Appendix 01

Using evolutionary methods of biology in linguistics - and getting results

by George Weber

It has been estimated (ref. 1) that languages through history change roughly 20% of their basic vocabulary every 1,000 years. If two languages diverge from each other at this rate, the remaining cognate words common to both languages will have shrunk

- after 1000 years to 64%

- after 2,000 years to 41%

- after 10,000 years to just over 1%

The rate of vocabulary change The half-life of a word is the amount of time required for there to be a 50% chance that it will be replaced by a new word. Most words have a half-life of 2,000 years. However, a small number of words have a half-life of greater than 10,000 years. This shows that despite the fast average pace of language evolution, some meanings, like highly-conserved genes, evolve at a slow rate. The y axis in the graphic is the number out of a sample of 200 meanings. (ref. 1)

Two languages sharing only 1% cognates with each other would, of course, be mutually unintelligible. Beyond 8,000 years it is impossible for linguists to distinguish homology (i.e. truly related words) and unrelated words that look similar merely by chance.

All attempts such as Ruhlen's claim (ref. 4) that there is linguistic evidence for a putative 12,000-year old American language family, have remained controversial, neither widely accepted nor conclusively disproved. 

A method established by M. Pagel of the McDonald Institute of Archaeology, Cambridge (ref. 6) dispenses with the assumption that linguistic change is at a steady rate and applies instead a stochastic (i.e. variable speed, depending on chance) model to lexical change, a model that has worked well in analyses of genetic change. His results show a much better fit to the available data than the steady-rate model did. He showed that there are a few words that couould remain cognates to 20,000 years.

It is on these long-lived words that research now concentrates. The problem of distinguishing long-lived cognates from chance correspondence remains, however.

Not only words change &endash; grammatical structures also changes over time. But linguists think (ref. 7) that grammatical structures change more slowly than the meaning and form of words. Linguists now hope that comparing grammars with tools long and successfully used in the biological sciences will provide the key to deep linguistic prehistory and to establishing the ancient evolutionary relationships of languages.

As Charles Darwin has pointed out in 1871 (ref. 9), languages evolve in a remarkably similar way as biological species. Languages, too, tend to split into different, new languages (through the intermediate stages of dialects) , mutate and survive or become extinct. Linguists, so far, have tended to shy away from using the phylogenetic methods that have revolutionized evolutionary biology over the past decades. This is now about to change.

One group who has begun to apply the phylogenetic methods of data analysis has now presented ist results. Dunn at al (ref. 3) have borrowed two tools from biology:

(1) they constructed a database of 125 structural (grammatical) features from 16 Austronesian and 15 Papuan languages.

(2) they used a computer program (used so far mostly by biologists) to establish a set of optimal trees for their linguistic data sets.

They then compared the result with the conventional linguistic classification of the Austronesian language family &endash; and the two trees matched remarkably well.

This result suggests that the linguistic structural features do indeed contain historical links going back at least 4,000 years . This is the age that the Austronesian language family is thought to have.

The Papuan language family is certainly much older, though it is an open question just how much older. Archaeological finds documenting the earliest human settlements in the area go back to more than 30,000 years. Of course, nobody can say what languages the first modern humans in the area spoke. Butt he new phylogenetic analysis of these languages produced a tree that, though weak in ist oldest parts, is consistent with an age of more than 10,000 years. Even more astonishing ist he fact that the tree shows the presently correct geographic clustering at the top (i.e. most recent in time). In other words, the present-day Papuan languages most closely related to each other have been in their present location for a very long time and there has been little migration.

For a method that has not been used for this purpose before and will no doubt be capable of much further refinement in future, this is a splendid result that promises well for future studies on many other language families, including, we hope, the Andamanese language family. 

Linguists have been given a new tool to penetrate much further into the prehistoric past. As Russell Gray notes (ref. 1):

The Dunn et al paper sets new standards for the systematic collection and analysis of structural features. Its approach is likely to be widely emulated by researchers working on languages in other regions. In the future, we may see the development of web-based databases for the languages of the world similar to the GenBank repository for DNA sequences. The task of making accurate inferences about our past is a demanding one that requires the integration and triangulation of inferences from genetic, linguistic and archaeological data. The Dunn et al approach is an important step forward in this interdisciplinary endeavor. 

For a summary of the detailed results by Dunn M. et al see our chapter on Papua-Newguinea.

References

1. Gray R. 2005. „Pushing the Time Barrier in the Quest for Language Roots." Science 309:2007-2008

2. Dunn M., Terrill A., Reesink G., Foley R.A., and Levinson S.C. 2005. "Structural Phylogenetics and the Reconstruction of Ancient Language History", Science, 309:2072-2075

3. Ruhlen M. 1994. The Origin of Language: Tracing the Origin of the Mother Tongue. Wiley, New York

4. Pagel M. 2000. In: Time Depth in Historical Linguistics, eds. Renfrew C., McMahon A. and Trask L., pp. 189-207. McDonald Institute for Archaeological Research, Cambridge

5. Nichols J. 1992. Linguistic Diversity in Space and Time. University of Chicago Press, Chicago

6. Darwin C. 1871. The Descent of Man, London, Murray

[ Go to HOME ]

   

Last changed 15 October 2005