Stylometry
The statistical analysis of writing style. Historically used for author identification.
Stylometry is the statistical analysis of writing style. Historically, it's been used to identify the authors of anonymous texts. Disputed Shakespeare plays, the Federalist Papers, anonymous online posts. It analyzes features like average sentence length, function-word frequency, punctuation patterns, and vocabulary richness.
Stylometric features are some of the most stable signals available to AI detection. Even a careful prompt can't easily strip the statistical fingerprints of an LLM's training distribution. AI Detector API uses stylometric features alongside perplexity-based signals and classifier confidence.
The downside: stylometry can encode demographic or cultural patterns, so a detector that leans too heavily on it may show bias against non-native English speakers. We tune to minimize this. But no detector is perfectly neutral.
Related terms
- Burstiness· A measure of variation in sentence length, structure, and complexity across a piece of text.
- Perplexity· How surprised a language model is by a given piece of text. Lower means the text looks more model-generated.
- AI detection· The task of identifying text that was written by a large language model rather than a human.
Move from definition to code
Free 1,000 requests/month, no credit card. Be detecting AI text in 5 minutes.