Young children respond best to other people who speak with the same accent, dialect or language as themselves, even if they grow up multilingual and exposed to a diverse language environment, according to a new meta-analysis of international studies.

In a paper published Wednesday in Child Development, researchers conducted the first meta-analysis of literature on infants and children's linguistic-based preferences, encompassing 38 studies. The authors found that infants and children overall prefer to engage with peers and adults who speak and sound like they do from early in their development, which was consistent with the majority of prior research. 

Jessica Spence, a Ph.D. candidate at the University of Queensland in Australia and lead author of the paper, told The Academic Times that children who grow up only speaking one language in relatively homogeneous, Western cultures are usually the ones thought to have biases about accents and dialects. 

But Spence was motivated to investigate whether children from all cultural and linguistic backgrounds exhibit preferences for those who speak like them, and she ultimately determined that bilingual children displayed just as much preference for speakers of their own linguistic variety as monolingual children did, if not more.

Bilingualism likely affects children's social preferences by allowing them to identify with multiple linguistic-based in-groups, and fostering opportunities for them to become accepting of people in other social groups, according to the paper.

"In particular, switching between two languages facilitates exposure to different cultural worlds, leaving bilinguals better equipped than monolinguals to understand individuals of diverse social groups and to cope with differences," the authors said.

But when choosing between native and non-native speakers, both monolingual and bilingual children in the analyzed studies displayed significant native-speaker preference, and this effect appeared to be stronger for bilingual children.

The 38 studies included in the meta-analysis were published between 1980-2020 and examined children's linguistic-based social preferences. They involved 2,680 infants and children ranging from just a few days old to 11 years old who lived in 13 countries — the U.S., the U.K., Australia, Brazil, Canada, France, Germany, Israel, Japan, the Netherlands, Singapore, South Africa and Spain. 

The children of the studies were also generally healthy, had diverse socioeconomic backgrounds and spoke 15 primary languages. Many of the studies required the children to make social preference decisions based on linguistic cues such as accent, dialect or language between monolingual native and non-native speakers.

To analyze the results, the researchers calculated the overall strength of linguistic-based social preference across all children from the 38 studies. They then compared the strength of linguistic-based preferences among monolingual versus bilingual children, and among those with varying levels of exposure to non-native speech. 

The authors noted that there was a considerable lack of reporting on details about the children's linguistic background in the studies; only about half reported information on children's exposure to non-native speech.

Overall, the children displayed significant preferences for speakers who had their native accent, dialect and language over non-native counterparts. The authors said the discovery that bilingual children appeared to show greater native-speaker preference than monolingual children was a "striking finding," because there has been substantial theoretical speculation that the opposite would be true. 

The children's cultural backgrounds, which were categorized as Western U.S., Western non-U.S., and non-Western, also did not impact their display of preferences.

"Children's preferences did not vary according to their cultural background, providing the first evidence that children's linguistic-based biases may transcend cultural differences," Spence said. "Moreover, contrary to the idea that being raised in a linguistically diverse environment may reduce linguistic-based biases, our findings suggest that these children may actually exhibit greater linguistic-based biases."

This analysis uncovered a range of factors that contribute to the development of linguistic-based biases in early childhood. Understanding these patterns can help inform efforts to reduce biases based on speech throughout adolescence, the authors said.

"Linguistic-based preferences for native speakers over non-native speakers are not only evident in infants and children, but are also present in adulthood," Spence said. "While linguistic-based preferences do persist throughout our development, it is likely that what is underlying these preferences does change with age." 

Infants who are too young to speak may have their preference for native speakers driven primarily by a familiarity bias that aids their linguistic learning, according to the authors. But when they learn to speak, they may be more likely to "meaningfully categorize others based on conceptual representations of linguistic-based group membership." 

"These representations may initially revolve around factual information about cultural and geographic background that linguistic varieties reveal," the authors said. "As children progress through the primary school years, their preference for native speakers may then evolve to incorporate societal attitudes."

For future studies on linguistic biases in children, Spence recommended digging deeper into environmental factors such as the occupations non-native speakers have, or the use of labels in classrooms such as "school English" and "home English." These may be reinforcing linguistic-based group differences and influencing the strength of children's linguistic preferences.  

"Our follow-up empirical work is really trying to focus on the 'why' question, including exploring how awareness of social norms against discrimination may impact children's display of linguistic-based biases," Spence said.

The study, "Something About the Way You Speak: A Meta-analysis on Children's Linguistic-based Social Preferences," published March 24 in the Child Development journal, was authored by Jessica L. Spence, Matthew J. Hornsey and Kana Imuta, the University of Queensland.