This page (still in progress) shows a preliminary sample of our studies in Historical Glottometry.
The Torres–Banks islands, in northern Vanuatu, are home to seventeen different languages (see François 2014):
Several of these languages are endangered. François has collected fieldwork data on these languages since 1997 (see his homepage).
Here is a small sample of historical isoglosses from that Torres–Banks area:
a) Regular sound change: | *r > /j/ |
b) Irregular sound change: | *malate → *malete ‘broken’ |
c) Irregular sound change: | *ʔaŋaRi → *ʔaŋai ‘almond’ |
d) Morphological change: | metathesis in trial pronouns (Plural+three → three+Plural) |
e) Morphological change: | *toɣa ‘stay’ → Prohibitive |
When represented on a map, these innovations define isoglosses which clearly intersect:
This intersection of historical isoglosses is constantly observed in this region (as well as, in our view, in most language families in the world).
In 2011-12, François collected information on 474 historical innovations based on the principles of the Comparative Method. He then encoded their presence or absence in the 17 languages of Vanuatu. Here is what the coded data looks like:
Based on this data, Siva Kalyan proposed an analysis in terms of “cohesiveness” and “subgroupiness” (see Kalyan & François, f/c). Essentially, these are two ways to measure the degree of cohesion of a genealogical subgroup: the more innovations were shared by its members together, the more that subgroup is “cohesive” and “subgroupy”.
After a long analysis which cannot be detailed here, we were able to rank the language groups with highest subgroupiness in this region:
Subgroup | Subgroupiness |
---|---|
Volow–Mwotlap | 12.82 |
Hiw–Lo-Toga | 12.45 |
Vurës–Mwesen | 9.34 |
Lemerig–Vera’a | 6.78 |
Koro–Olrat–Lakon | 6.63 |
Dorig–Koro–Olrat–Lakon | 6.01 |
Olrat–Lakon | 5.34 |
…and so forth, for all the genealogical subgroups attested in the area, whether they intersect or not. The final result was the following Glottometric diagram. It shows the patterns of distribution of historical innovations in this group of languages; the more cohesive a subgroup is, the thicker is the line around it.
Languages are here shown from northwest (left) to southeast (right). Languages adjacent on the diagram are also adjacent geographically (though the reverse is not true).
The crucial observation is that the genealogical subgroups which can be identified (based on the Comparative method) commonly intersect: this runs contrary to most people’s expectations, due to the assumptions tied to the Tree model. In fact, nothing justifies that language genealogy should happen following a tree-like (or “cladistic”) division, with discrete subgroups; instead, language genealogy evidently involves intersecting subgroups. While a family tree is incapable of representing this situation in any satisfactory way, Historical Glottometry is a methodology designed precisely with the purpose of capturing the historical reality of language families.
*
We hope to enrich this page in the near future, with more illustrations of our methodology.
Watch this space!