Approaching Reflex Predictions as a Classification Problem Using Extended Phonological Alignments

TL;DR

This paper introduces an extended alignment approach that transforms cognate reflex prediction into a classification task using multilayered phonological information, evaluated with a random forest model.

Contribution

It presents a novel multilayered alignment technique for reflex prediction, generalizing the problem as classification and demonstrating its effectiveness with a random forest model.

Findings

The method achieves competitive reflex prediction accuracy.

Multilayered alignments improve phonological feature encoding.

The approach offers a flexible framework for cognate analysis.

Abstract

This work describes an implementation of the "extended alignment" (or "multitiers") approach for cognate reflex prediction, submitted to "Prediction of Cognate Reflexes" shared task. Similarly to List2022d, the technique involves an automatic extension of sequence alignments with multilayered vectors that encode informational tiers on both site-specific traits, such as sound classes and distinctive features, as well as contextual and suprasegmental ones, conveyed by cross-site referrals and replication. The method allows to generalize the problem of cognate reflex prediction as a classification problem, with models trained using a parallel corpus of cognate sets. A model using random forests is trained and evaluated on the shared task for reflex prediction, and the experimental results are presented and discussed along with some differences to other implementations.