Lingose

Connecting English pronunciation, spelling and meaning.

Lingose develops English lexical data and tools for researchers, AI developers and language learners. Our work brings pronunciation, word structure and English–Chinese meanings together, making them available through a downloadable dataset, Python and JavaScript packages, and a browser extension.

Learn about the IPSM Approach · Explore the dataset

The IPSM Approach

The IPSM Approach — Integrated Pronunciation, Spelling and Meaning Approach connects three perspectives on a word:

IPS — Integrated Pronunciation and Spelling: marks pronunciation directly on the word's spelling, keeping its letters visible.
ISM — Integrated Spelling and Meaning: uses word divisions to explore structure and meaning within Lingose's conventions.
IPM — Integrated Pronunciation and Meaning: explores connections between sound and meaning through listening and speaking.

Lingose provides IPA transcriptions alongside IPS, ISM and the IPSM representation, which combines IPS pronunciation marks with ISM divisions. This representation is one component of the wider IPSM Approach.

Our projects

Project	What it offers	Get started
IPSM English Lexicon	A public dataset with 50,335 English word entries and 23 fields: IPA, IPS, ISM, IPSM, root lemmas, parts of speech, English–Chinese meanings, vocabulary categories, available frequency ranks and field diagnostics.	Dataset and usage guide
Lingose for Python	An offline English lexicon for pronunciation and word lookup, word structure, meanings, vocabulary lists and IPS text annotation.	PyPI: `lingose`
Lingose for JavaScript / TypeScript	Offline lexical lookup and IPS text annotation for JavaScript and TypeScript applications. Published under two package names with the same functionality.	npm: `lingose` / `@englishword/lingose`
Lingose browser extension	Adds IPS pronunciation marks to English words on web pages, bringing pronunciation support into everyday reading.	Explore IPS notation

Use the data and tools

The dataset is available as Parquet, with CSV and JSONL downloads of the same records. Load it with Hugging Face Datasets:

from datasets import load_dataset

lexicon = load_dataset("Lingose/IPSM-English-Lexicon", split="train")

For application development, install the Python package with pip install lingose, or choose either npm package: npm install lingose or npm install @englishword/lingose.

Use Lingose to investigate pronunciation–spelling relationships, explore word structure and bilingual meanings, build vocabulary tools, or add IPS notation to reading interfaces. See the dataset card for field definitions, coverage, decoding rules and reproducible research guidance.

Data and interpretation

The lexicon preserves exact word keys, Unicode notation and missing-value diagnostics. ISM divisions follow Lingose's conventions and may differ from other morphological analyses. The dataset is a lexical resource; it does not establish learning-effectiveness claims or provide a benchmark evaluation split.

Original Lingose data and notation are by ZHAN Xianjue, lingose.com, under CC BY 4.0. Consult the dataset's sources and attribution and source-trace notes for third-party credits and provenance limitations.

Visit lingose.com for explanations, notation guides and learning materials.