MATH · IN · MODELS

Phonological features are linear vector-offset directions in speech models

measured in 1 paper

Choi et al. extract phoneme-level representations from three frozen self-supervised speech models (wav2vec2-large-lv60, HuBERT-large-ll60k, WavLM-large) [choi-etal-2026-phonological-vector-arithmetic-speech-ssl] Phonological feature contrasts (voicing, place, manner) are encoded as approximately consistent diff-in-means directions that transfer across phoneme pairs [choi-etal-2026-phonological-vector-arithmetic-speech-ssl] Word2vec-style vector-offset arithmetic (e.g. [b] = [d] - [t] + [p]) succeeds above chance on the real extracted representations, with per-model accuracy reported [choi-etal-2026-phonological-vector-arithmetic-speech-ssl] The result is a quantified linear-direction/vector-offset claim on real speech models; no causal intervention is performed [choi-etal-2026-phonological-vector-arithmetic-speech-ssl]

Context

phonological feature vectors, cross-model transfer, vector-offset arithmetic

Papers

[b]=[d]-[t]+[p]: Self-Supervised Speech Models Discover Phonological Vector Arithmetic — Choi, Kwanghee, Yeo, Jaemin, Cho, Eunjung, Harwath, David, Mortensen, David R.2026 · arXiv:2602.18899