=12-16) was just marginally greater than among the Chinese adoptees (SD=1.58, range=10-14). As the data above suggests, possible remnants of the birth language among the Chinese adoptees could be a relevant factor explaining these results.
Given the results from the pre-AX test, one may expect similar results from the pre-AXB test. The results are presented in Figure 26. The native Chinese controls scored almost at ceiling again, M=21.57 (out of 24). The Chinese adoptees scored higher than the young Dutch learners, M= 19.89 and 18.86, respectively. But again no significant difference among groups was found by a one way ANOVA test, F (2, 20) = 0.695, P=0.511> 0.05. In terms of variation; the pre-AXB test differs from the pre-AX test. The variation was bigger among the Chinese adoptees (SD= 4.57, range=12-24) than among the native Chinese controls (SD=3.46, range=14-24) but similar between the Chinese adoptees and the young Dutch learners (SD= 4.85, range=12-24). As suggested above, a certain degree of birth language attrition among the Chinese adoptees could account for the results.
15
26
Figure 25 represents the difference between the three groups in pre-AX test, and showing the maximum, minimum, mean scores and upper quartile and lower quartile for each group.
Figure 26 represents the difference between the three groups in pre-AXB test, and showing the maximum, minimum, mean scores and upper quartile and lower quartile for each group.
II. Comparison of improvement between test and post-test within groups. The same
pre-AX test was used in the post-testing phase in order to elicit improved perception of Mandarin tones in three groups after receiving training. The results are presented in Figure 28. Both the native Chinese controls and young Dutch learners improved their scores by M16=1.7 and 2.3 respectively. However, the Chinese adoptees showed a decrease in their scores in the same pre-AX test by M17=-0.5. Repeated measures ANOVA revealed that no significant effects were found
16
M represents the improvement value between the same pre-AX and pre-AX tests. 17
27
in the tests, F (1, 8) =4.78, P=0.06 > 0.05, which indicated that each group performed similar in same two AX tests. However, with a significance value of just above 0.05, one can speak of a trend. Moreover, there was no significant interaction between the independent variable and dependent variable: three groups of participants and the same two AX tests, F (2, 8) =1.996, P=0.198 > 0.05. But there were significant differences between groups, F (2, 8) =16.282, P=0.002 < 0.05. A Turkey’s Post-hoc analysis showed that only the difference between control group A and control group B was significant at P=0.001< 0.05. The experimental group did not differ significantly from any other groups, but the significance value between the experimental group and control group B was at 0.065, which is just above the p value, therefore, we may assume there is a trend of particular significance.
Figure 28 represents the different mean values of both pre-AX test and the same pre-AX test for three groups, and showing the general trend of development through two tests for each group. The factor represents the two tests, namely 1 is for pre-AX test and 2 for the same pre-AX test.
III. Comparison among groups in the relearning investigation. Three new tests were used in
this section. Firstly, a new post-AX test. The results of the new post-AX test are presented in Figure 29. As expected, the 618 native Chinese children scored at ceiling, M=15.33 (out of 16). But different from the pre-test, the 319 young Dutch learners scored slightly higher than the 220 Chinese adoptees this time, M= 12.33 and 12.0, respectively. A one-way ANOVA test showed that the three groups performed significantly differently from each other, F (2, 8) =6.727,
18 As mentioned in the methodology section that the youngest Chinese girl Liya dropped off during the relearning section.
19 As mentioned in the methodology section that only three Dutch children followed the training section with the tester. 20
28
P=0.019 < 0.05. A Turkey’s Post-hoc analysis revealed that significant differences exist between the experimental group (M=12.0, SE=1.0) and control group B (M=15.33, SE=0.33), and between control group A (M=12.33, SE=1.33) and control group B (M=15.33, SE=0.33) at P =0.048 < 0.05 and P=0.041 < 0.05 respectively. The difference between the experimental group and control group A was not significant.
Figure 29 represents the different performance in the new post-AX test among all three groups, and showing the maximum, minimum, mean scores and upper quartile and lower quartile for each group.
29
Figure 30 represents the different performance in the new post-AXB test among all three groups, and showing the maximum, minimum, mean scores and upper quartile and lower quartile for each group.
30
31
DISCUSSION & CONCLUSION
The current study looks at perception of Mandarin Chinese tones in school age Chinese adoptees, who were adopted when they were infants between age 1 and 2 by Dutch families. Two control groups, namely young Dutch learners of Mandarin Chinese and native Chinese children have been studied as well. Three main research questions have been addressed and the answers are delivered as follows:
Research question one: Whether the Chinese adoptees have completely lost their childhood memory of Mandarin Chinese tones after a moderate-period (at least 5 years) of non-exposure to the language. Based on the observed data and quantitative analysis from the attrition/retention investigation, the answer to the research question one should be negative. The Chinese adoptees outperformed the young Dutch learners in both tasks in the pre-test. Although the difference was found to be non-significant, the clear difference between their mean values has indicated that these adopted Chinese children maintained their sensitivity to perceiving Mandarin tones in certain degree; they did not totally lose their birth language. However, they did show some extent of attrition or loss of their birth language, as compared to the native Chinese children.
Research question two: Whether the Chinese adoptees can perform as well as Chinese children who have been raised in Chinese families after relearning Mandarin Chinese tones. The native Chinese children outperformed the Chinese adoptees in all four tasks in the post-test phase. Quantitative analyses also revealed significant differences between the two groups in most tasks. In particular, compared to the pre-test, the Rep-measures showed that the native Chinese children improved their perception of Mandarin tones after training, but the Chinese adoptees did not improve at all; instead, their accuracy decreased after relearning. Therefore, the answer to the research question 2 is also negative. 6-hours’ relearning did not reactivate their childhood memory in an explicit ways; they still lag behind the Chinese children.
32
up the tones faster than the young Dutch learners and did better in the in-class quizzes than the young Dutch learners during the relearning session. The negative result may due to various reasons, which is discussed in the following sections.
These data do not support the findings obtained in other studies involving attrition of a childhood language in two ways. (1) According to Pallier et al. (2003) and Ventureyra et al. (2004), this particular population who experienced a sudden disconnection of their language environment during their childhood may totally lose their birth language skills. However, the data from the 9 school age Chinese adoptees in the present study showed very variable results. They outperformed the young Dutch learners in both tasks in the pre-test and showed advantage of their prior experience to their birth language. Although they did not perform as well as the Chinese children, their performance did not significantly differ from each other. This evidence showed that the Chinese adoptees did not completely lose or forget their birth language, particularly in suprasegmental perception.
The discrepancy in results may be due to the fundamental difference between my population of adoptees and the populations in the above-mentioned studies. The populations in those studies are adult adoptees who have been severed from their initial maternal language for more than 20 years. My Chinese adoptees were school age children. They had only been isolated from their birth language for a moderate time period of five years.
33
On the other hand, the different length of non-exposure, namely time since onset of attrition, may have resulted in the different results. Time since onset of attrition is one of the extralinguistic factors; it seems to have an obvious relationship with the degree of erosion of the first language. However, the preliminary studies into the effect of time on attrition have been inconclusive so far. Some studies in the context of emigration have shown that the time elapsed since emigration can have a strong impact on language capability (e.g. Waas, 1996; Soesman, 1997) while others suggested that this factor may only have a restricted effect (e.g. de Bot & Clyne 1994; de Bot, Gommans & Rossing, 1991; Jaspaert & Kroon, 1992; Giesbers, 1997). Nevertheless, de Bot, Gommans & Rossing (1991:94) also claimed that there was a linear connection between “time” and attrition when there was little or no contact with the first language available. Therefore, the results from the attrition investigation in the current study stays in line with the previous findings (e.g. Waas, 1996; Soesman, 1997) and the notion (de Bot, Gommans & Rossing, 1991:94) that the time since onset of attrition has a close relation with the degree of attrition in the case of birth language attrition in international adoptees.
(2) Based on the findings from Oh et al. (2003) and Hyltenstam et al. (2009), relearning the birth language may refresh the adoptees’ childhood memory of the language. This gives them advantage, particularly in phonetic perception, compared to the foreign learners of the language. But the present data from the post-test phase failed to demonstrate any advantage in the childhood over-hearers. No improvement was found in all four tasks in the post-test; instead, they showed a decrease on the same two AX tests. In contrast, the young Dutch learners showed great improvement after the instruction. They outperformed the Chinese adoptees in most tasks. Three possible reasons may explain us the inconsistent results in the current study.
The first possible explanation is the different amount of early birth language input between the Korean adoptees in the above-mentioned studies and the Chinese adoptees. On average, the Korean adoptees heard Korean 40.3 h/week for 5 years before adoption in Oh et al.’s (2003) and 29 months in Hyltenstam et al.’s (2009), but the Chinese adoptees only heard their birth language for an average of 15 months before adoption. Moreover, there might be a possibility that these Chinese children were only exposed to their local dialect or a combination of both local dialect and official Mandarin Chinese. Therefore, the early input of Mandarin Chinese for the Chinese adoptees was very limited.
34
2.5 and 4.5 years respectively, 1 of them watched Chinese children songs through videos on an average of 2 h/week for two years after the adoption, but stopped for six years since then. 1 of them had a short visit to China for 10 days, 2 of them received 6-hours’ training on Mandarin tones with the tester within two months and a half, and the last 2 had never been re-exposed to Mandarin Chinese except taking the tests in this study. The different lengths of re-exposure or relearning provide different extents of language input for the populations of adoptees.
As one of the most crucial resources, language input enables learning and growth (Krashen, 1985). It has been proved that language input has close relationship with language development, particularly early first language acquisition (Tomasello, 2000; Diesel & Tomasello, 2000). With continuing meaningful input, the language system is able to develop and self-organize over time (Van Geert, 1991). As commonly acknowledged longer period of exposure to a language leads to more input. Therefore more language knowledge and skills will be possible to be acquired. This accounts for the reason why the Korean adoptees in the previous studies showed advantage in relearning their birth language but the Chinese adoptees did not. However, considering the advantage in perception of Mandarin tones they showed in the pre-test phase, with more exposure to and more extensive instruction on relearning the language, these adopted Chinese children are likely to demonstrate their strength.
35
played explicit roles are education, age, language learning abilities and language learning strategies.
1) Age. All 23 participants age between 4 and 16. It has to be admitted that this age gap is quite big. Different age levels give people different life experience, especially when learning is involved. Learning experience and effects are quite closely related to age levels. Previous studies have proved that the elders learn faster than the youngers (Snow & Hoefnagel-Höhle, 1978; Bialystok, 1997). The similar finding has been observed in the current study as well. Take the young Dutch learners, 6 to 9 year-olds learnt slower than 13 to 16 year-olds, especially for those who did not have foreign language learning experience. More information about learning experience and abilities is presented in the next section.
2) Education. 19 out of 23 participants in the present study are studying in primary schools in Groningen; the last 4 (Dutch learners) are studying in middle schools in Groningen and Leiden. When dividing groups, only nationality was considered, so participants with mixed educational levels were combined together, especially in the Dutch learners’ group. This combination has neglected their different learning experience and abilities. As we all know that older children with more educational experience are quicker and better at understanding and learning new knowledge. Likewise, younger children with less educational experience are slower at understanding and learning new knowledge. This notion has been confirmed through the tests and relearning process in the current study. Younger participants need more repetition both in the tests and relearning sections to understand the questions and testers, but older learners, especially the 4 Dutch teenagers think the tests were too slow and too easy for them.
