Using support vector machines and state-of-the-art algorithms for phonetic alignment to identify cognates in multi-lingual wordlists

Abstract

Most current approaches in phylogenetic linguistics require as input multilingual word lists partitioned into sets of etymologically related words (cognates). Cognate identification is so far done manually by experts, which is time consuming and as of yet only available for a small number of well-studied language families. Automatizing this step will greatly expand the empirical scope of phylogenetic methods in linguistics, as raw wordlists (in phonetic transcription) are much easier to obtain than wordlists in which cognate words have been fully identified and annotated, even for under-studied languages. A couple of different methods have been proposed in the past, but they are either disappointing regarding their performance or not applicable to larger datasets. Here we present a new approach that uses support vector machines to unify different state-of-the-art methods for phonetic alignment and cognate detection within a single framework. Training and evaluating these method on a typologically broad collection of gold-standard data shows it to be superior to the existing state of the art.

Predicting Historical Phonetic Features using Deep Neural Networks: A Case Study of the Phonetic System of Proto-Indo-European

“…Using machine learning algorithms is not new to the field of linguistics, though it is one of the more recent methods. 1 While these approaches are found in an increasing number of studies in lin-guistics in general, in historical linguistics in particular the method is less used although some studies have been published in this or adjacent fields such as cladistics (Jäger et al, 2017;Jäger and Sofroniev, 2016). Since this approach of predicting sound features by the features in the phonetic environment only works synchronically, the deep neural network used for this needs to be trained on better known phonological features as the basis for predicting unknown features.…”

Global-scale phylogenetic linguistic inference from lexical resources

“…We created a goldstandard dataset from the data used in [22] (which is is drawn from the same sources as the data used in [21] but has been manually edited to correct annotation mistakes). Only the 40 ASJP concepts were used.…”

Beyond edit distances: Comparing linguistic reconstruction systems

“…B-Cubed scores offer a straightforward way to compare partitioning analyses (or cluster analyses) with each other. In the task of automatic cognate detection in computational historical linguistics, for example, B-Cubed scores are frequently used to compare how well an algorithm performs in comparison with a gold standard (Hauer and Kondrak 2011;Jäger, List, and Sofroniev 2017;List, Greenhill, and Gray 2017).…”