Explore our peer-reviewed publications, open datasets, and linguistic methodologies driving the future of NLP and multilingual AI.
We propose a transformer-based architecture that generates contextual definitions for morphologically complex languages using cross-lingual transfer learning. Achieves 18% improvement over baseline LLMs on zero-shot evaluation.
An open, crowdsourced alignment dataset containing semantic triples, phonetic mappings, and usage contexts. Designed to benchmark cross-lingual embedding models and support dictionary construction pipelines.
Longitudinal analysis of lexical meaning shifts across social media platforms (2010–2024). Reveals accelerated semantic broadening in tech-adjacent terminology and proposes a computational drift metric.
Addresses the limitations of ASCII-based phonetic transcription for tonal languages. Introduces a tone-aware mapping layer that improves pronunciation accuracy by 32% in ASR pipelines.
A novel approach to building dense synonym networks by distilling contextual embeddings from LLMs into lightweight graph models. Enables real-time thesaurus queries with 99.4% precision.
A structured dataset tracking lexical origins across 200 years of printed corpora. Includes proto-language roots, borrowing pathways, and semantic evolution timelines.
Our work bridges computational linguistics, AI, and open science to build more accurate, inclusive language tools.
Improving definition generation, synonym mapping, and contextual disambiguation for language models.
Building resources for underrepresented languages to close the lexical gap in global AI systems.
Curating, aligning, and publishing high-quality linguistic datasets for academic and commercial use.
Advancing pronunciation modeling, tone recognition, and dialect-aware speech synthesis.
Led by computational linguists, data scientists, and lexicographers committed to open research.
Head of NLP Research
Computational Linguist
Corpus Linguistics Lead
Phonetics & Speech AI