lemminflect

What it is and what it does

LemmInflect is a standalone English lemmatization and inflection library that converts words to their base forms or inflected variants based on part-of-speech tags. It combines a dictionary approach (using the NIH SPECIALIST Lexicon) with neural network rules to handle out-of-vocabulary words, and can work independently or as a spaCy extension. The library accepts Universal Dependencies or Penn Treebank tags to specify the target word form and returns a tuple of possible spellings.

The package depends only on numpy for its neural network computations. It is designed for Python 3 and runs on Linux, Windows, and macOS. The description reports lemmatization accuracy of 95.6% overall on a test set of 119,194 inflected words, though the library acknowledges inherent ambiguities in English morphology (e.g., context-dependent forms like 'further' vs. 'farther', or verb 'be' forms that cannot be fully disambiguated by tag alone).

Use it for:

• Normalize inflected words to base forms in text preprocessing pipelines for NLP or information retrieval tasks.
• Generate specific word forms (past tense, plurals, adjective variants) programmatically for text generation or templating.
• Extend spaCy NLP workflows with lemmatization and inflection methods on individual tokens without replacing spaCy's lemmatizer.
• Handle out-of-vocabulary words by applying morphological rules when dictionary lookup fails.
• Build rule-based text analysis systems that need reliable English morphology without heavy ML dependencies.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Lemmatizes and inflects English words using dictionary lookup and neural network rules, with support for Universal Dependencies and Penn Treebank tags, and optional integration with spaCy.

Yes, if you need reliable English lemmatization and inflection with low dependencies. The permissive MIT license and pure-Python wheel make installation straightforward. However, the dormant maintenance status (last update 2022-10-02) means no active development or bug fixes; install only if the current feature set meets your needs and you can tolerate a static codebase. The reported accuracy is solid for standard use cases, but verify it against your specific word forms before committing to production.

Install

lemminflect on PyPI

pip install lemminflect

uv add lemminflect

poetry add lemminflect

Installing lemminflect

Before you install

Low friction—only numpy as a runtime dependency, distributed as a pure Python wheel. Maintenance is dormant (last commit 2023-09-14, 1412 days since release), so expect no active bug fixes or feature updates.

License in practice

MIT license (permissive) allows commercial and private use with minimal restrictions.

Quickstart

pip install lemminflect

from lemminflect import getLemma, getInflection

# Lemmatize
getLemma('watches', upos='VERB')
# ('watch',)

# Inflect
getInflection('watch', tag='VBD')
# ('watched',)

Verify before relying

• Actual accuracy on modern English corpora outside the AGID baseline; whether the 96.1% verb accuracy still holds in production use.
• Whether dormant maintenance affects real-world reliability or if the library is stable enough for long-term production use without updates.
• Support for Python versions beyond 3; the description says 'untested under Python 2' but does not specify a minimum or maximum Python 3 version.

Package facts

Evidence: lemminflect-0.2.3-py3-none-any.whl

Tags

More Linguistic packages