Open Research Europe presents an interview with Prof. Johann-Mattis List, Chair of Multilingual Computational Linguistics at the University of Passau, where he heads the ERC-funded research group “ProduSemy“. He is the Gateway Advisor of the Languages and Literature Community Gateway on Open Research Europe.
This Community Gateway aims to provide a space to publish the latest European Commission-funded language research, research on modern languages and linguistics as the scientific study of the structure and development of human language. The Gateway also includes research on Literature, including literary history, genres, aesthetics and literary discourse.
Read on to find out more about the work of Prof. List, the challenges facing his field, and why he thinks that open science is important for the field of linguistics.
Please describe your specific area of research and how it relates to the wider research available in the Languages and Literature Gateway.
In my work, I try to develop automatic methods to compare languages in order to understand how they evolved, how they differ with respect to their structure, and which factors contribute to the emergence and distribution of linguistic diversity. As linguists, we use a mixture of different computational methods, ranging from classical algorithms that we design or transfer from different fields in order to tackle individual problems in historical linguistics and linguistic typology, up to methods from machine learning.
What are the biggest challenges of being a researcher in your field?
The biggest challenge lies in the evaluation and curation of linguistic data. Datasets for more than 7000 languages are still in use today and have been published in digital form. But these are often difficult to compare, since the scholars who create them often do not consider the potential reuse of their data and therefore ignore existing standards. Standardizing these datasets is tedious, since there are so many.
Additionally, we must evaluate how robust these datasets are to solve problems in our field. This task is much more difficult. It is often unclear how well the data we use to approach the construct of language reflects the spoken or signed languages out there.
What are the important topics in your field? Have you noticed any new areas of study emerging in your field?
We look at language from three perspectives, from the perspective of evolution, typology, and from the perspective of the language users. New topics that have emerged in the past years which attempt to combine these perspectives. They seek to understand how specific typological characteristics of certain languages have evolved over time, or trying to understand to which degree the conditions in which people speak (their culture or their environment) constrain certain developments.
Another important topic is that of computational methodology. While earlier approaches to historical linguistics (dating back more than 200 years) were based on manual data inspection, our interactive tools and computational methods are now getting closer to modelling these manual approaches, allowing us to work in a computer-assisted way for our comparative work.
What aspects of open science are particularly beneficial to your field of research?
Our field has made significant progress in the past decade thanks to open data and open research (in the form of shared code). Both open data and open research are integral to our research as we need to be able to combine individual datasets to carry out new studies or to modify existing datasets in the search for new patterns. We also need open-source code to both control what colleagues do and how we learn from them.
As linguists, we use a variety of open-source tools to assist our research. LingPy (https://lingpy.org) is a free software library that offers state-of-the-art-algorithms for computer-assisted language comparison, that we have been maintaining for more than 10 years now. Additionally, Cross-Linguistic Data Formats (https://cldf.clld.org) allows us to establish standards for the exchange of cross-linguistic datasets in historical linguistics and linguistic typology that have allowed us to publish large cross-linguistic databases covering several thousand different language varieties.
Are there any key challenges when it comes to participating in open science that are specific to your research area?
In 2024, participating in open science is fairly easy. With Zenodo, there is a large repository where we can host our data in Europe for free. Using basic programming languages like Python is much easier than ten years ago, and there are numerous tutorials on the internet that help people to get started. The problem is that scholars often lack the awareness that the code they write and the data they curate should be treated with the same care as the words and phrases they put in their articles. As a result, code is often written in a form that cannot be used on foreign computers, and datasets are poorly described.
Explore more of the Languages and Literature Community Gateway today.
