Transformed Protoform Reconstruction

TL;DR

This paper enhances protoform reconstruction by applying a Transformer-based seq2seq model, outperforming previous RNN models on Latin and Chinese datasets, and explores phylogenetic signals in the model.

Contribution

It introduces a Transformer model for protoform reconstruction, achieving better performance than prior RNN-based models and analyzing phylogenetic signals.

Findings

Transformer outperforms RNN models on multiple datasets

Model achieves state-of-the-art metrics on Latin and Chinese datasets

Phylogenetic signals are detectable in the model

Abstract

Protoform reconstruction is the task of inferring what morphemes or words appeared like in the ancestral languages of a set of daughter languages. Meloni et al. (2021) achieved the state-of-the-art on Latin protoform reconstruction with an RNN-based encoder-decoder with attention model. We update their model with the state-of-the-art seq2seq model: the Transformer. Our model outperforms their model on a suite of different metrics on two different datasets: their Romance data of 8,000 cognates spanning 5 languages and a Chinese dataset (Hou 2004) of 800+ cognates spanning 39 varieties. We also probe our model for potential phylogenetic signal contained in the model. Our code is publicly available at https://github.com/cmu-llab/acl-2023.