Skip to content
Author:
  • Audio
  • 30 mins

Darwin Now pod 6: Language diversity

Updated Tuesday, 10th November 2009

Do languages evolve in the same way as species do in nature? If so, how can evolutionary biologists shed light on the study of language diversity?

This page was published over five years ago. Please be aware that due to the passage of time, the information provided on this page may be out of date or otherwise inaccurate, and any views or opinions expressed may no longer be relevant. Some technical elements such as audio-visual and interactive media may no longer work. For more detail, see our Archive and Deletion Policy

Audio

Copyright British Council

Transcript

Music: Variations on a Theme by Giuliani

 

Cecile Whidborne:

Bonjour et bienvenue.

Martha Lisk:

Hola, bienvenido.

Ursula Stickler:

Hallo, und willkommen.

Joseph Hopkins:

Hola, benvingut.

Becky Jones:

Buongiorno e benvenuto.

Malihe Sanatian:

Salam, khosh oomadid.

 

Rissa de la Paz:

Welcome to Darwin Now. We’ll be turning our attention from the variety of species in the natural world to another flowering of diversity – language. It’s probably the trait that distinguishes us from all other species on earth; it allows us to plan, to communicate, formulate rules, give directions and to share knowledge or experience. And through it, we’ve evolved a rich array of ways in which to express ourselves.

Quentin Atkinson:

In the world today there are about six thousand recognised languages. A lot of those we know are related into large families like the Indo-European language family spread across Europe and the near east, the Bantu language family in Africa is a huge family of some five or six hundred languages. The largest language family in the world is the Austronesian language family which comprises about fifteen hundred languages and stretches from Hawaii and Easter Island in the east to Madagascar in the west so it covers about a third of the globe.

 

Rissa de la Paz:

Quentin Atkinson, an evolutionary biologist now working at the University of Oxford has for some years been collaborating with colleagues at the University of Reading, in the Evolutionary Biology Group headed by Professor Mark Pagel. What’s the attraction of language diversity for evolutionary biologists? Mark Pagel.

 

Mark Pagel:

I think what got me interested in language evolution was that people like myself, evolutionary biologists, were interested in things that evolve not surprisingly, and when you look around the world you see that a single species, human beings, speak about seven to eight thousand mutually unintelligible languages, and so it’s evident that we have this system that has evolved a lot of diversity, and so someone like me he wants to understand that diversity in the same way I might try to understand the diversity of species, of biological species.

 

Quentin Atkinson:

I got interested in the evolution of language when I realised that the processes that generate language diversity are very similar to the processes that generate biological diversity and so that means that the methods that we use to study species evolution can be applied to study the evolution of languages.

 

Mark Pagel:

There are many parallels between biological evolution and linguistic evolution. When we study biological evolution, we’re blessed by the fact that we know there are genetic systems composed of these discrete elements we call genes, and those genes are inherited from parents to their offspring, and we know those genes can change over time by processes of mutation, and we know that some genes are favoured over others because of what they do for the bodies that they reside in and we call that favouring natural selection. And when you look at languages we also see that they’re comprised of discrete elements, we call them words, and there are other aspects of language that are discrete, like the grammar or the syntax. Those discrete elements, those little chunks, are handed by and large from parents to their offspring, so there's a process of replication. We also know that mistakes can be made to those processes of mutation and we also know that in some circumstances we can say the right thing and we can say the wrong thing and so there's a process as much like natural selection as a right way to express oneself or a better way to express oneself and a worse way.

 

Rissa de la Paz:

When we communicate through language, what filtering processes might operate to favour certain words or expressions over others? In other words, what’s the cultural analogue to natural selection?

 

Mark Pagel:

One might be that if we want to be understood we have to follow a very, very simple rule which is that we should speak like everybody else, and so there isn’t selection so much on the particular word we use, but on the fact that we should use the same words that everybody else is using, and so in that sense there’s what we call frequency dependent selection – do what everyone else is doing. In some other senses there may actually selection on words, that is to get them right. If I say to you that there are two guys coming over the hill and they’re going to attack us, and in fact I should have said two hundred, consequences could follow, and so in that sense there may be selection to get our language correct.

We think that linguistic selection is acting on the words we use and the sentences that we speak, and the reason it acts is that those sentences have to make sense, they have to be understood by the people we’re speaking to, and they have to be clear and replicable, that is they have to be something we can learn and teach to our offspring.

 

Rissa de la Paz:

Given that language isn’t a physical chemical entity in the way that genes are, it’s remarkable how closely its transmission mimics a genetic system in one crucial respect.

 

Mark Pagel:

Genes are replicated when they’re transferred from parents to their offspring. We make a copy of that gene and we pass it on to our offspring. Now in language evolution there is no analogous physical chemical unit that gets passed on from parents to their offspring, and so it’s remarkable that language can be transmitted with such accuracy, given that it’s not a physical replication process. It's a process of me imitating what you're saying to me.There’s a fundamental way that language evolution differs from genetical evolution, and yet language evolution mimics genetical evolution in a remarkably high fidelity way.

It’s so high fidelity that not only is it extremely easy for children to learn from their parents and they might be separated by twenty yearsbut children can speak to their grandparents and they may be separated by maybe fifty years of language evolution, but we know that we can read texts in our language that are hundreds of years old, sometimes even thousands of years old. And so we know this process of what Darwin would have called descent with modification, that is the language being passed on from parent to offspring over and over and over, is of such fidelity that the language is recognisable over maybe hundreds of thousands of years.

 

Rissa de la Paz:

If languages do evolve by a process of descent with modification, we might expect them to form family trees of related languages. Indeed such linguistic family trees were an inspiration for Darwin as he pondered on biological evolution. Now, 150 years later, can biologists return that favour by offering some of their theoretical tools to the study of language diversity?

 

Mark Pagel:

We can borrow or sometimes adapt mathematical and statistical methods that we can use to characterize the evolution of species and apply them to languages and one of the things we have done is just that. So we have dreamt up the sort of mathematical models that make predictions about how language might evolve, and we apply those to languages, and they then allow us to draw these very precise phylogenetic trees we call them, and that’s just a technical term for a for a family tree, and what's remarkable about is that we’ve had great success adapting these methods from biology to use on this linguistic data.

 

Rissa de la Paz:

Pagel, Atkinson and their colleagues have been applying their models to studying the evolution of languages, particularly from the Indo-European family. What are the key questions that interest them? Quentin Atkinson.

 

Quentin Atkinson:

The first is how the languages in the world today are related, how different language families are related, where did they come fromand when did they arise? A second question is if we can characterise how languages are related, we might be able to also start studying the process by which they change.

 

Mark Pagel:

One of the things we can do when we’ve got a phylogenetic tree of languages is that we can begin to ask how do various aspects of language evolve along that tree.

If we look at individual elements of language on that tree, we see that some of the elements evolve very, very rapidly and other elements evolve very, very slowly,

So for example the word that Indo-European language speakers use to describe two objects is the same sound across all of the Indo-European languages and so the Spanish say ‘dos’; the French say ‘deux’, the Italian say ‘due’ the German say ‘zwei’: all of these sounds are what linguists call homologous and what Darwin would have recognised that as they are sounds that all evolve from a common ancestor.

Cecile Whidborne:

Un, deux, trois, quatre, cinq …

Joseph Hopkins:

Un, dos, tres, quatre, cinc ...

Becky Jones:

Uno, due, tre, quattro, cinque ...

Martha Lisk:

Uno, dos, tres, cuatro, cinco ...

Jim Donohue:

Een, twee, drie, vier ...

Ursula Stickler:

Eins, zwei, drei, vier, fünf ...

Malihe Sanatian:

Yek, doe, se, char, panj.

 

Mark Pagel:

But the puzzle is that that other words in the same Indo-European languages evolve very, very rapidly, so, for example, what English speakers call a bird the Italians call ‘ucello’, the Spanish call ‘pajaro’ the Germans call ‘vogel’ and Caesar back in Roman times would have said ‘avis’ and so here we have a concept, ‘bird’, that has acquired a whole lot of different words.

 

Quentin Atkinson:

So the first question we were interested in was how quickly are these words changing and we were able to model that process on these language family trees, and quantify the rate of change. And we calculated what we call a word half-life, which is the time after which there’s a fifty percent chance that the word has changed to something else. The shortest half-lives for the most rapidly evolving words were about 4 or 5 hundred years; the longest half-lives were of the order of 70 thousand years, so these words really don’t change very much at all.

 

Rissa de la Paz:

This degree of conservation for some words is remarkable when you consider that languages are evolving simultaneously and independently of each other.

 

Mark Pagel:

When we study languages we have to bear in mind that each of these languages is existing independently and evolving through time, and so if we have ten different languages each of which has been around for one thousand years what we really have is ten thousand years of language evolution. These are language years, and if a single word has been used by all of those language speakers in all of those different languages for those years that represents ten thousand language years that that word has been conserved.

 

Rissa de la Paz:

But why are some words so highly conserved while others seem free to evolve very rapidly? What’s the basis for this variation?

 

Mark Pagel:

Drawing an analogy from biological evolution we hit on the idea of a word’s ‘expression level’. How often is that word used in everyday speech? In biological evolution sometimes how often a gene is expressed in your body, how often it’s used in your body is related to how rapidly it evolves. So we tried to investigate a word’s expression level, and we were able to find information on the frequency with which words are used in everyday speech.

Quentin Atkinson:

We chose four languages: English, Greek, Spanish and Russian as examples from across the Indo-European language family, and these were high quality data sets with millions of conversations, and what we found was the more frequently used words changed much more slowly than the less frequently used words.

Mark Pagel:

Right across the board, in the Indo-European languages, words that are used frequently in everyday speech are highly conserved; they tend not to change. So words like two objects that we describe or three objects that we describe or a pronoun like ‘I’ or ‘who’ or ‘what’, these words all evolve rather slowly and they come up in our speech over and over and over .But words like ‘bird’ or ‘belly’ or ‘dirty’, these words that we don’t use so frequently in our everyday speech, they seem to evolve rather rapidly.

 

Rissa de la Paz:

In some ways the result was a surprise: one might argue that words that we use more often might have more opportunities to change. But to explain why frequently-used words evolved more slowly, the group came up with two possible explanations.

 

Mark Pagel:

One possible explanation is that the more we say a word, the better we get at saying it. So when we use the word we're less likely to make a mistake. We introduce fewer mutations into that word, if you want to use a genetic analogy. So it could just be that there are fewer production errors in words that we use frequently because we’re better at using those words. Another possibility was that if we use a frequently used word incorrectly, we will hear somebody else using it correctly, rather soon and we’ll correct our mistake. And so in some sense our mistake will not be allowed to propagate because others will be using the word correctly, whereas with an infrequently used word, we can use it incorrectly and that mistake can propagate for a long time before somebody corrects it.

 

Rissa de la Paz:

But could it be that how often a word is used is simply a feature of what part of speech it is, that is, whether it’s a noun such as ‘dog’ or ‘cat’ or a verb such as ‘run’ or ‘walk.’ So Pagel and his colleagues took this into account when they performed their analysis. They found that for any part of speech like a noun or a verb, the more often it’s used, the more slowly it evolved. But overall, what part of speech a word corresponds to, does affect how quickly it evolves.

 

Quentin Atkinson:

What we found was that numbers evolve most slowly, given how often they’re used, then pronouns, things like ‘he’, ‘she’, ‘I’, then nouns, naming words like ‘cat’, ‘dog’ and so on, then adjectives, the describing words like ‘yellow’, ‘tall, ‘thin‘, then the verbs were slightly faster again, and then the most rapidly evolving word types were conjunctions and prepositions. So conjunctions are words like ‘and’, ‘but’, ‘because’ and prepositions, words like ‘in’ and ‘on’ and ‘at’, and they evolved relatively quickly given how frequently we used them.

 

Rissa de la Paz:

What could the explanation for these findings be? Pagel offers some suggestions which could provide exciting avenues for future research.

 

Mark Pagel:

Nouns and verbs carry a lot of information. It's extremely important to get them right in a sentence, whereas conjunctions and prepositions are sort of place-fillers in sentences, and so maybe it's the case that parts of speech influence the rate at which a word evolves by how much information they carry. And so it may be there’s been a lot of pressure on us to get nouns and verbs right, and so when we look across a selection of words, it’s words like the nouns and the verbs that evolve slowly and it's interesting that numerals, the number words, evolve the most slowly, and so here again if I’m trying to describe a band of people coming over the hill who might be trying to rob me and I say there's two when in fact and twenty or two hundred there could be big consequences, and so it may be the case that some words are very, very informative and so there’s been pressure to get those right and as a result they evolve rather slowly.

 

Rissa de la Paz:

Given how complex the structure of language can be, it’s remarkable how well the proposed models were able to explain linguistic evolution.

 

Quentin Atkinson:

These two simple predictors of rate of change: the frequency with which we use words and the part of speech, can explain 50% of the variation in the rates at which different meanings change. By any standards that’s quite a lot of the variation in rate of change. Contrary to what a lot of people think, languages aren’t changing in a completely unpredictable way. We’re able to use these two simple predictors to really make a pretty good guess about how quickly different meanings are changing. So that’s really promising for any research attempting to model the process of language change, to really get an understanding of what is going on when we’re using languages day-to-day and how that shapes the form of languages over centuries and millennia.

 

Rissa de la Paz:

Viewing language change from this longer-term perspective has been another fruitful avenue of research. What might influence how rapidly a language evolves from its point of origin to the present?

 

Quentin Atkinson:

What factors affect how the whole language changes through time? Does a language change at a roughly constant rate through time? Does it evolve in what you might call punctuated bursts? One theory in biology is that species evolve in punctuated bursts, so that when new species are formed their rates of evolution increase. Mark Pagel and some of my colleagues at the University of Reading tested this idea in a biological context and were able to show that indeed the rate of genetic evolution does speed up when new species are formed. What we wanted to try and do was test the same idea in languages. Do languages increase in their rate of change when new languages form?

 

Mark Pagel:

And so we got together collections of languages that form family trees of languages such as the Indo-European languages and we had a group of African languages known as the Bantu languages and a group of Austranesian languages; these are the people who moved out into the Pacific, the Polynesian islanders and so on. And we studied these languages to see f they ever showed these bursts of change in their histories. And one of the interesting things we found was that languages do show these bursts of change, and in particular it seems to be the case that if we follow a language over a long period of time, tracing it back to its ancestral language, that languages that have experienced many, many events in their history of splitting and forming into sister languages, seem to suffer more change than languages that don’t have this history of splitting events. And what we found that the more often that splitting has happened, these sub-populations have formed, the more likely the languages were to have changed in these rapid bursts. So it almost seems to be the case that there’s a social effect, that when languages split into two sister languages they somehow try to differentiate themselves from each other; they try to change in some way, almost as if forming an identity around the language.

 

Rissa de la Paz:

One of the most striking examples of a rapid burst of linguistic evolution associated with two languages wanting to establish their own identity can be traced back to Noah Webster around the time of the American Revolution.

 

Mark Pagel:

Noah Webster was an educator who was putting together the first dictionary of American English, and Noah Webster says in the preface to that dictionary that the American people need a system of language that is independent as well as a system of government, and it’s well known to American schoolchildren that Noah Webster in that dictionary introduced, you might say overnight, a series of peculiar or idiosyncratic American spellings for a whole lot of words that differ from those in British English.

 

Rissa de la Paz:

So much for language as a marker of identity. How about language as a marker of origin and migration? We can use phylogenetic methods to infer the ancestry of languages and probe human history, to find out, for instance, where different groups of people have come from and where they moved to.

 

Quentin Atkinson:

Biologists are using human DNA from around the world to reconstruct the human story from Africa around the globe – how we colonised the globe. But at more recent timescales, the genes aren’t changing fast enough to provide a signal over the last ten thousand years. So genetic data – its resolution breaks down at more recent timescales. What’s nice about the language diversity is that because languages change much more quickly we can use them to kind of fill in that last little bit of human history. And so just like biologists can look at the differences in gene sequences between species, and they can then work out how long ago those species must have separated, we can use the same technique and apply it to languages. Look at the different words in the different Indo-European languages and count how many changes we see and then work out what time must have elapsed to produce that amount of change. We used that kind of approach to test between two competing theories for the origin of the Indo-European languages.

 

Rissa de la Paz:

One theory suggested that the Indo-European languages were spread with a warrior horse-riding culture called the Kurgans, from the Russian steppes about five or six thousand years ago. The other proposed that the languages spread more passively from Anatolia, what’s now Turkey, with the spread of farming, eight to ten thousand years ago. Could the family tree of the Indo-European languages itself shed light on the problem? For Atkinson, the answer emerged through a collaboration with Professor Russell Gray at the University of Auckland.

 

Quentin Atkinson:

What we found was that the age of the Indo-European language family was around eight thousand five hundred years. Crucially the range when you allow for some statistical error in our estimate was between 7500 and 9500 years. That fits nicely with the farming theory but it allows us to rule out the 5 to 6000 year age suggested by the Kurgan theory. So that’s really quite strong evidence in support of the spread with farming.

 

Rissa de la Paz:

So much for illuminating the past. What of the future for language diversity? With increased migration, mass media and the rise of the Internet, will the global melting pot yield a recipe for language that loses some of its richness and flavour? Mark Pagel.

 

Mark Pagel:

Well we’re going to see this great melding together of languages and we’re already seeing this in some ways in that there's a kind of Spanglish, and there’s a kind of Franglais and there’s a kind of language of the Internet in which English words are used in many, many different languages around the world. But I think what’s far more interesting is that the languages are homogenising at a far, far slower rate than they might, given the Internet, given mass communication, we could expect this process of language homogenation to happen almost overnight but it’s not. And so what we see is a great resistance in languages to change from the outside, and we know that people hold their languages very, very dear to their identities. We know that many countries erect ministries to prevent words, foreign words they call them, from coming in from the outside. And it tells us that languages really are these rather discrete entities that protect themselves a little bit like a species protects itself from outside genetic influences.

 

Rissa de la Paz:

Even though globalisation isn’t changing the major languages as rapidly as we might think, the fact remains that about forty languages are going extinct every year, a rate that vastly exceeds the extinction of biological species.

 

Quentin Atkinson:

We’re losing a lot of languages in the world today. Currently there are about six thousand recognised languages but over half of those are classified as endangered and that means that they’re expected to disappear within a couple of generations. So that’s a huge loss of language diversity and something that I think a lot of people don’t appreciate. There’s a lot of emphasis on the species diversity crisis, which is definitely something we should be worried about, but there’s also a kind of cultural diversity crisis that also warrants attention I think.

 

Mark Pagel:

It isn't the case that these languages are inferior languages; they just had the misfortune of being associated with people who've been squeezed out economically or socially or politically in some way. And the question really arises what do we want to do about it .Do we want to save these languages? To the speakers of these languages there’s a great cultural tradition and temporal tradition and one associated with their artefacts that is associated with these languages, and so I think really we have to ask the people who speak these languages if they want to preserve them in some way.

 

Rissa de la Paz:

Providing the resources to preserve this linguistic diversity is a challenge for all who value our human cultural heritage.

Music: Variations on a Theme by Giuliani

Rissa de la Paz:

We’ll be looking at other aspects of cultural evolution in another podcast. Till then, we bid you a multilingual farewell.

Cecile Whidborne:

Au revoir.

Joseph Hopkins:

Adeu.

Malihe Sanatian:

Khoda hafez.

Martha Lisk:

Adios, hasta luego.

Ursula Stickler:

Auf widersehen.

Jim:

Tot ziens.

 

Rissa de la Paz:

This podcast was produced as a collaboration between the British Council and the Open University.

 

Author

Ratings

Share

Related content (tags)

Copyright information

For further information, take a look at our frequently asked questions which may give you the support you need.

Have a question?