Physicists in Spain are challenging the idea that two languages cannot continue to exist side-by-side within a society. But while the findings may spell good news for some languages, it still leaves doubts over the long-term survival of more isolated languages such as Welsh and Quechua.

Jorge Mira Pérez, who led the research, became interested in the issue of language survival because of the situation in his own region of Galicia in north-west Spain where the population contains speakers of both Spanish and the local language, Galician. Teaming up with his colleagues at the University of Santiago de Compostela, Mira Pérez used a mathematical model to investigate whether these two languages could continue to coexist in the years to come.

Their approach built on an earlier study carried out at Cornell University in the US, which had modelled a two-language society by dividing a fixed population into two distinct language groups. In the Cornell model, speakers are free to switch between language groups driven by factors such as economic incentives, and the weaker language always dies out eventually.

Accounting for bilingualism

Mira Pérez's team realized, however, that this model does not take account of bilingualism and the impact this could have on the stability of each language. So they have developed a more advanced model to include three distinct groups – the two monolinguists and the bilingual – where people can shift between all three groups.

In Mira Pérez's model, the chance of each language group losing speakers is related to the "status" of each language, a parameter that takes into account the social and economic advantages of that language. It is also related to the number of speakers in each population to start with and the similarity of the languages in question.

To test against a real-world situation, the researchers compared their model with historical data for Spanish and Galician spanning the 19th century to 1975, and found that the fit was quite good. After varying the parameters with more than 400 different values, in order to span all the possible combinations of status and similarity, they concluded that it is possible for two languages to coexist indefinitely.

The key to survival

The key to survival is that two languages must be spoken by enough people to begin with and they must be sufficiently similar. "If the statuses of both languages were well balanced, a similarity of around 40% might be enough for the two languages to coexist," says Mira Pérez . "If they were not balanced, a higher degree of similarity – above 75%, depending on the values of status – would be necessary for the weaker tongue to persist."

The findings are good news for languages such as Galician and Catalan, spoken in autonomous communities in Spain, which have relatively steady numbers of speakers and share many similarities with Spanish, the dominant national language. It could spell bad news however, for more distinctive languages such as Quechua in South America, which is very different from Spanish and it is already being marginalized to rural communities.

Mira Pérez acknowledges that his model is based on an "ideal" society with static populations. It does not take into account of many other factors that could influence the balance between languages, including migration and the unpredictability of social dynamics.

Political factors

This is a view shared by David Crystal, a linguistics expert at Bangor University and author of the book Language Death. "They seem to be using a crude notion of lexicostatistics to define similarity, which is not a measure everyone respects, and they haven't taken important social variables into account," he says. Crystal also points out that political factors can also determine the fate of a language, as in the case of Welsh, which has seen a resurgence in recent years fuelled by two Language Acts and significant activism.

Andrea Baronchelli, a physicist at the Polytechnic University of Catalonia who has also developed an interest in the mathematical modelling of language, agrees that politics does mediate the number of speakers of a language. But Baronchelli argues that, so long as there is no pretension that mathematical studies of language can provide a complete picture, then they can still offer useful insights. "Simplification is however not automatically a problem. The role of modelling is in this case that of testing different hypothesis to show their consequences."

Mira Pérez tells that he would now like to develop his research by applying his model to other language pairings in different countries. He says he is particularly interested in the case of Belgium whose population speaks French and Dutch along with a number of minority languages. "It will be fascinating to see how similar French and Dutch although the geographical and political situation in Belgium is complex," he says.

This research is described in a paper in New Journal of Physics. See this video abstract from the paper: