Not only do humans possess the ability to use language but they also have the ability to recognize language as systematic patterns. Speakers of a language can analyze the language use. They are able to tell if it is consistent with the language's grammatical rules and correct it if necessary. This type of language ability, namely, metalinguistic awareness, is quite different from the basic language skills and is generally acquired afterwards (Cazden, 1976; Bialystok, 1991; Romaine, 1997). In fact, in this study, Thai children aged 5-9 years could correctly produce /ï¢ï€¬ï€¿ï€¬ï¨ï€¬ï´ïƒ¾ïˆï€¬ï¬ï€¬ï¡ï®ï¤ï€ ï²/ in the onset position, but they had difficulty identifying these onset phonemes in real words (metalinguistic task).
One of the most important features of human language is Cultural transmission (Hockett, 1960). Specifically, infants are able to (consciously or unconsciously) acquire more than one language without any disruption. Using the age of language acquisition as a criterion, bilingualism can be categorized into 2 groups: simultaneous bilingualism and sequential bilingualism. The former occurs when a child is exposed to two languages simultaneously - mostly within the first three years. The latter occurs when a child learns a second language after having acquired the mother language. This usually occurs between the ages of 3 and 12. Sequential bilingualism is sometimes called second language acquisition, a process which can be due to many different environments. For example, in the Philippines, a child might learn one of the 70 dialects at home before being exposed to English and Tagalog.
Although bilingualism has become a common phenomenon in Thai society, issues in bilingualism of Thai (and other languages) have not been fully investigated. Of interest here is metalinguistic awareness in Thai bilingual population. In other languages, contradicting findings have been reported regarding metalinguistic awareness in (simultaneous) bilingual children. A number of studies suggested that bilingual children possess a higher level of metalinguistic awareness (especially in lexical arbitrariness domain) than monolingual children (Leopold, 1961; Vygotsky, 1986) although Ben - Zeev (1977) Cummins (1978), Edward and Christopherson (1988) suggested the contrary. Those who supported the bilingual advantage claim believe that because bilingual children have words in two different languages to refer to the same thing, they can more easily grasp the concept of the arbitrariness between word and referent, and can understand that the sound of a word has no relation to the meaning. Moreover, the fact that bilingual children see a language as one system among many might lead them to be more aware of the different phenomena in each language (Leopold; 1961, Vygotsky; 1986).
Researches conducted on metalinguistic awareness have focused on many linguistic domains, i.e. phonological, lexical, and structural domains. Concerning phonological domain, relations between phonological metalinguistic awareness and children's reading skills have been reported (Carroll, 2004). Moreover, with bilingual children having more phonological awareness, one can deduce that bilingualism lends itself to better reading skills.
DurgunogïƒŸïƒŸïƒŸlu, Nagy, and Hancin-Bhatt (1993) conducted a study on native Spanish-speaking students who are learning English as a second language and found that phonological awareness and word recognition skills in Spanish can predict word recognition skills in English. The study concluded that there is cross-language transfer. However, Bialystok, Majumder, and Martin (2003) argued that Spanish and English share a number of similarities, which could be mistaken for phonological awareness. They carried out a new study which showed that Spanish-speaking children have more phonological awareness of English, but native Chinese-speaking children do not develop the same phonological awareness of English.
As for the lexical domain, Yelland, Pollard and Mercuri (1993) explored the awareness of referential arbitrariness of kindergarten and Year 1 students in Australia. They tested the children's understanding of how the length of word for an animal has no relation to the size of the actual animal. For example, "caterpillar" is a long word that refers to a tiny insect while "whale" is a short word that refers to a large animal. The results show that English-Italian bilingual children did better on the test than monolingual children. Even children who had just started learning a second language surpassed the monolingual group on the same test.
In the structural domain, Ricciardelli (1992) compared structural metalinguistic awareness in Year 1 English-Italian bilingual students and English-speaking monolingual students. The students were asked to correct ungrammatical sentences said by a puppet. The results show that balanced bilinguals had more metalinguistic awareness than unbalanced bilinguals.
Most of the findings previously mentioned lend support to the claim of bilingual advantage. Therefore, the main focus of this work is to investigate if the claim holds up among the Thai monolingual and Thai/English bilingual population.
The scope of this study includes metalinguistic awareness of Thai children in 3 domains: phonological, lexical, and structure-meaning across age groups. It is generally assumed that different stages of cognitive development could explain the difference in development of language skills. This study takes this into account and therefore divides subjects into 2 age groups: 5-6 year-old and 8-9 year-old. According to Piaget (1959), the cognitive ability of the 5-6 year-olds is in the preoperational stage, which is when the child starts using symbols, develops memory and uses imagination. At this age, the child starts to develop full linguistic abilities, although logical thought and the ability to look at a situation from another person's viewpoint are still not achieved. The 8-9 year-olds have their cognitive ability in the concrete operational stage, which is when the child learns to use symbols in a concrete, systematic and logical way, and can adopt other people's viewpoints when looking at a situation.
It is noteworthy that the intention of using educational programs as a variable in this study, rather than the number of acquired languages, is to represent a realistic language/ bilingual situation in Thailand. Specifically, a large number of Thai bilinguals might be considered unbalanced bilinguals with Thai being their dominant language and English their second language. Thai children who are enrolled in an international program are influenced by the two languages, unlike those enrolled in normal regular schools who are only exposed to Thai in the classroom.
Subjects are 80 Thai students attending Grades 1 and 4 at Kasetsart University Laboratory School (Bangkok). They all took Raven's Coloured Progressive Matrices test (designed by John C. Raven). This test was designed to overcome the language and culture limitations of 5-12 year-old children. Out of 80 students, 40 students within designated ranges of age and levels of IQ (Wechsler Intelligence Scale for Children) were selected. Their WISC scores are shown in Table 1.
International program (IP)
5-6 years old (G1)
89 - 104
91 - 106
8-9 years old (G4)
98 - 106
99 - 106
Table 1. Wechsler Intelligence Scale of 40 children participated in the study
With regard to language use in the classroom, Thai students in the regular program are taught in Thai in all classes, except in their English class which they start learning from Year 3 onwards. Thai students in the international program have English as a medium of instruction, except in their Thai language class. They usually start learning and using English from pre-primary school.
II. Testing materials
There are 10 tests testing 3 metalinguistic domains of the Thai language: phonological, lexical, and structural-meaning. All tests were conducted individually for each subject and performed in Thai in a game-like manner with short familiarization sessions.
1) Three phonological tests: a rhyme-tone detection, a phoneme identification, and a phoneme substitution
1.1 rhyme-tone detection (adapted from Ho, C.S. and Bryant, P. (1997)) (6 test items: 3 for rhyme and 3 for tone)
Students are asked to choose one of the two words that has the same rhyme and tone as that of the first word. The test takes around 4 - 5 minutes.
Example: first word choice I choice II
à¹à¸ˆà¸ /ï´ïƒ¾ï…ï‚™ïƒ¹ï«/ à¹à¸‚à¸ /ï«ïˆï…ï‚˜ïƒ¹ï«/ à¹à¸‚à¹ˆà¸‡ /ï«ïˆï…ï‚™ïƒ¹ïŽ/
1.2 phoneme identification (32 test items: 16 for onset and 16 for coda)
Students are asked to identify an onset (or a coda) of each given word. The test lasts about 6 - 10 minutes.
Examples: test word target phoneme
onset identification test à¹€à¸•à¹ˆà¸² /ï´ï¡ï·/ /ï´/
coda identification test à¹€à¸‡à¸² ï€¯ïŽï¡ï·/ /ï·/
1.3 phoneme substitution (adapted from Bialystok, E., Majumder, S., and Martin, M.M. (2003)) (10 test items: 5 for onset and 5 for coda)
Students are asked to substitute an onset (or a coda) in the second word with that of the first word. The test takes around 14 - 18 minutes.
Examples: first word second word target answer
onset substitution test à¸Šà¸™ /ï´ïƒ¾ïˆï¯ï®/ à¸¥à¸´à¸‡ /ï¬ï©ïŽ/ à¸Šà¸´à¸‡ /ï´ïƒ¾ïˆï©ïŽ/
coda substitution test à¸«à¸¢à¸´à¸š /ïªï©ï‚™ï°/ à¹€à¸ˆà¹‡à¸” /ï´ïƒ¾ï¥ï‚™ï´/ à¹€à¸ˆà¹‡à¸š /ï´ïƒ¾ï¥ï‚™ï°/
2) Four lexical tests: a noun classifier test and 3 referential arbitrariness tests
2.1 noun classifier test (15 test items)
Students are shown a set of pictures and asked to justify and correct a noun classifier of each word. The test takes about 2 - 5 minutes.
Example: noun given classifier target classifier
à¸à¸£à¸°à¸ˆà¸ à¹ƒà¸š à¸šà¸²à¸™
/kraÂ™ï€¿Ã-tþoÂ™k/ /ï¢ï¡j/ /ï¢ï¡ùï®/
mirror classifier for tree classifier for leaf, fruit, mirror and document, and window
2.2 referential arbitrariness: 3 different tests including
Lelu-Bobe (2 test items), Sun-Moon (3 test items), and Dog-Cat (3 test items) (adapted from Osherson, D.N., and Markman, E. (1975)). Each test takes no longer than 5 minutes.
Example: a picture of a leaf is shown and subjects are asked if it can be called 'snow'. If yes, what color would it have?
(the correct answer is 'yes' and 'it would have the color of green as the way
leaves often are'.)
3) A structure-meaning test (23 test items)
A set of pictures is shown and students are asked to justify and correct syntactical structure and meaning of each given sentence. The test takes about 5 minutes.
Example: given sentence target sentence
à¸œà¸¹à¹‰à¸£à¹‰à¸²à¸¢ à¹„à¸¥à¹ˆà¸ˆà¸±à¸š à¸•à¸³à¸£à¸§à¸ˆ à¸•à¸³à¸£à¸§à¸ˆ à¹„à¸¥à¹ˆà¸ˆà¸±à¸š à¸œà¸¹à¹‰à¸£à¹‰à¸²à¸¢
'thief' 'chase' 'policeman' 'policeman' 'chase' 'thief'
A thief is chasing a policeman. A policeman is chasing a thief.
All tests were carried out such that the subjects start with the less complicated and progress to the more complicated ones, i.e. the structural-meaning domain, the lexical domain, and the phonological domain respectively. They were divided into two sessions on two separate days. The first session tested the subjects' structural and lexical awareness. The second session tested the subjects' phonological awareness. Each session lasted about 20-30 minutes.
For the classifier and structural-meaning tests, apart from analyzing the students' responses, error corrections made by each student were collected and categorized for further investigation.
Figure 1 compares the test results across three different language domains. It shows that all age groups in all programs have the highest awareness in the structure-meaning domain, followed by the phonological domain. The lexical domain is the domain in which all the subjects showed the lowest metalinguistic awareness.
Figure 1: Percent average correct responses from 3 metalinguistic domains
(pooled across 10 tests)
I. Phonological domain
Figure 2: Percent average correct responses from 5 phonological tests
Overall, as shown in Figure 2, every group performed best in the onset identification test and worst in the coda identification test. A notable difference in test scores in the test of coda substitution was found between the two programs (RP: 58 vs. IP: 49). However, no statistically significant difference was reported between the two educational program groups (RP: 73.5 vs. IP: 72.8 as shown in Figure 1). Moreover, The 8-9 year olds from both programs showed significantly higher test scores in the onset substitution and coda substitution tests than the 5-6 year olds.
II. Lexical domain
Figure 3: Percent average correct responses from 4 lexical tests
As shown in Figure 3, every group did best on the classifier test and worst on the Lelu-Bobe test. The 8-9 year olds from both programs showed significantly higher test scores in the noun classifier test than the 5-6 year olds. No statistically significant difference between the two educational program groups was found (RP: 48 vs. IP: 47.1 as seen in Figure 1).
III. Structure-meaning domain
Figure 4: Percent average correct responses from structure-meaning test
The two educational programs showed no statistically significant difference in the structure-meaning test score (RP: 92.9 vs. IP: 91.9 as seen in Figure 1). Moreover, the 8-9 year olds of regular program had significantly higher test scores in the structure-meaning test than the 5-6 year olds of the same program.
When correcting linguistic errors, most children succeeded, and very few failed to correct classifiers that are not frequently used. Likewise, most of them could correct clearly unacceptable meanings and syntactic errors. Only a few did not succeed in correcting complex sentences. In fact, some justifications and corrections from the children did not identically match the target answers, yet they were suitable with the use of words and structures which are possible and more familiar.
DISCUSSIONS AND CONCLUSION
Even though the findings do not support the argument that bilingual children possess higher metalinguistic awareness in any linguistic domain than monolingual children, they strongly suggest that sequential bilingualism does not inhibit development of the mother language (Thai). Moreover, stages of metalinguistic development of the two groups (regular and international programs) do not seem to differ. From these findings, the metalinguistic development seems to emerge first in structure-meaning domain, then phonological, and lastly lexical. Interestingly, at around the ages of 5 and 6, Thai monolinguals and bilinguals show a strong level of metalinguistic awareness in structure-meaning domain. That is, they not only are able to produce grammatical sentences, but also have the knowledge of syntactic rules, consistent with previous studies on other languages (Carroll, 2004).
In both educational programs, the 8-9 year olds performed significantly better than the 5-6 year olds in all tasks except for the test on onset - coda identification and those on word arbitrariness. Different stages of cognitive development as proposed by Piaget (1959) might explain different levels of metalinguistic awareness found between the two age groups.
We are grateful to Kasetsart University Laboratory School Center for Education Research and Development for allowing and assisting us in the data collecting process. A complete version of this work was the first author's MA thesis (2010) at Department of Linguistics, Faculty of Liberal Arts, Thammasat University.