Vortrag: Open Data in Computational Historical Linguistics: Upcycling of Chibchan wordlists
This work presents presents an ongoing research project on the phylogeny of the Chibchan language family which adheres to the current state-of-the-art of computational methods in Historical Linguistics. The aim is to show how we can upcycle already published data and infer new results by applying models of Bayesian inference (or others). In times of a pandemic, these seem to be reasonable methods to gather data when we cannot do fieldwork and the data can be used for a huge variety of goals.
Link to data on GitHub: https://github.com/lexibank/constenlachibchan/releases/tag/v0.2
The following steps of the workflow will be introduced:
i. Collecting already published data for the Chibchan family (Constenla Umaña 2005)
ii. Applying cross-linguistic data format standards (Forkel & List 2020)
iii. Detection of cognacy and working with Edictor (List 2017)
iv. phylogenetic analysis through Bayesian inference (Greenhill et al. 2020)
All data of this project was published by Constenla Umaña (2005) and will be upcycled
into the Cross-Linguistic Data Format (Forkel & List 2020) which adheres to certain standards which ensure comparability across languages. In the next step, this data is first analyzed automatically for cognates and then manually corrected where necessary (List 2017). In a final step, the cognates will be analyzed with computational methods, such as NeighborNetworks and Bayesian inference of phylogenetic trees (Greenhill et al. 2020).
The main goal of this presentation is to show how computational methods can enrich our
workflows also in more ‘traditional’ areas of linguistics and how we can upcycle old data to arrive at new conclusions about the languages of the world.