An Automated Framework for Fast Cognate Detection and Bayesian Phylogenetic Inference in Computational Historical Linguistics

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 6225–6235 Florence, Italy, July 28 - August 2, 2019. c⃝2019 Association for Computational Linguistics 6225 An automated framework for fast cognate detection and Bayesian phylogenetic inference in computational historical linguistics Taraka Rama Department of Linguistics University of North Texas taraka.kasi@gmail.com Johann-Mattis List Dep. of Ling. and Cult. Evolution (DLCE) MPI-SHH (Jena) list@shh.mpg.de Abstract We present a fully automated workflow for phylogenetic reconstruction on large datasets, consisting of two novel methods, one for fast detection of cognates and one for fast Bayesian phylogenetic inference. Our results show that the methods take less than a few minutes to process language families that have so far re- quired large amounts of time and computa- tional power. Moreover, the cognates and the trees inferred from the method are quite close, both to gold standard cognate judgments and to expert language family trees. Given its speed and ease of application, our frame- work is specifically useful for the exploration of very large datasets in historical linguistics. 1 Introduction Computational historical linguistics is a relatively young discipline which aims to provide automated solutions for those problems which have been traditionally dealt with in an exclusively manual fashion in historical linguistics. Computational historical linguists thus try to develop automated approaches to detect historically related words (called “cognates”; J¨ager et al. 2017; List et al. 2017; Rama et al. 2017; Rama 2018a), to infer lan- guage phylogenies (“language trees”; Rama et al. 2018; Greenhill and Gray 2009), to estimate the time depths of language families (Rama, 2018b; Chang et al., 2015; Gray and Atkinson, 2003), to determine the homelands of their speakers (Bouckaert et al., 2012; Wichmann et al., 2010), to determine diachronic word stability (Pagel and Meade, 2006; Rama and Wichmann, 2018), or to estimate evolutionary rates for linguistic features (Greenhill et al., 2010). Despite the general goal of automating tradi- tional workflows, the majority of studies con- cerned with phylogenetic reconstruction (includ- ing studies on dating and homeland inference) still make use of expert judgments to determine cog- nate words in linguistic datasets, because detect- ing cognates is usually regarded as hard to auto- mate. The problem of manual annotation is that the process is very time consuming and may show a lack of objectivity, as inter-annotator agreement is rarely tested when creating new datasets. The last twenty years have seen a surge of work in the development of methods for automatic cog- nate identification. Current methods reach high accuracy scores compared to human experts (List et al., 2017) and even fully automated workflows in which phylogenies are built from automatically inferred cognates do not differ a lot from phylo- genies derived from expert’s cognate judgments (Rama et al., 2018). Despite the growing amount of research de- voted to automated word comparison and fully au- tomated phylogenetic reconstruction workflows, scholars have so far ignored the computational effort required to apply the methods to large amounts of data. While the speed of the current …

Content truncated to 500 words