Phylomilia – Phylogenetic Linguistic Inference from Acoustic Speech Data

Welcome to the Phylomilia Project

Exploring the evolution of languages through advanced deep learning methodologies.

Project Steps

About

The Phylomilia project, named after ‘phylogeny’ (evolutionary family tree) and the Greek ‘omilia’ for speech, proposes a groundbreaking method to investigate the evolution of languages using speech data directly. Traditionally, comparative linguistics depends on written data, which requires laborious manual transcription and annotation, limiting the scope and speed of research – not least with regard to under-documented languages. Phylomilia aims to transform this field by applying sophisticated Deep Learning techniques to analyze acoustic speech data, thereby bypassing the need for manual transcriptions and annotations. This innovative approach will use advanced AI-based speech recognition technology to convert spoken language into a format that machines can analyze, then employ cutting-edge statistical algorithms to trace the history of languages and reconstruct earlier language stages. By focusing on spoken language, Phylomilia will capture nuances and details lost in written records, providing deeper insights into language history.

The ultimate goal is to develop an automated system that can identify and interpret features of spoken languages remaining consistent over long stretches of time, thereby offering a new perspective on how languages evolve and interact. This project not only aims to advance linguistic research but also to enhance our understanding of human cultural heritage.

Funded by the Volkswagen Foundation

Team

Principal Investigator

Prof. Dr. Gerhard Jäger, University of Tübingen.

Job Announcements

We are currently hiring a postdoc positions. Please see the job announcement for more details.

Contact

For more information, contact Prof. Dr. Gerhard Jäger at gerhard.jaeger@uni-tuebingen.de