Exposure share predicts vocabulary
Two-year-olds’ vocabulary in each language tracks the share of input in that language (Place and Hoff, 2011).
For families keeping a language alive
A language survives on hours, not lessons
If your child’s heritage language has no school and no street, it lives or dies on how many hours of real talk it gets each week, from how many speakers. Yearkeep makes those hours visible and asks the adults to change the week, not to buy an app for the child.
The budget
Set a target share for each language, by age
Pick the languages your child will hear. Set a percentage for each by age band. The default over-weights the language with no environmental support, because that is the one that fails first.
Assign each adult a language. Video calls with grandparents count as speakers, and speakers matter as much as hours.
The forecast
See the gap before the week starts
Every planned block carries a language. The week view shows the planned split against the target, and turns amber when a language is short. You fix it on Sunday, not at the end of the month.
The evidence
Quotes keep their language. Words become a vocabulary.
Log “she said” in any script. The language is detected or picked. New words go to an inbox you approve, merge, or reject, and the Language page shows first words, counts per language, and the share each language actually got.
When a language drops under 15 percent for weeks, or vocabulary stalls, a calm card says so. It never diagnoses.
The speakers
Grandparents are part of the plan
Relatives on Highlights get a weekly link in their language, with call prompts: three questions to ask the child. They can record a story, which appears in the books list for the child to hear again. Every speaker you add is hours the language would not otherwise get.
What the research says
Three facts the budget is built on
Speakers matter independently
Hearing a language from several native speakers helps beyond the raw hours.
Below roughly a fifth, it fades
Languages that fall well under 20 percent of waking input rarely become productive. The default budget keeps every planned language above 15 percent.
How a heritage-language family starts
- 01
Name the languages and the speakers
Including the grandparents on video calls.
- 02
Accept the default budget for the age
Then move the language with no street up by ten points.
- 03
Run one week and read the pie on Sunday
Change one block. That is the whole method.
FAQ
Is my language supported?
The language plan today covers Dhivehi, English, Malay, Indonesian, and Mandarin; more plan languages are on the roadmap. The interface itself is in 19 languages, including Dhivehi, Arabic, Hindi, Tamil, Bengali, Urdu, Chinese, Spanish, Portuguese, and Tagalog.
Does it teach my child the language?
No. Adults do. Yearkeep measures and plans the hours and gives every adult the same prompts. There are no child lessons and no screen time.
How accurate is the hours count?
It is an estimate from planned blocks, logged blocks, and quotes. The dashboard says so and shows weekly trends rather than daily precision.
Can I write in a non-Latin script?
Yes. Quotes accept any script, including right-to-left and Thaana. Mixed-direction text is handled.
Give the language its hours
Set the budget, assign the speakers, and read the pie on Sunday.