As someone who has a background in ecology and many things biology, there are often many times I have tried to observe parallels between biodiversity and linguistic diversity. In fact, I have written a couple of posts before about these topics, breaking down published academic or scientific papers that explore these concepts. We see that some places have far more languages spoken in there compared to others, from the island of New Guinea to the Caucasus, while some islands are known to only have a few languages spoken on them.

Today, we will discuss an article by Michael C. Gavin and Nokuthaba Sibanda published in 2012 in Global Ecology and Biogeography, which aimed to explore which factors could explain the variation of language richness between islands. These islands in question spanned the geographic region of the Pacific, from Rapa Nui (Easter Island) to Indonesia, and from Hawaii to New Zealand. These islands present a stunning variation in language richness — New Guinea is famous for having the most number of languages spoken at over 900 of the world’s around 7 000 languages, while Rapa Nui is known to have only one or two languages spoken.

To explore the underlying factors influencing this observed variation, the authors presented four categories of such factors, of which three are rather consistent with ecological concepts, and the other is more, well, anthropogenically based.

Firstly, time. As humans migrated out of Africa, and settled out into the islands of the Pacific, different islands got settled in later than others. For instance, New Guinea may have had human settlement as far back as 50 000 years ago, while New Zealand was uninhabited until Eastern Polynesians settled in it around the 13th century CE. Citing a paper by Dynesius and Jansson (2000), language richness tends to increase over time, and hence regions with the longest continuous occupation would be the most diverse. So we would expect islands that were settled last to have the fewest number of languages spoken, and vice versa.

Next up, would be the most well-known factors often brought up when discussing island biogeography — that of area and isolation. From The Theory of Island Biogeography by MacArthur and Wilson in 1967, larger islands tend to harbour more species, and more isolated islands often support fewer species. The species-area relationship could be driven by context-specific factors, although the underlying mechanisms driving this relationship are under debate, with many hypotheses proposed. Immigration and extinction rates, habitat heterogeneity, and resource availability are all proposed reasons.

In terms of isolation, more isolated islands would have lower immigration rates, resulting in fewer species colonising those islands. Furthermore, there is this rescue effect whereby continual immigration from the mainland (or continent) to the island could lower the likelihood of the corresponding species going extinct. This effect is less pronounced in more isolated islands. In fact, such an explanation was proposed for the pattern of language diversity in Melanesia back in 1976 by Terrell. However, the authors remarked that many previous studies were rather confined in spatial scale, involving individual archipelagos, while the few relatively large-scale studies found that diversity was best predicted in combination with environmental variables.

But which environmental variables are we talking about here?

Environmental factors can come in various categories, such as the climate, and how diverse the environment is. After all, mountains, large rivers and other inhospitable places could function as barriers of human-human contact, forming geographical barriers between two potentially different languages. Another factor that is relevant to this topic is known as “ecological risk”. According to Nettle (1999), rainfall and temperature influences the mean growing season, which in turn determines the ecological risk of a location. What does a low ecological risk mean for human language?

With low ecological risk, a group of humans could obtain all required resources (think food, shelter, etc.) within a small area without the need to trade extensively with other groups of humans. This results in a greater language diversity within a set area. Taking all of these in, a gap in knowledge would be determining the role of such environmental factors on language diversity in islands and island chains.

And lastly, society. This is a more human-centric thing, since languages and their diversification and extinction could be driven by social processes. Group boundaries can be drawn within societies that do not necessarily correspond to geography, isolating people and thus resulting in divergence. Social hierarchies, conflict, and the strength of networks for trade and communication could all affect language diversity. The issue is that such social processes are hard to quantify, and used to determine if there is a correlation between language diversity and societal prcoesses.

So, how did the authors assess which of these factors influences language diversity in the Pacific?

The language data came from Ethnologue, perhaps the most well-known source of the world’s living languages, although pidgins, creoles, and national languages not captured by Ethnologue were excluded. Environmental variables like area and isolation, and temperature, rainfall, and soil conditions were sourced from various databases, some were quantified as continuous variables, while others were categorised discretely. For time, since first human settlement on Pacific islands could not be accurately determined, the authors used three categories to classify these islands by time since first human settlement. An additional consideration was made for this thing known as language phylogeny. This takes into account the likelihood of each language to diversify into more languages. With these factors, the authors determined the relationship with language diversity through the use of various statistical models

The authors found that among these factors studied, area and isolation played the largest roles in influencing language diversity in the islands of the Pacific, followed closely by habitat heterogeneity, time since first human settlement, and ecological risk. Interestingly, climate played a smaller role than expected in language diversity, and contrasted with several previous studies on ecological risk, like that by Nettle (1999).

A large talking point about the results would be the finding that the models could only explain half of the variation in language richness in the Pacific islands, being criticised as having a lower predictive power than models that investigate their species diversity counterparts, which models could explain at least 85% of corresponding variation in species richness. What could explain this discrepancy?

Could these differences be attributed to social processes, the factor that is substantially more difficult to objectively quantify than the environmental variables? Or were there additional factors the authors may have omitted due to poor data coverage? While these factors could potentially cover the remaining variation, it would be substantially more difficult to measure its relative impact compared to area and isolation.

While this study appears to be rather consistent with MacArthur and Wilson’s theory of island biogeography, there could be different underlying processes that explain this area-isolation and language diversity relationship compared to what you would find in ecology papers. More discussion was focused on how social processes could have affected language diversity, such as how stratified each society is, and how power is distributed across the social classes. Nevertheless, the authors concluded that a combination of biogeographical, environmental, and societal processes formed a complex web of factors determining how language richness is distributed across these Pacific islands.

This study has left me wondering, if there were more available data for additional factors, or if a rather objective measure of societal processes were devised, how their results would have changed. Having half of the variation unexplained, not all of these could be fully captured by society and omitted factors, leaving some of the influences rather idiopathic. Could there have been some stochasticity influencing these models as well? After all, there could be some variation attributable to just randomness. Not knowing how much of these variation is possibly attributable to which factor is rather concerning, since there could be a possibility where societal processes could outweigh environmental ones to explain variation in language richness, or that area and isolation are actually dominant factors. As such, the conclusion made by the authors is reasonable and properly represents their findings. It has definitely presented another parallel between language and biodiversity that could be drawn, although there are human-specific factors that could add to the complexity of the full picture.

Further reading

Gavin, M.C. and Sibanda, N. (2012), The island biogeography of languages. Global Ecology and Biogeography, 21: 958-967. https://doi.org/10.1111/j.1466-8238.2011.00744.x.