phonotactics

A deterministic, 8-bit alternative to statistical sub-word tokenizers like BPE. Maps language to the physical rules of phonotactics to eliminate massive active embedding tables, maximize sequence information density, and optimize pre-training storage pipelines with the .pvb format.

Exploring statistical regularities in the syllable canon of Sino-Vietnamese loanmorph phonology

R package for automatic phonological report generation

A simple GUI to create and model phonological processes.

Random English-sounding (phonotacticaly correct) word generator that runs on a full-stack web app with public API. (Vue + Node + Express)

Bayesian model comparison framework for evaluating language family hypotheses (Proto-Mixe-Zoquean vs. Proto-Huastecan) for the undeciphered Isthmian (Epi-Olmec) script, using distributional phonotactic analysis without requiring proposed sign readings.

This program uses diphone (bigram) and triphone (trigram) models to represent Mainstream American English (MAE) phonotactics.

Phonetic-first language phylogeny: unified datasets (PHOIBLE, WALS, Glottolog, Lexibank…), feature builders (PanPhon/CLTS), and UMAP/HDBSCAN clustering with evaluation.

Supplementary materials for the LabPhon journal publication

Diphone probability analysis for speech perception research

🌍 Unify and analyze phonetic data to reveal language relationships with PhonoPhylo, a scalable tool for phonetic-first language phylogenetics.