November 1, 2011 Features

The Cantonese-English Bilingual Project

see also

After a year in preschool, 4-year-old Ming, who speaks Cantonese as a home language and who is learning English in school, seldom talks to his teachers or interacts with other children. Although he seems to understand the general classroom instructions and routine, his teacher is concerned about his ability to communicate.

Are Ming's difficulties related to his environment—limited experience with language, for example—or his language-learning ability?

Through our University of Colorado program, we hope to find answers to questions about Ming and other bilingual children. The San Francisco- and Boulder-based project investigates the language development and skills of preschool children who first learn Cantonese—a tonal language—at home and then learn the typographically and phonetically different language of English at school.

We are examining children's language knowledge across domains, including semantic and syntactic skills, and their language-learning skills in their first and second languages. By comparing our data with those from previous studies of bilingual children from other language backgrounds, we would like to determine universal patterns and variability in bilingual language acquisition. Clinically, the systematic investigation from the large sample of Cantonese-English bilingual children can serve as normative data for early identification of language disorders.

Cantonese originated in the Guangdong and Guangxi provinces in Southern China. This tonal language is spoken by 71 million people worldwide (Lewis, 2000) and has been the dominant language of North American Chinatowns. Many waves of Cantonese-speaking immigrants have immigrated to the United States since the 17th century, but little is known about the language development of Cantonese-English bilingual children.

Project participants—more than 100 so far—are recruited and tested at the six centers of San Francisco's Kai Ming Head Start Program for children from low-income families. Most participants are from families of recent immigrants who speak Cantonese (or dialects of Cantonese) and have limited English proficiency.


Language learning occurs within socio-cultural contexts. Many areas of San Francisco— Chinatown, Sunset, and Richmond, for example—have high concentrations of immigrants from Cantonese-speaking regions in China and in Asia. These communities share characteristics, such as Cantonese restaurants, Chinese-English bilingual store signs, Chinese grocery stores, and small herb shops.

Chinatown in particular also has unique, historical Chinese-looking buildings, such as the pagoda-stacked Sing Chong Building, one of the first places rebuilt after the 1906 earthquake. But the buildings do not quite look like the traditional buildings in China, perhaps because the oldest Chinatown in the United States was re-built mainly for tourist purposes (e.g., Tsui, 2009). This image of Chinatown may convey the skewed impression that the Chinese people in San Francisco enjoy living in an insulated community and refuse to assimilate; in fact, the densely populated Chinatown serves as a temporary home for many new Chinese immigrants who typically will move to other areas as soon as they can afford it (e.g., Tsui, 2009; Du, 2010). This stereotypical image is not representative of all Cantonese-speaking immigrants and Cantonese-English bilingual individuals in San Francisco.

Another misconception about children who learn the same home language—such as Cantonese in San Francisco—concerns the presumed homogeneity of their home social-cultural experiences. Socio-cultural contexts—such as place of origin—within this minority community vary greatly and contribute to the diverse contexts in which children learn their home language.

For example, although all participants are from low-income Cantonese-speaking households, their families come from various regions of the Guangdong province in China, from Hong Kong, or from other parts of Asia (e.g., Vietnam). Many Cantonese dialects are spoken in the Guangdong province. Standard Cantonese—the language spoken in Hong Kong and Guangzhou—is generally considered more socially prestigious and "educated." Some families in the study speak standard Cantonese at home; some speak Taishanese (a Cantonese dialect) along with standard Cantonese; a few speak Cantonese and Vietnamese.

In addition, some immigrant parents have experienced the political and economical upheaval in China, whereas others have experienced a more stable environment. Some live with their extended family and some do not. The complex traditions, the histories of the immigrants, and the diverse socio-cultural contexts affect how immigrant families raise their children and the parents' views about how their children learn Cantonese and English—views ranging from favoring bilingualism to favoring English.

Developing Measures

Given the diverse backgrounds of this group of bilingual children, this research project is designed to examine typical language development in this group of diverse language-learners. The development of the battery of measures used in this study is based on several theoretical perspectives on bilingual language development:

Bilingual children are not a sum of two monolingual children. Testing children with measures developed for either English-speaking or Cantonese-speaking monolingual children may not provide enough information about the dynamic interactive language system in bilingual children. Therefore, instead of using tests from each language, we develop single measures that allow assessment of children's language skills in both languages.

Social-cultural-linguistic contexts in which language input is embedded are important for children learning two languages. Stimuli should be relevant to the cores of the two cultures in which the children learn Cantonese and English. For the receptive and expressive vocabulary measures, for example, items were selected from both cultures in accordance with cross-linguistic findings. Previous research demonstrates that cross-cultural and cross-linguistic differences in the jnput are related to early word learning (e.g., Choi, 2000; Tardif, Gelman, & Xu, 1999). For example, Chinese-speaking monolingual children appear to have relatively more verbs than English-speaking monolingual children (e.g., Tardif et al., 1999). Therefore, we include not just nouns in vocabulary measures, but also other word types (e.g., verbs and adjectives).

Moreover, we consider children's cultural experience when selecting testing items. For example, we include the word "bitter" in the picture identification task. The word "bitter" is not commonly used in standardized tests developed for monolingual children who grow up in American culture. But it is a common word in Chinese cultures, because children who grow up in Cantonese-speaking households typically drink very bitter herbal tea when they get sick. Our preliminary data show that participants can rapidly identify the word "bitter" in both Cantonese and English.

Children use a common cognitive system in learning both languages. Children are able to learn and use two languages when they have enough language-learning opportunities across social settings. In this project, we particularly investigate children's skills at learning new words across different learning conditions in Cantonese and in English. To control for children's experience with the new words, we developed novel words (word forms that follow the phonological rules of a language but have no meanings in that language) in Cantonese and in English when testing their word-learning skills. We present each participant with each novel object along with its novel word-form in each language in different learning conditions. Consistent with previous studies (e.g., Kan & Kohnert, in press), our preliminary data show that Cantonese-English children's word-learning performance in both languages is directly associated with the amount of input.

Clinical Application

Clinically, the data from the large sample will provide norms for the diagnosis of language disorders in Cantonese-English bilingual children, like Ming, who exhibit language difficulties. Ming's parents, who immigrated to the U.S. about seven years ago, work more than 10 hours every day in a local restaurant. Ming spends a lot of time with his grandmother, who speaks Taishan and Cantonese. Ming was referred to a speech-language pathologist for an evaluation after the classroom teacher raised concerns about his language skills.

The SLP tested Ming in English using the Peabody Picture Vocabulary Test and the Expressive Vocabulary Test. Both test scores were two standard deviations below the mean. In addition, the SLP (along with an interpreter) elicited spontaneous language samples in English and Cantonese. Ming was able to produce some one- to two-word utterances when describing a trip to the zoo in Cantonese. He refused to speak when he was prompted in English.

The SLP consulted our lab for recommendations. We suggested testing Ming in two stages. Stage 1 focused on his language knowledge in both languages. The SLP used the picture-naming task and picture-identification task to test Ming's language knowledge in Cantonese and English. Ming's vocabulary skills were stronger in Cantonese than in English. Ming's scores in both languages were combined, revealing vocabulary skills significantly lower than the scores of the age-matched participants in our study. But were these scores related to his limited experience or to his language-learning ability?

Stage 2 focused on Ming's learning capability. The procedures used in this stage were similar to those in the dynamic assessment. Specifically, we suggested the SLP select real words new to Ming in both languages and teach him the new words in different contexts. The SLP found that, with similar exposures and prompts, Ming seemed not to learn the new words in either language. His performance in each language was certainly lower than the word-learning performance in the typically developing Cantonese-English bilingual children in our study. The overall results confirm that Ming's lack of language experience is probably not the sole reason for his lower language performance on the vocabulary knowledge measures and suggest that he may need intervention for language impairment.

Research Outcomes

We are analyzing the data collected from about 100 Cantonese-English bilingual children, and collecting data from more participants. We anticipate that the overall data yielded by this project will contribute to scientific advances in bilingual language acquisition and add to the bilingual literature by defining the various learning conditions that lead to optimal gains in word knowledge in both languages.

Pui-Fong Kan, PhD, CCC-SLP, is an assistant professor in Department of Speech, Language, and Hearing Science and principle investigator of the Child Language and Learning Lab at the University of Colorado at Boulder. Her general research investigates language acquisition in young children learning two languages, beginning at different points in their cognitive and social development. Contact her at

cite as: Kan, P. (2011, November 01). The Cantonese-English Bilingual Project. The ASHA Leader.


Choi, S. (2000). Caregiver input in English and Korean: Use of nouns and verbs in book-reading and toy-play contexts. Journal of Child Language, 27, 69–96.

Du, L. (2010). Learning to be Chinese American: Community, education, and ethnic identity. Plymouth, U.K.: Lexington Books.

Kan, P., & Kohnert, K. (in press). A growth curve analysis of novel word learning by sequential bilingual preschool children. Bilingualism: Language and Cognition.

Lewis, M. P. (2000). Ethnologue: Languages of the world, 14th Edition. Retrieved from

Tardif, T., Gelman, S. A., & Xu, F. (1999). Putting the "noun bias" in context: A comparison of Mandarin and English. Child Development, 70, 620–635.

Tsui, B. (2009). American Chinatown: A people's history of five neighborhoods. New York: Free Press.


Advertise With UsAdvertisement