Research collaboration

YAKKI is research-informed by design — and we want it research-validated. This module describes the platform as a research instrument, the studies we are seeking partners for, and our position on data ethics.

The instrument

What the platform offers a researcher

Controlled task generation

Exercises are generated to specification — topic, CEFR level, format, plus the L1 of the student interface. For a researcher this means stimulus control that a fixed textbook cannot offer: the same construct, instantiated fresh, at scale.

Event-level process data

The platform sees practice as it happens: answers, retries, hint usage, response patterns across game formats. Process data of this granularity is usually expensive to instrument; here study-specific capture can be built into the platform’s native operating mode.

An L1-diverse population

Israeli classrooms bring Hebrew, Arabic, Russian, and Amharic first languages into the same schools — a natural setting for crosslinguistic transfer research that most platforms cannot reach.

A real intervention surface

Conditions can differ by session: game formats, feedback styles, L1 interface on or off. Comparative designs can run inside ordinary lessons, not in a lab simulacrum.

What we are seeking

Studies we want to run with partners

Efficacy pilots

Does structured YAKKI practice improve vocabulary retention, reading, or willingness to speak — against business-as-usual instruction? Pre/post designs in pilot schools.

Dyslexia accessibility validation

Our adaptations follow published guidance; whether they help actual students with dyslexia is exactly the kind of claim that needs independent validation. This is our most urgent ask.

L1 transfer in practice

Per-L1 adaptation of errors and support is the direction the platform is being built toward. Do per-L1 paths outperform generic ones? Israeli multilingual classrooms are the right place to ask.

Formative assessment validity

Hint usage and retry patterns will feed the learner profile now in development. How well does this mediated-assessment signal correlate with established measures?

Terms

What we offer and what we ask

We offer: platform access for study designs, engineering support for experimental conditions, and direct access to the team that builds the system
We ask: independence — findings are the researcher’s, including negative ones. We would rather learn the truth than publish flattery
Publication: we place no restrictions on publishing results; we ask only standard factual review of platform descriptions

Data ethics

Our position, before you ask

Guest mode by default

Classroom sessions run without student accounts — there is no covert profile accumulation to worry about.

Consent is the researcher’s frame

Any study-specific data collection follows the partner institution’s IRB/ethics approval and informed-consent procedures. We build to your protocol, not around it.

Schools stay in control

Participating schools approve study scope. Data leaves the platform only as specified in the agreed protocol.

The full scientific foundation — nine research pillars and the Israeli research the platform builds on — is documented on the Science page. Read the science page

Start the conversation

Write to us with your field and a sentence about what you would want to study. We answer substantively — and in English, Hebrew, or Russian.

Contact for research