What we can actually show about how people learn
Language-learning marketing is full of numbers with no method behind them. This is where we set out what we study, how we study it, and what learner data is involved.
How we work
Four commitments that decide what we are willing to claim in public.
Publish the misses too
A study that shows a feature made no difference gets written up the same as one that shows it worked. Selective publication is how an industry convinces itself of things that are not true.
Consent is opt-in and revocable
No learner is enrolled in a study by default. You opt in, you can see what was collected, and withdrawing removes your data from future analysis.
Pre-register the question
We write down what we expect to see before we look, so a null result cannot quietly become a different, more flattering finding after the fact.
Methods in public
Sample size, cohort definition and the statistical approach are published with every result, so the finding can be argued with rather than merely believed.
What we study
Three questions occupy most of the work. How much speaking time in a lesson actually predicts fluency gain, versus time spent on grammar explanation. Whether spaced repetition built from a learner's own errors outperforms a shared deck at the same level. And how closely our automated writing and speaking estimates track a human examiner on the same script.
That third question is the one we are least comfortable with, and the reason those two scores are presented as a range rather than a single number on your dashboard.
Published work
Nothing published yet
We would rather say this plainly than pad the page. When the first study is complete it will appear here with its full method, its sample size and its raw effect size — including if the answer is that the thing we built did not help.
Your data
Research analysis runs on aggregated, de-identified records. Names, email addresses and payment details are never part of a study dataset, and no individual learner is identifiable in a published result.
If you have opted in and change your mind, withdrawing takes effect immediately for anything not already published. You can check or change that setting from your account at any time.
Researching language acquisition?
We are open to collaborations with universities and independent researchers, including access to de-identified datasets under an agreement.