Finding Words in Speech: An Investigation of American English

Abstract

Two computational analyses of phoneme transitions are presented for American English. A strong bimodal distribution was found such that any given phoneme pair is expected to occur either only across a word boundary or only within a word. This is a type of regularity that language learners have been shown to be particularly adept at detecting. This result indicates that highly accurate word segmentation of American English speech is possible by a system that attends only to the statistics of single phoneme transitions.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 39 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.