Mot-à-Mot

Product

Mot-à-Mot is a French practice companion that helps learners actively use the French they already know.

Write a sentence, see how a native speaker might say it in everyday conversation and writing, build a personal French Toolbox from what you encounter, and practice using that growing vocabulary.

Live app: https://mot-a-mot-nine.vercel.app

Problem

The gap isn't knowing the rule. It's getting enough practice using it.

Most language tools optimize for correction, translation, or structured lessons. They don't necessarily create a loop around the vocabulary a learner is personally accumulating.

A learner can check a sentence, receive a correction, and move on without ever reusing what they just learned.

Who is the user?

Adult French learners who want to actively use French in everyday life and build confidence through repeated practice.

Particularly learners who already have some vocabulary and grammar, but need more opportunities to produce, reuse, and internalize it.

Why this problem?

I encountered this gap while learning French in Paris.

I could study vocabulary and grammar, but the harder problem was finding a way to repeatedly use what I already knew, understand how it would actually be expressed by a French speaker, and turn those encounters into future practice.

But I didn't want to assume my experience was universal.

What research told me

Don't just show learners the French. Give them a reason to produce, retrieve, and reuse it. This became the foundation for Mot-à-Mot.

• Produce — Language production can help learners notice gaps in what they know. (Izumi et al., 1999)
• Retrieve — Retrieving vocabulary from memory can outperform simply repeating it. (Kang et al., 2013)
• Revisit — Spacing and repeated retrieval can improve longer-term L2 vocabulary retention. (Nakata, 2015; Nakata et al., 2023)
• Structure AI — AI feedback can support engagement, but learners also report limitations and concerns. (Teng, 2024)

Opportunity

Turn every interaction into part of a personal learning loop: Write → Understand → Add to Toolbox → Practice → Reuse

The longer-term opportunity is to extend this loop across writing, reading, listening, and speaking.

Product hypothesis

If learners repeatedly practice with vocabulary drawn from their own French, they will become more confident and capable of producing the language independently.

This is a hypothesis, not a research finding.

The research supports the mechanisms behind the product: production, retrieval, feedback, and repeated practice. Mot-à-Mot tests whether connecting those mechanisms to a learner's own accumulated French makes practice more relevant and actionable.

Success criteria

North star: Are we helping learners actively use more of the French they know?

• Learners build a growing personal French Toolbox
• Learners complete practice sessions using accumulated vocabulary
• Learners reuse previously learned vocabulary in new contexts
• Learners become more confident producing French independently

Research

Product direction evolved through iterative PRDs and hands-on testing from V1.0 → V1.3.

The product began as a sentence-correction tool. Testing and reflection challenged that premise: Correction tells me what I did wrong. Practice helps me use what I know.

V1.2 began shifting toward a personal vocabulary loop. V1.3 introduced the Practice Lab and connected Check → Toolbox → Practice → History.

Validation so far has primarily been through self-testing and iterative product development. Broader user testing is the next validation step.

Product strategy

From correction tool → practice companion

Start with writing. Build the learner's vocabulary. Turn that vocabulary into practice.

Rather than prescribing a curriculum or artificial CEFR progression, Mot-à-Mot focuses on the learner's existing language inventory.

Check gives feedback → Toolbox captures what the learner has accumulated → Practice asks them to retrieve and use it again.

Then eventually: Writing → Reading → Listening → Speaking — with the Toolbox acting as the connective tissue.

PRD / Requirements

• Check: French input → Everyday French (Speaking) + Writing: Foundation, Expanding, Fluent
• Meaning: Show what the learner's sentence actually communicates
• What Changed: Highlight meaningful differences and compare each version with the original
• French Toolbox: Store vocabulary by lemma, surface form, meaning, and grammatical function
• Import: Add existing vocabulary or notes to the Toolbox
• Practice Lab: Generate sessions from the learner's Toolbox
• Practice modes: Spot & Match + Write in French
• History: Save and revisit previous checks
• Pronunciation: Results and Toolbox entries
• Feedback: Built-in feedback form connected to email

UX decisions

• Removed CEFR labels from the core experience
• Everyday French first: prioritize natural conversational language
• Writing progression: Foundation → Expanding → Fluent
• Meaning over complexity: never make French more complicated just to appear advanced
• Meaningful changes only: unchanged text is hidden
• Practice without hints: test retrieval rather than recognition
• 10-question sessions: short, focused practice with immediate feedback
• Mobile-first: low-friction interactions throughout

Build

A mobile-first AI web application deployed on Vercel.

Built iteratively from PRD → prototype → deployment → testing → refinement, with AI-powered sentence analysis, vocabulary extraction, local Toolbox storage, practice generation, pronunciation, history, and an integrated feedback system.

Iteration

Product evolution: Correction → Collection → Retrieval → Reuse

• V1.0 — Check my French: Sentence correction + explanations
• V1.1 — Remember what I'm learning: Meaning clarification + French Toolbox + History
• V1.2 — Make the correction more useful: Meaning preservation + writing variation
• V1.3 — Practice what I've accumulated: Practice Lab + Toolbox + 4-tab navigation + import + pronunciation

What I learned

• More complexity isn't automatically more value. If one expression works naturally across contexts, don't manufacture alternatives.
• Meaning preservation matters more than sophistication.
• The learner's own vocabulary is a powerful practice dataset.
• Correction becomes more valuable when it feeds future practice.
• AI needs structure. Better prompts, schemas, validation, and deterministic rules materially improved the experience.
• Building changed the product strategy. The product I started with was fundamentally different from the practice companion I ended up building.

What's next

Validate the loop. Does Check → Toolbox → Practice actually make learners return, retrieve, and reuse more of their accumulated French?

If the core loop proves valuable, expand the same philosophy beyond writing: Writing → Reading → Listening → Speaking — with the French Toolbox connecting all four.