Skip to main content

Verified by Psychology Today


What Languages Do Bilinguals Count In?

New insights into the relationship between bilingualism and numerical cognition

Post written by Aneta Pavlenko.

It is often said that bilinguals continue using their first language for simple arithmetic operations, such as addition or multiplication, long after they shifted to the second language in other domains. I am not an exception to this phenomenon. After two decades in the U.S., I live, lecture, and write in English but when it comes to balancing my checkbook, calculating a tip, or counting the number of reps at the gym, I often switch to Russian. Do others also count in their first language while living in the second, and if so, why? And what does this adherence mean for kids who study math in a second language or shift languages mid-way through the schooling process?

To study the relationship between bilingualism and numerical cognition, researchers commonly use experimental tasks that range from number recall to complex mathematical problems. They also use large-scale surveys that ask participants about their language preferences for everyday numerical activities, such as object counting, calculator use, telling time, memorizing telephone numbers, and figuring out discounts. The findings of more than three decades of research confirm that bilinguals who learned a second language in late childhood or adulthood favor their first language for mental computations. They are also faster at remembering numbers and solving mathematical problems in that language.

The first language advantage, however, is limited to speakers whose early schooling was in their home language. When kids are schooled in languages different from those of the home, they tend to favor the language of early schooling as the language of mental arithmetic. For the world’s leading expert on numerical cognition, Stanislas Dehaene, this makes perfect sense. He argues that even the most fluent bilinguals favor the language of instruction, because the laborious process of learning and reciting arithmetic tables imprints them as word sequences in the brain structures, and it is more efficient to automatically activate these sequences than to relearn arithmetic in a new language.

Yet the picture emerging from bilingualism research is significantly more complex. To begin with, mental arithmetic is not the only area where we deal with numbers—we also have to retrieve numbers from memory, as dates, pin codes, or phone numbers. In my own case, the number of my old apartment in Kiev may pop out in Russian, while my social security number comes out in English. Such language dependence was also observed by an American psychologist, Elizabeth Spelke, who discovered that she could readily provide American friends with her summer address in France but not with her telephone number. Retrieving the number required that she say it in (non-native) French, visualize the numerals, and then mentally read them off in English.

Intrigued by this phenomenon, Spelke and her Russian-speaking graduate student at MIT, Sanna Tsivkin, conducted a training study with eight Russian-English bilinguals, who learned elementary arithmetic in Russian and favored Russian for everyday mental calculations. The participants were taught new numerical operations, new arithmetic equations, and new geographic or historical facts containing numerical information, in either Russian or English. When asked to recall these numerical facts and to solve problems or equations similar to those presented in the training, they performed faster and more efficiently in the language of the training.

More recently, Nicole Wicha and her colleagues in San Antonio, Texas, came to a similar conclusion when they examined the relationship between bilingualism and math using event-related potentials (ERPs). In their first study, they measured electrical activity in the brains of 22 Spanish-English bilinguals performing basic multiplication calculations. All of the volunteers were college students who grew up in Spanish-speaking families but became proficient in English by the age of 15. They were asked to solve simple math problems, some of which were presented in digits and others in words, in both Spanish and English. The results confirmed the advantage for the language of instruction but they also showed that some individuals responded faster in the language they used regularly and not in the language in which they initially learned basic math.

The researchers then repeated the experiment with 14 elementary school teachers, who were bilingual in English and Spanish and had an extensive experience of teaching arithmetic. Half of the participants taught in the same language in which they themselves learned math and the other half taught in a different language. The teachers were asked to judge the correctness of multiplication problems in each of their languages. Both groups were very fast in their responses, regardless of the language, but their brains appeared to respond faster to incorrect solutions in the language of teaching, regardless of whether it was the language of the speaker’s early schooling.

These findings suggest that the language bilinguals count in may depend on the language of early schooling but the language of other numerical tasks depends on their subsequent experiences with language and math, so that some tasks may be handled faster and more efficiently in languages learned later in life. But, to add a bit of spice to the equation, there are also those who do not rely on words and vocalizations at all when dealing with math and prefer to think of mathematical relationships "in the language of math"!

For a full list of "Life as a bilingual" blog posts by content area, see here.


Martinez-Lincoln, A., Cortinas, C., & N. Wicha (2015) Arithmetic memory networks established in childhood are changed by experience in adulthood. Neuroscience Letters, 584, 325-330.

Salillas, E. & N. Wicha (2012). Early learning shapes the memory networks for arithmetic evidence from brain potentials in bilinguals. Psychological Science, 23, 745–755.

Spelke, E. & S. Tsivkin (2001) Language and number: A bilingual training study. Cognition, 78, 45-88.

Aneta Pavlenko's website.

More from Francois Grosjean Ph.D.
More from Psychology Today