Frederik Hartmann
University of North Texas, TX·Social Science
Every modern language descends from older languages that were never written down. One example is the shared ancestor of English, German and Swedish.
- Fields
Social Science
Computer Science
- Topics
historical linguistics
comparative method
proto-language reconstruction
- Grants
- 1 known grant, through Aug 2028
What they work on
- Build artificial neural networks that learn to reconstruct proto-forms, the ancestor words of a language family, from their descendant words, using large datasets spanning many language families.
- Compare different neural model types, network architectures and dataset-construction approaches to find which best capture the regular sound correspondences used in comparative reconstruction.
- Use cross-linguistic patterns learned from multiple language families to improve inference of ancestor forms beyond what a single family's data provides.
- Apply the trained reconstruction models to current, unresolved reconstruction problems in various language families.
More about the research
Rebuilding these lost "proto-languages" shows how languages change over time, and it sheds light on the histories of the peoples who spoke them. Linguists have long done this with the comparative method. They line up words in related languages, find sounds that correspond regularly, and work backward to the ancestral forms. The method has several limitations as it is practiced today, and it is hard to apply to very large and varied collections of language data.
Why it matters
Reconstructing ancestor languages helps scholars trace how languages change and how human populations and societies developed. Neural reconstruction tools could let historical linguists work with far larger and more diverse data than the hand-applied comparative method allows. These tools may help address reconstruction problems that have remained open in several language families. The work also adds to AI tools developed for use in the sciences.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Kyle Mahowald
University of Texas at Austin, TX·Social Science
Measure how speakers of a variety of languages choose deictic words such as "this," "that," "here," and "there" to refer to objects and locations, using cross-linguistic behavioral experiments.
Elena Koulidobrova
Central Connecticut State University, CT·Social Science
Build treebanks, collections of sentences annotated for grammatical structure in the Universal Dependencies format, that capture how evidential markers (grammatical signals of information source) and focus features appear in sentences.
Jonathan Amith
Gettysburg College, PA·Computer Science
Build POWSM, a foundation speech model that transcribes any language as sequences of consonants and vowels, and prompt it with vector representations of sound correspondences so it can recognize low-resource varieties as transformed versions of well-resourced ones.
The school
University of North Texas: 4 labs on PI Pages, $124M a year on research, 283 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.