Final Report Summary - LANGELIN (Meeting Darwin's last challenge: toward a global tree of human languages and genes)
For two centuries now historical linguists have tried to reconstruct the history and genealogical relations of languages mostly comparing their sounds and words. Despite their successes, the language families so discovered are relatively circumscribed in scope and time depth. As a consequence, linguistics has failed to meet the challenge proposed by biological anthropology (Darwin, Cavalli Sforza) of measuring the congruence of languages with the genes of their speakers; this is so because genetics can reconstruct relations/events spanning over continent-wide or cross-continental domains and over long stretches of time.
The original intuition behind the LANGELIN ERC Advanced Grant Project is that the comparison of language differences and similarities in syntax (the mental rules and principles combining words into grammatical sentences) may remedy the (relative) lack of success of sounds and words as tools to reconstruct long-range linguistic relations. So far, hardly any work has tried to use syntax to historically classify languages. This has remained largely true even with the development of formal models of grammatical diversity in the cognitive science of the past 40 years: theoretical linguists have reiterated the claims that syntax is ineffective for reconstructing language phylogenies.
The recent Parametric Comparison Method has defied such claims. The Langelin project started with three innovative objectives, all revolving around this original idea of using syntax to better reconstruct human history:
1) proving that the syntax of human languages contains a historical signal, in other words that syntactic differences alone can identify with accuracy the known language families;
2) showing that the historical signal encoded in syntax extends beyond the boundaries of established language families and can be used for cross-family comparisons;
3) exploring long-range correlation between syntactic diversity and genetic diversity.
The results we obtained discovered large support for the unconventional conclusions in 1) and 2) above, against the traditional scepticism about historical syntax. Syntax proved to retrieve deeper insights into the history of languages than many methods comparing words or sounds.
Our work required the construction of a unique dataset of 95 abstract syntactic differences, encoded as binary distinctions (syntactic parameters), among more than 80 contemporary languages belonging to at least 15 language families from Europe, Asia and partly from Africa and America. These parameters affect a unique module of the syntax of these languages, namely the structure of nominal phrases. The Parametric Comparison Method was used to calculate various types of distances among these languages and to perform quantitative experiments, both on the global language set and also on specific subdomains of closely related languages.
On this basis, we could address point 3). Correlations between syntactic and genetic distances were investigated, first in Europe and then across Eurasia. We discovered macro-areas of the two continents (especially Western Europe and East Asia) with high congruence of genetic and linguistic diversity: in these regions, syntax turned out to predict the degree of genetic diversity of the populations of speakers better than does their mere geographical position. In other regions, the distribution of syntactic and genetic similarity diverges. These discoveries also led to some preliminary generalizations about ‘glottogenetic history’, holding at least in such domains: for example, in areas of extended admixture of peoples, genetic uniformation of populations is stronger than convergence and assimilation of languages.
The results proved the possibility of applying quantitative methods of biostatistical derivation to the output of formal syntactic analyses: this suggests that modern cognitive theories of language can interact and side profitably with genetic anthropology in the attempt to establish a scientific approach to long-term human history.
The original intuition behind the LANGELIN ERC Advanced Grant Project is that the comparison of language differences and similarities in syntax (the mental rules and principles combining words into grammatical sentences) may remedy the (relative) lack of success of sounds and words as tools to reconstruct long-range linguistic relations. So far, hardly any work has tried to use syntax to historically classify languages. This has remained largely true even with the development of formal models of grammatical diversity in the cognitive science of the past 40 years: theoretical linguists have reiterated the claims that syntax is ineffective for reconstructing language phylogenies.
The recent Parametric Comparison Method has defied such claims. The Langelin project started with three innovative objectives, all revolving around this original idea of using syntax to better reconstruct human history:
1) proving that the syntax of human languages contains a historical signal, in other words that syntactic differences alone can identify with accuracy the known language families;
2) showing that the historical signal encoded in syntax extends beyond the boundaries of established language families and can be used for cross-family comparisons;
3) exploring long-range correlation between syntactic diversity and genetic diversity.
The results we obtained discovered large support for the unconventional conclusions in 1) and 2) above, against the traditional scepticism about historical syntax. Syntax proved to retrieve deeper insights into the history of languages than many methods comparing words or sounds.
Our work required the construction of a unique dataset of 95 abstract syntactic differences, encoded as binary distinctions (syntactic parameters), among more than 80 contemporary languages belonging to at least 15 language families from Europe, Asia and partly from Africa and America. These parameters affect a unique module of the syntax of these languages, namely the structure of nominal phrases. The Parametric Comparison Method was used to calculate various types of distances among these languages and to perform quantitative experiments, both on the global language set and also on specific subdomains of closely related languages.
On this basis, we could address point 3). Correlations between syntactic and genetic distances were investigated, first in Europe and then across Eurasia. We discovered macro-areas of the two continents (especially Western Europe and East Asia) with high congruence of genetic and linguistic diversity: in these regions, syntax turned out to predict the degree of genetic diversity of the populations of speakers better than does their mere geographical position. In other regions, the distribution of syntactic and genetic similarity diverges. These discoveries also led to some preliminary generalizations about ‘glottogenetic history’, holding at least in such domains: for example, in areas of extended admixture of peoples, genetic uniformation of populations is stronger than convergence and assimilation of languages.
The results proved the possibility of applying quantitative methods of biostatistical derivation to the output of formal syntactic analyses: this suggests that modern cognitive theories of language can interact and side profitably with genetic anthropology in the attempt to establish a scientific approach to long-term human history.