Free academic integrity tool · 6 stylometric signals

Free authorship and writing style checker for teachers

Paste a known authentic classroom sample from a learner alongside their new submission to compare sentence rhythm, vocabulary diversity, contractions, openers, and punctuation habits.

6 interpretable style dimensions 100% deterministic · no AI training Built for supportive check-in conversations

Compare baseline sample vs new submission

Paste at least 40 words in each box (120+ words each recommended for steady stylometrics).

Example output (crawlable, no JavaScript)

Sample authorship report: GCSE Biology practical write-up vs homework

Style similarity: 19% — Marked style shift (<45%)

Multiple stylometric dimensions diverge from the baseline sample. Review the dimension cards below and hold a supportive check-in conversation with the learner about how they drafted the piece.

Sentence length & rhythm Consistent
Baseline: 22.5 w/sentence (variation 0.14)
Submission: 19.2 w/sentence (variation 0.24)
Sentence pace and rhythm variability match the baseline closely.
Vocabulary range (MATTR) Inconsistent
Baseline: 88.8% unique per 50 words
Submission: 97.4% unique per 50 words
Vocabulary density is markedly denser or flatter than the baseline sample.
Register & contractions Inconsistent
Baseline: 3.33/100w contractions · 7.78/100w first-person
Submission: 0.0/100w contractions · 0.0/100w first-person
Sharp register shift (for example natural contractions disappearing into formal prose).
Sentence openers & connectives Inconsistent
Baseline: 25.0% connective starts · 0.0/100w hedges
Submission: 75.0% connective starts · 2.6/100w hedges
Heavy shift toward or away from formulaic connective starters and hedging phrases.
Punctuation & paragraph symmetry Moderate shift
Baseline: 0.0 typographic marks/sentence
Submission: 0.5 typographic marks/sentence
Minor difference in punctuation or paragraph structure.
Phrase & opener repetition Consistent
Baseline: 100.0% unique openers · 0.0% repeated pairs
Submission: 100.0% unique openers · 0.0% repeated pairs
Rate of repeated word pairs and varied sentence openings matches the baseline.
Writing style shifts can happen for legitimate reasons — moving from creative to analytical writing, following a teacher's writing frame, or spending extra time redrafting at home. Use this profile to prepare a supportive conversation, never as sole proof of misconduct.
Step-by-step instructions

How to run a fair authorship check & read the 6 style signals

Comparing a learner against their own baseline avoids the biases of generic one-number AI detectors.

1

Choose a like-for-like baseline

Pick an in-class essay, timed assessment, or supervised draft in a similar genre (analytical with analytical). At least 120 words per box gives the most reliable MATTR and burstiness readings.

2

Look for multi-signal divergence

A single shift (e.g. fewer contractions in a formal report) is normal. Pay attention when 3+ dimensions shift at once — such as vocabulary density jumping while sentence rhythm flattens and em-dashes appear.

3

Ask process & vocabulary questions

Bring the specific dimension cards to a calm check-in: ask the learner to define two high-register words from their submission or talk through how they structured a key paragraph.

Want multi-sample learner baselines and whole-class ZIP checks?

In the full Feedback Flows Writing Fingerprints tool, each student profile stores a portfolio of verified writing samples over the year, combines stylometrics with deep linguistic comparison, and checks an entire class ZIP in one batch.

Explore Writing Fingerprints →
FAQ

Authorship checker FAQ

It compares a known authentic piece of student writing against a new submission across 6 stylometric dimensions: sentence length and rhythm (burstiness), 50-word vocabulary diversity (MATTR), register and contractions, sentence openers and connectives, typographic punctuation, and phrase repetition.

Generic AI detectors guess whether prose looks statistical "AI-like" in isolation, which can misfire on polished or EAL writing. An authorship checker asks a fairer classroom question: does this new piece match how this specific learner usually writes?

At least 40 words per box is required, and 120–400 words per sample gives much steadier vocabulary (MATTR) and sentence-rhythm measurements.

No. Legitimate shifts happen when a learner moves from creative to analytical writing, follows a structured writing frame, or spends extra time redrafting. Use the dimension cards to guide a supportive conversation about how they wrote the piece.

No. This tool runs deterministically in Python with zero AI calls, and pasted text is discarded after the short job TTL (~1 hour).

Inside Feedback Flows, Writing Fingerprints stores a multi-sample baseline portfolio on each student profile, combines stylometrics with contextual analysis, and lets you run whole-class ZIP comparisons on deadline day.
Get started

Build persistent writing fingerprints for your classes

Store authentic writing baselines on every learner profile and combine authorship checks with evidence-led AI detection.

Full Writing Fingerprints Tool Guide: Authorship checks in UK schools