# PhiloBERTA: A Transformer-Based Cross-Lingual Analysis of Greek and Latin Lexicons

arXiv: https://arxiv.org/abs/2503.05265 · HTML: https://mlsanity.com/paper/2503.05265/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 3 of its 16 references that are in the map.
- First posted on arXiv 7 Mar 2025.

## Builds on (most cited first, 3 of 3)

- [Making Monolingual Sentence Embeddings Multilingual using Knowledge Distillation](https://mlsanity.com/paper/2004.09813/) (2020)
- [Sentence Embedding Models for Ancient Greek Using Multilingual Knowledge Distillation](https://mlsanity.com/paper/2308.13116/) (2023)
- [Lexical semantic change for Ancient Greek and Latin](https://mlsanity.com/paper/2101.09069/) (2021)

## Abstract (arXiv)

We present PhiloBERTA, a cross-lingual transformer model that measures semantic relationships between ancient Greek and Latin lexicons. Through analysis of selected term pairs from classical texts, we use contextual embeddings and angular similarity metrics to identify precise semantic alignments. Our results show that etymologically related pairs demonstrate significantly higher similarity scores, particularly for abstract philosophical concepts such as epist=em=e (scientia) and dikaiosyn=e (iustitia). Statistical analysis reveals consistent patterns in these relationships (p = 0.012), with etymologically related pairs showing remarkably stable semantic preservation compared to control pairs. These findings establish a quantitative framework for examining how philosophical concepts moved between Greek and Latin traditions, offering new methods for classical philological research.