Or, to take the other side of the coin: “Why are OLDER immigrant children so much better at heritage language maintenance?” These two questions were tackled in a 3-year longitudinal study of 10 Chinese children/teenagers who immigrated to New York and New Jersey in the mid-1990s. Gisela Jia (City University of New York) and Doris Aronson (New York University) tracked these students’ English development and Chinese language maintenance over 36 months, and found age 9 to be a critical point: the children 9 or younger became English-dominant, while the children older than 9 maintained Chinese while learning English as an additional language. Are these findings evidence that a language is best learned when young? Not really, as the younger children were as limited in their heritage language maintenance as the older ones were slower at English learning. The authors conclude that “Age on Arrival” (AoA) is NOT the cause of Chinese loss or faster English gain. Rather, AoA influences easiest language of communication and preferred input environments, and these are huge factors that affect language development at any time of life.
Jia, G., & Aaronson, D. (2003). A longitudinal study of Chinese children and adolescents learning English in the United States. Applied Psycholinguistics, 24(1), 131-161. https://doi.org/10.1017/S0142716403000079
A persistent finding in language learning research is that age of arrival in a new country is a strong predictor of first language maintenance or speed of second language acquisition, regardless of the immigrant group or how closely or distantly related the two languages are. Some researchers suggest that this is due to a “critical period” in language learning, but what they tend to focus on is acquisition of the second language, not maintenance of the first or heritage language. To deepen understanding of Age on Arrival (AoA) effects by looking at both, Jia and Aronson argue that contexts of language use must be studied along with age.
People, they point out, are not input-driven machines but “intentional beings with feelings associated with their linguistic experiences. These feelings may well influence how they approach a language and how well they learn it” (p. 133). Preferences for language use in conversation and input sources (books, TV, etc.) shape language repertoire development over time, and what is interesting about this study is the patience with which the researchers tracked changes in 10 young people’s language environments over three years, gathering multiple rich sources of data.
The participants were 10 Chinese children and teenagers, 5 girls and 5 boys, who ranged from 5 to 16 years old when they immigrated to New York or New Jersey (neighbouring east coast U.S. states that are small in size) in 1995-96. They were:
- Anna (5)
- Betty (5)
- Carl (7)
- Dianna (8)
- Eric (9)
- Frank 9)
- Gary (12)
- Hua-lei (12)
- Jing-lan (15)
- Kang-da (16)
Seven were from Mainland China and three were from Taiwan. Seven spoke only Mandarin at home, while Betty, Dianna, and Gary knew Mandarin and another Chinese language. Their parents had college degrees, but held relatively low-level positions in their fields at the early stage of immigration, and spoke English to varying degrees.
Each child was visited once a month during Year 1 of the study (12 times), once every season in Year 2 (4 times), and twice during Year 3. Each home visit lasted half a day (4 hours) and consisted of language tasks, an interview with the child, an interview with the child’s parent(s), questionnaires, and observations. The following data were collected at each of the 18 points in time (12 + 4 + 2):
1. Answers parents gave on a written questionnaire (in Chinese) about their child’s language use in various situations. These included (a) number of hours of T.V. their child watched in each language during the week, (b) number of books their child read in each language in the semester/break, (c) number of friends their child had who spoke predominantly one language, and (d) estimated percent of each language use at home.
2. Observed language preferences, i.e., watching the child interact with people at home, and the researchers’ conversations with the child.
3. Test results on grammaticality judgment and translation tasks.
4. Parents’ ratings of their children’s ability in the two languages for listening, speaking, reading, and writing. For each of the four skills, parents had to rate BOTH the proficiency on a 4-point scale (1=nothing, 2=a little, 3=some, 4=fluent) AND the change since the last visit on an 11-point scale (0=no change, +5=great improvement, and -5=great decline).
The researchers found the findings from the different data sources highly consistent. These results are presented next.
The six children with arrival ages of 9 or younger developed quite differently from the four 12- to 16-year-olds. At the beginning of the study, all 10 participants reported preferring to speak and read in Chinese. However, by the end of Year 1, the two groups started to show “considerable age differences in their language preferences” (p. 138). All participants aged 8 and younger switched their preference to the second language (L2) for speaking and reading except the youngest participant, Anna (age 5/6), who had not started reading in either language.
From Year 2 on, ALL participants spoke English less than 30% of the time. (This actually suggests that if they were followed beyond 3 years, many or all of the older ones might also have become English dominant; i.e., the process was the same but at a slower rate.) However, while the younger ones did not use Chinese, the older ones maintained low levels of Chinese, suggesting that their heritage language development reached a limit relative to their age of arrival and remained steady, while the younger children spoke Chinese very little or not at all.
These results might have been different if the community had had more Chinese speakers, but all the study participants were “mainstreamed” in English classes at schools with predominantly English-speaking populations and only 10-15% ethnic Chinese students. Schoolmates might have spoken other languages at home, but for the purposes of everyone understanding each other, no language could serve the purpose but English.
Preferred Input Sources: Friends, Books, T.V.
A younger arrival age predicted a larger proportion of English-speaking friends, and over the three years, the six younger arrivals dramatically increased their number of English-speaking friends, while maintaining low numbers of Chinese-speaking friends. In contrast, older arrivals had low numbers of English-speaking friends and high numbers of Chinese-speaking friends, even in Year 3. In Table 2 below, one can see the same findings for books as friends (it seems the younger children did not know how to read in Chinese). However, for TV, everyone seemed to equally like watching in English, and unfortunately, the younger children were found to watch much more television (the group averages were 493 vs 292 hours in Year 1, and 615 vs 388 hours in Year 2).
Qualitative Data: Why?
“Complex interactions among initial L1 [first language] abilities and cognitive, social, cultural, and psychological factors provided some of the underlying causal mechanisms for the language switch or maintenance phenomenon” (p. 143). Basically, people like to be comfortable—and children with no study, work, or travel goals yet tend to go with the easiest, most comfortable language practices. Age differences led to relative difficulty or ease of L1 or L2 use, affecting motivations towards L2 learning and L1 maintenance, and ultimately a significantly richer environment in the preferred language. For example, the bar chart below shows the composite scores for L2 English environment richness (Friends + Books + Television + Conversation).
On the other hand, recall that the older children could still read between 2,000 and 5,000 characters in Chinese, whereas the younger ones could not read Chinese at all. While it is unlikely that the older children would have built on what they knew, it is also unlikely that the younger children would have learned anything—we don’t learn things for no reason, when our time and energy and pleasure could be spent elsewhere—and the older children, who could benefit from what they could read in Chinese, enjoyed doing so.
While the younger children had more ethnolinguistically diverse friends (European, Hispanic, Asian), the older children preferred Asian friends, both Chinese and Korean, and used Chinese with the former and English with the latter. The two oldest children, Jing-lan and Kang-da (15 and 16 at the start of the study) “were the most selective. During the first two years of L2 immersion, they had almost exclusively L1-speaking friends who came from the same region of China as they did” (p. 145). This seems to suggest that the closer to adulthood you get, the more aware you are of social mores and boundaries, which already have an impact on you (e.g., if you are older, you know the group that will most likely welcome you to sit with them if you are a new student at school). The participants’ perceptions of how many Chinese were in their school were affected by their chosen social circles. While Anna, the youngest participant, said, “Oh, there is a lot of Chinese people, but… they don’t speak Chinese,” Hua-lei, one of the older children, reported that she preferred to speak Chinese with peers in school because “there’s so many Chinese in the school” (pp. 145-146).
This did not mean that the older students were linguistically incapable of dealing with English speakers—they could do so perfectly well, but preferred to interact with peers from the same culture, which kept them updated on Chinese movies and music, celebrities and singers, while younger children shared American cultural experiences with their chosen friends.
What About Students’ Actual Language Proficiencies?
Participants’ scores on the grammaticality judgment tasks (in English) and translation tasks were equal in Years 1 and 2, but the younger ones did better in Year 3. Note that these tasks were probably related to nativelikeness; by Year 3, the difference would probably have concerned articles and prepositions, not, say, level of academic vocabulary. There is evidence that older arrivals may perform better on academic tasks than younger arrivals because of transferring previous subject knowledge from the first language (see Roessingh, 2008, another longitudinal study comparing older and younger Age on Arrival worth looking at).
For ALL participants, more severe Chinese loss occurred in reading and writing than listening and speaking. This should come as no surprise, as there would have been no opportunities for literacy development compared to maintenance of conversational Chinese. The younger students were most affected by the underdevelopment of literacy skills (i.e., they forgot what few characters they knew, and ultimately could not read or write in Chinese at all), but the older ones, too, were affected, in that they lost a little or did not gain literacy-wise. Some participants, of various ages, spoke Chinese quite well regardless of AoA, including Carl (7), Dianna (8), Jing-lan (15), and Kang-da (16), and a few even improved over time.
This study’s findings are similar to those of another study that Jia conducted with 147 native Spanish speakers who immigrated to the U.S. (Jia, Pantin, Alvarez, & Williams, 2002). Older and younger students in this study with Aronson performed more or less equally on grammar and sentence translation tasks (which featured quite basic, everyday language; all the students came to the U.S. quite young after all, and were immersed in English), but the younger children preferred to use English and subsequently used it more (and spoke at least a little Chinese), while the older ones preferred to use Chinese and used it more (and could handle some English).
Jia and Aronson thus propose that existing language proficiencies—and the comfort zones associated with them—shape social choices and input preferences, which in turn shape language proficiencies further. It is important to note that the younger children who preferred English were not more willing to go beyond their comfort zone; they arrived at such a young age that English was their comfort zone. “Older participants with higher L1 abilities at the time of arrival found it more comfortable to continue using L1, and younger arrivals with lower L1 abilities had less motivation to do so” (p. 153).
Of course, you will find it easier and more comfortable to hang out with people who speak your preferred language, and they, in turn, will share with you input sources (books, music, T.V. shows, etc.) in that language. If you are socially obligated to use the other language, you may try your best. Interestingly, because the older children would have had to learn English too, they ended up more bilingually proficient, and ultimately had more choice of what language to use; however, this phenomenon was also shaped by what was more natural and expedient, given the particular domain of language use. Since all the children were schooled in English, the older children were equally engaged as their younger peers in English for school purposes, though younger arrivals used English for more diverse purposes.
All in all, it was clear that the first/heritage language, with no great opportunities for maintenance in this setting, was less developed in ALL 10 participants over time. The older participants’ maintenance was not gain, even though they continued to use the language, and eventually all 10 participants—who today would be in their 30s and 40s—would have had their postsecondary education in the U.S. and taken up jobs in English, if they didn’t move back to China or Taiwan. [Side note: Given the Internet age (Lam & Warriner, 2012) and access to transnational social networks for heritage language maintenance, as well as the rise of Mandarin as a global language, this study would be well worth replicating in 2021 to see if findings would be different.]
In terms of theoretical implications, Jia and Aronson conclude that arrival age seems to point to neurobiological causes of language development “because it covaries with environmental factors” (p. 156). However, there is no reason not to believe that it is these environmental factors, and not the AoA or neurobiological mechanisms, that lead to older versus younger immigrants’ patterns of development in both the new language and the home language.
Jia, G., Pantin, Q., Alvarez, G., & Williams, R. (2002). Language proficiency and academic performance of Spanish–English bilingual college students: Predictive factors. Unpublished manuscript.
Lam, W. S. E., & Warriner, D. S. (2012). Transnationalism and literacy: Investigating the mobility of people, languages, texts, and practices in contexts of migration. Reading Research Quarterly, 47(2), 191-215. https://doi.org/10.1002/RRQ.016
Roessingh, H. (2008). Variability in ESL outcomes: The influence of age on arrival and length of residence on achievement in high school. TESL Canada Journal, 26(1), 87-107. https://doi.org/10.18806/tesl.v26i1.392