Grammarticle

GrammArticle is a RoBERTa-based grammar checker for English article usage. It detects three types of article errors:

1. Missing – when an article is absent but required
2. Wrong – when an incorrect article is used (e.g., "a apple" instead of "an apple", or "the" instead of "a/an")
3. Redundant – when an article is unnecessary (e.g., "the furniture")

Installation

GrammArticle is trained on publicly available GEC datasets with synthetic augmentation and is available as a SpaCy pipeline.

pip install grammarticle

or

pip install spacy-transformers
python -m spacy download en_core_web_trf
pip install https://huggingface.co/iproskurina/en_grammarticle/resolve/main/en_grammarticle-1-py3-none-any.whl

Usage

import grammarticle
nlp = grammarticle.load()
text = "This is sentence"
doc = nlp(text)
for span in doc.spans.get("sc", []): 
    print(f"[{span.label_}] {span.text}")

Outputs:

[Missing] sentence

---

Easy-to-use examples can be found in the /examples folder:
annotate_texts.py for text annotation and html_render.py for HTML highlighting (see example below).

---