Look. See. Write. 0

📝 Handwriting Assessment for occupational therapists

Upload a photo of a child's handwriting sample and get a consistent, rubric-anchored screen: letter-level legibility (ETCH-style percent-legible), component observations in the Minnesota Handwriting Assessment's categories (formation, sizing, spacing, alignment), reversal notes, and a draft paragraph for your documentation. Same rubric, same model, every time — which makes it most useful for week-over-week progress tracking.

Screening aid, not an evaluation. This tool analyzes one static photo. It cannot see stroke order, grasp, posture, or pressure, and it is not equivalent to a standardized instrument administered by a clinician (ETCH, MHA, DASH, Print Tool). Use it to structure observations and track trends; rely on your standardized tools and clinical judgment for eligibility and diagnostic decisions.
Privacy: the photo is sent to our grading service (which calls Anthropic's Claude API), scored, and immediately discarded — never stored. This is the only feature on LookSeeWrite that sends anything over the network. Please keep names out of frame — crop or fold the page so only the writing shows.

1 · The sample

Add a photo of the writing sample

Best results: flat page, straight-on, even light, writing lines fully in frame, no name visible.

A few sentences is ideal (up to about half a notebook page). We'll ask you to crop anything much longer — LookSeeWrite is free, and focused samples keep assessments fast, consistent, and affordable on our limited AI credits.

2 · Context (grade + task condition change what "typical" looks like)

How the scoring works (the rubric, with sources)
  • Letter legibility (primary): a letter counts as legible if recognizable in isolation with all parts completed. Letter-level scoring is the most reliable unit in ETCH research (interrater ICC .84); numerals are excluded (unreliable).
  • Word legibility (secondary): ETCH-C general criterion — quickly, easily, and correctly read as the intended word, at a glance, out of context.
  • Components: Minnesota Handwriting Assessment categories — formation, sizing, spacing, alignment (baseline convention: within ~1/16″), plus slant for cursive. Formation and spacing are weighted first: they're the weakest subskills in beginning writers (Mathwin et al. 2024, n=408).
  • Developmental calibration: grade level and task condition are part of the input because only ~20% of grade-1 / ~40% of grade-2 children write an easily readable self-generated sentence; copied text is reliably neater.
  • Speed (optional): letters/minute vs HW21/Zaner-Bloser consensus benchmarks (grade 2 ≥25 … grade 8 ≥100) — only for timed copying tasks, grade 2+.
  • What a photo can't tell you: stroke order and direction, grasp, posture, pressure, fatigue. The published automated-assessment literature relies mostly on tablet-captured dynamics and still treats clinician assessment as ground truth — which is why this is a screen.

Full evidence base with citations and verification notes: HANDWRITING-RESEARCH.md in the project repository.