Campus & Community

Language can be an ambiguous heritage

7 min read

Maria Polinsky studies the range of competence in heritage language speakers

Two images fill a computer screen in Maria Polinsky’s language lab. On the left, a young boy is painting a portrait of a girl. On the right, the roles are reversed — the girl paints a portrait of the boy. Once the images are shown, Polinsky, professor of linguistics, plays a single recorded sentence for the test subject seated at the computer: “The girl is being painted by the boy.” As the clock ticks, the test subject must choose the appropriate image.

Though it seems like a straightforward task, this test can be highly challenging when the sentence is played in a language one does not know well — which is precisely why Polinsky employs it in her lab. Passive constructions are difficult for non-native speakers to learn and can therefore provide a useful indicator of language proficiency.

The portrait test is just one of many tools that Polinsky employs to evaluate foreign language competency. For the past eight years, she has been studying heritage speakers, a term that refers to individuals who grow up in a family where a language other than English is spoken, but who learn English from their earliest days at school. Polinsky hopes that her studies will help answer questions about what it means to be competent in any given language, and reveal how humans learn and process foreign languages in the absence of formal instruction.

“Heritage speakers are those who never truly learned the family language,” Polinsky says. “They have certainly absorbed some from their environment, but they have never taken any formal classes.”

Heritage language speakers, says Polinsky, demonstrate a broad range of language ability. Some can understand the heritage language when it is spoken to them, but cannot read or write. Others may only be able to speak a few words. There are hundreds of thousands of heritage speakers in the United States, says Polinsky, and an especially large population exists among college students. Harvard is no exception.

“We estimate that nearly one-third of Harvard undergraduates are heritage language speakers,” Polinsky says. “In this group, the most common heritage languages are Cantonese, Korean, Mandarin, and Russian.”

The prevalence of these particular languages at Harvard reflects a broader contemporary trend in the United States, Polinsky explains.

“The most common heritage languages change over time to reflect immigration patterns,” says Polinsky. “A century ago, Italian and Yiddish were at the top of the list. Today, we see primarily Asian languages. Who knows what it will be in 30 or 50 years?”

Polinsky is able to draw on the large community of heritage language speakers at Harvard and in the Boston area for her research. She is primarily interested in those who have very limited knowledge of their heritage language.

“My goal is to figure out what these people know and don’t know in terms of the language spoken by their parents or in the home,” Polinsky says. “Cognitively, they are fully developed adults, but they come across as 5-year-olds when speaking in their heritage languages.”

This presents a challenge for Polinsky, who has had to create testing methods that will effectively capture the limited knowledge that the subjects do possess.

“You can’t just ask them to start talking, because you might only get mumbles or fragments of a sentence,” Polinsky says. “So that’s where the lab comes in.”

Aided by Ming Xiang, a postdoctoral fellow in the linguistics department, and a team of research assistants, Polinsky has designed a set of experiments to evaluate heritage language competency. During one test, subjects listen to a string of words — some of which are not actually part of the language’s lexicon — and click a button after each word to indicate whether they think the word is real.

“In addition to accuracy, we look at reaction times during the word recognition test,” Polinsky explains. “Competent speakers take an average of just 200 milliseconds to click, whereas heritage speakers can take up to four times longer.”

Another test, which Polinsky calls “elicited imitation,” requires the heritage speaker to repeat a sentence in the heritage language that has been read aloud by Polinsky or her assistants. She has found that heritage speakers have difficulty repeating and often tend to change things in a predictable way.

“If I pose the question, ‘Who was the shopkeeper yelling at?’ the respondent might instead ask, ‘Who yelled at the shopkeeper?’” says Polinsky. “This test demonstrates that there is something interfering with the subject’s ability to remember sentences in the heritage language. If we were testing a native-speaking child we might blame it on attention difficulties, since children often have trouble repeating. But if you are working with a fully developed 20-year-old, then you can think of the mistake in terms of language barriers.”

Polinsky has found that in general, even very good heritage speakers take about two to three times longer to read the language than do native speakers.

“Initially, Ming blocked out 45 minutes of time for the experiments,” Polinsky says. “But in many cases, we’ve ended up needing almost three hours!”

This year, the Polinsky Language Lab has evaluated heritage speakers in Chinese, Korean, Russian, and Spanish. Polinsky chose these languages primarily because they are well-represented in the Boston area and reflect contemporary immigration patterns.

To find subjects, Polinsky relies on Facebook, a popular networking Web site.

“Facebook is a great source for a lot of things people don’t imagine,” says Polinsky. “It’s easy to find speakers of exotic languages and get a sense for what language groups exist within the Harvard community.”

She also reaches out to potential subjects through colleagues who are linguists, by sending e-mails to undergraduates and by advertising on Craigslist. Forty subjects are required for a successful study, and Polinsky says she has to screen about 100 people to ensure she gets a sufficient number.

To confirm that the candidates will in fact serve as good test subjects, Polinsky and her colleagues interview the heritage speakers at length about their family background and administer a pretest, in which subjects watch a short video and are then asked to describe what they saw using the heritage language.

“These steps help us to ensure that the subsequent studies will yield accurate test results,” Polinsky says.

According to Polinsky, there are many practical applications for this kind of research. Most important, she says, are the possibilities for improving language instruction.

“Those who are interested in teaching heritage speakers certainly find this information useful,” she says. “We have discovered that heritage speakers who re-learn their family language have a distinct advantage, because they already have some vocabulary, generally have better pronunciation, and learn faster overall.”

Polinsky argues that heritage speakers should not be put in classrooms with those who are learning the language for the first time.

“You can’t mix and match these students,” she says, “because their needs are fundamentally different.”

Most rewarding for Polinsky are the cultural benefits of research and instruction in heritage languages.

“It is good for these individuals to learn more of their home language because it can further enhance their cultural knowledge,” she says. “In that sense, this is a socially responsible endeavor that can help boost self-awareness.”

Since she began her research, Polinsky has received several letters from grandparents expressing gratitude for the fact that they can finally communicate with their grandchildren in their native tongue.

“It was a very touching and unexpected outcome,” she says. “This project has taken on a life of its own!”

Polinsky and other heritage language scholars from across the United States will share the results of their research during the Second Heritage Language Summer Institute, June 23-27, at Harvard. For more information, visit

In addition, during the spring semester of 2009 Polinsky will teach Linguistics 162 — “Incomplete Acquisition: Heritage Speakers in Cross-Linguistic Perspective,” a new course that she has just designed. The course will focus on the linguistic study of heritage speakers’ knowledge of heritage languages and the testing and education of heritage speakers. During the course, students will examine data on heritage speakers of a variety of languages, including Russian, Polish, Czech, Armenian, Korean, Vietnamese, Tag-alog, Persian, and Arabic.