The week between Saturdays: a Mandarin practice plan that survives a real Tuesday

It is Wednesday night at my house. Homework is done. There is enough left in the tank for a bath, a book, or ten minutes of Chinese. Not all three. My eight-year-old went to Saturday school four days ago and will go again in three, and somewhere between those two Saturdays a character she wrote five times in class — 寫, say — has quietly stopped being a character she can produce on demand.

Every heritage parent I know has had some version of this Wednesday. The teachers say just do a little each day. The grandparents say just speak more Chinese at home. Both are correct and both are useless as plans, because a little each day loses every fight it picks with soccer, dinner, and a tired kid.

So here is what we actually do, plus what I’d hand a friend who cornered me at pickup. Five weekday slots — three short sessions, one informal exposure day, one rest — around 35 minutes of focused practice across the week. Built around kids roughly 6 to 10 with Saturday school on the calendar. Adjustments for younger and older kids are at the end.

One thing up front. At my daughter’s Taiwanese Saturday school in Seattle, roughly half the parents do not speak Mandarin themselves — second- and third-generation families, Cantonese- or Taiwanese-speaking households, interlingual marriages, adoptive families. The plan works in both cases. I flag the day-by-day shifts as they come up, and there’s a dedicated section near the end.

The two findings the plan rests on

Cepeda et al. (2006) pooled 317 experiments on study spacing. Same total minutes, distributed across days, produced better retention than the same minutes stacked into a single session. Ten minutes on three weeknights beats thirty minutes on Sunday afternoon, every time.

Roediger and Karpicke (2006) compared rereading material against actively retrieving it. Retrieval won by a wide margin on long-term retention, even though it felt harder to learners in the moment. The session that asks the kid to produce the character — flashcards with the hanzi hidden, a read-aloud where she’s the reader — does more than the session spent looking again at Saturday’s worksheet.

Stack those two findings on a Saturday-school calendar and you get a specific shape: three short weekday sessions built around retrieval, an informal exposure day, a rest day before class. Which day carries which role bends to fit the household. The shape doesn’t.

The default rhythm is the inefficient one

Class once a week, nothing in between, class again six days later. Each Saturday partly re-teaches the previous Saturday, because almost a week without retrieval is enough for a real chunk of new material to fade. The fade is gentler for heritage kids than the textbook forgetting curve — there’s usually some prior oral exposure anchoring meaning — but it is not zero, and it stacks up across a school year.

You don’t need more total minutes. You need better distribution of the minutes you already have, and the weekday touches need to ask the child to pull material back out rather than re-read it.

The week, day by day

What follows is one workable assignment of roles to days. Two days will collide with real life every week — a soccer Tuesday, a school project Wednesday, a Friday dinner out. Swap the days around. Keep the shape.

Saturday — class day. Structured instruction, new material, teacher-led. The longest session of the week. No home practice expected after; the class is the practice. Saturday evening at our house is fries, a movie, and zero Chinese. That is by design.

Sunday — listening and connection. The day to make Mandarin audible at home, in whatever form the household can sustain. For a household with a Chinese-speaking parent or grandparent, this is a meal where as much of the conversation as possible happens in Mandarin, a FaceTime with a relative, a bilingual bedtime book read by the fluent adult. For a household without a Chinese-speaking adult at home, the function is the same — sustained, unhurried Mandarin input from a real person on the other end — but the source shifts to a weekly call with a relative, a thirty-minute tutor session on italki or Preply, or a Mandarin cartoon and story-time channel watched together. Polinsky and Kagan (2007) identify naturalistic input and conversation with fluent speakers as disproportionately valuable for heritage maintenance. When the fluent speaker is not in the house, the work is to find one for thirty minutes a week somewhere else.

Monday — ten minutes, retrieve Saturday. Flashcards on the new characters. A read-aloud of the worksheet that came home, with the kid producing the words rather than me reading along. A listening drill on Saturday’s vocab. Monday is the first slot any real family week has room for; Saturday evening and Sunday are eaten by class fatigue. The job of the session is to drag Saturday back out before it sinks. That’s it.

This is the day Roediger and Karpicke (2006) are cashing in on — retrieval beats rereading by a wide margin for long-term retention, even though it feels harder in the moment. Lean into the harder-feeling version. If the kid says “I don’t remember,” that’s the session working, not failing.

Tuesday — fifteen minutes, characters as structures. This is the slow-down day. Two or three characters traced on a worksheet or in a notebook, carefully. Handwriting forces a kind of attention tapping doesn’t. If pencil work is a fight at our house that night — and some nights it is — we drop to a tracing app with the kid saying each character aloud. Pick the version your kid will actually do. A character app done is worth more than handwriting refused.

Research on Chinese literacy ties handwriting practice to character recognition, not just writing — the motor sequence appears to strengthen the visual representation. That is a real finding worth knowing. It is also not strong enough to be worth blowing up the evening over. If the kid will trace, trace. If she won’t, the app is the version that happens.

Wednesday — ten minutes, read-aloud. A graded reader at the right level, read aloud by the kid. If you read Chinese, sit next to her and catch errors as they happen. If you don’t, record it on your phone and either play it back yourself listening for fluency and confidence, or send the clip to the Saturday teacher for periodic feedback. If a graded reader isn’t on hand that night, Saturday’s worksheet re-read aloud is the fallback.

Pure listening is fine on Sunday or Thursday. Wednesday is too close to the next class to spend on input alone — read-aloud asks for output in connected text, which is what the retrieval research supports. Two days before class, you want to know whether she can still produce the words, not just recognize them.

Thursday — low-effort exposure. A Chinese cartoon at dinner. A song on the drive home. A YouTube story-time channel while the kid colors. Whatever ambient Mandarin the household already produces. No structured session, nothing graded, nothing that asks for output. Thursday is the buffer between three weeknight sessions and Friday’s rest — exposure stays in, demand stays out.

Friday — rest. Saturday’s class is the next day, and end-of-school-week fatigue is real. No structured session, no demand. The retrieval and character and read-aloud sessions earlier in the week have done their work. Arriving Saturday rested matters more than arriving primed.

Around 35 minutes of focused home practice across the week. Plus whatever Saturday school asks for. Plus whatever ambient Chinese the household produces on its own.

What counts as practice

When most parents picture practicing Chinese, they picture flashcards and worksheets. Those count. They are a narrow slice of what holds a heritage language across a week.

Formal practice is the structured stuff — the app session, the worksheet, the flashcard review, the read-aloud. Monday, Tuesday, Wednesday in the plan above are all formal.

Informal exposure is everything else. The cereal box. The menu in Chinatown. The cartoon at dinner. The song that’s been on repeat in the car for a month. The text message to grandma. None of it shows up on a tracker. All of it keeps the language audible, which the heritage-maintenance literature consistently ties to long-term retention. Polinsky (2018) describes how robust early exposure makes a language easier to reactivate later in life; the cereal box is feeding that exposure even on the days the formal session does not happen.

The schedule will slip. Tuesday’s worksheet gets displaced by a school project. Wednesday’s read-aloud collides with soccer. On those days, informal exposure is the backstop. A cartoon at dinner is not as strong as a focused ten minutes, but it is meaningfully better than silence, and it keeps the week from turning into a five-day English blackout between Saturdays.

The families whose weeks hold up across months tend to blur the line. A kid hunting for characters on the cereal box is doing informal retrieval. A parent asking what the teacher covered last Saturday — in English or Mandarin, depending on the household — and getting the kid to explain a Chinese concept out loud is running a retrieval session without calling it that.

Making it survive a real Tuesday

The plan works on paper. The question is whether it works on the Tuesday the kid is cranky and the parent is tired and fifteen minutes feels like fifty.

A few patterns separate the families whose weeks hold from the families who mean well and burn out by week three.

Anchor each session to an existing habit. Right after homework or right before bath beats sometime after dinner. A session with a specific anchor gets done; a session floating loose in the evening gets skipped. The anchor is the commitment.

Keep sessions short enough that starting isn’t a commitment. Ten minutes is deliberate. Fifteen is the ceiling for this age band. The moment a weeknight session feels like it requires clearing the table and announcing study time, it stops happening.

Give the kid real choice inside the session. Which five characters to review. Which graded reader. Which song. Tse (2000) describes a tween-years stage where children who feel Chinese is being done to them develop more negative associations with the language; children who have ownership over how it shows up in their day do not. The texture of the week matters as much as the quantity.

Track consistency, not performance. In the first month, the metric is we did a session four days this week, not we got 90% on the flashcards. Performance is lagging and noisy. Consistency is leading and real. Performance follows when consistency holds.

Build the bend into the plan. A rest day and an informal day are in the template on purpose. A missed weeknight isn’t failure — it’s a Tuesday with a fever or a Wednesday with a school concert. Tuesday’s character work is the most fragile slot at our house: handwriting on a heavy homework night is the easiest one to skip and the easiest one to swap to another evening without losing the week’s shape. Next Saturday still happens. The week is a rhythm, not a streak counter.

If you don’t speak Mandarin yourself

Roughly half the families at our Saturday school are in this situation — second- or third-generation Chinese-American parents who lost the language, households where the parents speak Cantonese, Taiwanese, or Hakka but not Mandarin, interlingual marriages, adoptive families. The weekday plan still works. What changes is which parts you run yourself and which parts you outsource.

What you can do at full strength: run the schedule, anchor the sessions, sit next to the kid during practice, ask about Saturday in English, manage the apps and worksheets, drive the consistency. Family-language-policy research (Curdt-Christiansen, 2009; King & Fogle, 2006) finds that parental management and commitment are separable from parental language modeling, and that the management piece moves outcomes on its own. What you can’t do is provide contingent feedback — the kid mispronounces a tone, you catch it, she adjusts. That function needs a fluent speaker somewhere in the week. Not necessarily at home.

Substitutes that hold up well. Pronunciation and listening input: native audio in flashcard apps, TTS for unfamiliar words, Mandarin cartoons, song playlists, audiobooks of graded readers. Roughly equivalent to a fluent adult modeling words, and in some ways steadier because the audio does not get tired on a Tuesday. Auto-graded retrieval: Skritter and similar character apps, Du Chinese for reading with built-in audio and dictionary, any flashcard app with native audio. These do not need a fluent adult to verify correctness. Monday’s worksheet review: ask the kid to explain in English what each item means, then play the audio — most heritage textbooks have audio companions or QR codes. Meaning in English plus pronunciation via audio covers most of what a fluent parent would do.

Substitutes that are partial. Wednesday’s read-aloud: record on your phone, play back yourself, or send to the Saturday teacher for periodic feedback. The contingent-feedback loop is weaker than a fluent parent listening live, but it is real practice and the recording itself often sharpens the kid’s attention. Pinyin or zhuyin annotation on worksheets lets you verify the right syllable, not the right tone.

The piece you can’t substitute with media. Real conversation with a fluent speaker. The cheapest way to get this is a thirty-minute weekly call with a child-friendly tutor on italki or Preply (often $10–20), a standing video call with a Mandarin-speaking relative even if they live in another city, or a Saturday-school playgroup that meets outside class hours. One of these, once a week, covers the gap. The heritage-language literature is consistent that conversation with fluent speakers is disproportionately tied to long-term retention. It does not say the speaker has to be the parent.

The strongest version of this household runs the weekday plan tightly, treats the apps as the pronunciation tutor, treats Saturday school as the structured-instruction layer, and finds one fluent adult somewhere in the week for thirty minutes of real conversation. Plenty of families at Saturday schools are doing exactly this, and their kids hold the language.

When you don’t have what the plan assumes

The plan references a worksheet from class, a graded reader at the right level, and Chinese books at home. Few families have all three. Substitutes exist for each.

No Saturday school on the calendar. The spacing principle still applies. Pick three or four days for short practice, bracket them with at least one rest day, and choose one day as the longest session — the day that carries more of the new-material weight. Whatever app or textbook the family uses then plays the role Saturday’s class plays here.

No worksheet from class. What the worksheet is really doing is constraining Monday’s review to the freshest material. A photo of the whiteboard at pickup, three characters scribbled on a napkin during class, or the last unit covered in whatever app the family uses does the same job. Review some Chinese doesn’t — the closed, defined set is the active ingredient.

No graded reader. The job Wednesday is doing is connected-text read-aloud at a level the kid can mostly decode. Free options that hit it: a Chinese children’s book from the public library (most metro library systems with heritage communities carry them), a free episode from a learner-targeted YouTube channel like Little Fox Chinese with the kid reading the on-screen text aloud, or Saturday’s worksheet re-read aloud as a fallback. Read-along videos work only if the kid actually reads — the temptation is to slide into passive watching, which does not do Wednesday’s job.

No Chinese books at home. Library Chinese picture books, or YouTube story-time channels watched with the parent rather than by the kid alone. Roughly equal if shared attention is preserved.

The day-by-day pattern — distribution, retrieval, at least one rest — is what drives retention. The specific format each day takes is negotiable.

Adjusting for younger and older kids

The template fits roughly 6 to 10 best. Two common adjustments:

Ages 4 to 6. Halve the session lengths — five minutes is the new ten, ten the new fifteen. Skip formal handwriting; tracing with a finger over a printed character, or matching games with character cards, do the same developmental work without the fine-motor strain. Sunday’s listening and connection time matters more at this age than any of the weekday sessions; protect it first. If no one in the house speaks Mandarin, that day becomes a weekly call with a relative or a tutor session plus shared cartoon-or-storytime time. Same function, different format.

Ages 11 to 13. Autonomy is decisive in this bracket. A schedule a parent imposes fails reliably; a schedule the kid owns has a real chance. Hand the template to the kid, ask which days she wants to do which slot, which day she wants to keep clear. Her version of the week will not be identical to the one above. That is fine. The pattern — several short sessions, retrieval over re-reading, at least one rest day — is the part that matters. The specific assignment of slots to days is the part to negotiate. Tween-years resistance to parent-led practice is well documented in the heritage literature, and giving up some control over the schedule is most of how families avoid the worst version of it.

Two hours, 168 hours

Saturday school is two hours. The week is 168. The math is unforgiving but not complicated: whatever happens in those two hours either compounds or partially resets, and the thing that decides which is what happens on the other 166.

The plan above is not a streak counter. It is a shape — three short retrieval-shaped sessions, one informal day, one rest, built around a class. Some weeks you will hit four out of five. Some weeks you will hit two, because Tuesday had a fever and Wednesday had a concert. Both are fine. The families whose kids hold the language across years are not the ones who hit five out of five. They are the ones whose week has a shape, who didn’t burn out by October, and who treated a missed Tuesday as a Tuesday rather than as evidence the project is failing.

The active ingredient is distribution, not minutes. Thirty-five minutes spread across three weeknights and a Sunday will outperform an hour stacked on Sunday afternoon, every time. You don’t need a better hour. You need a week with a pattern in it.

If you found this useful, you can join the waitlist on the homepage for the next piece in this series. The companion article, Heritage Mandarin between ages 8 and 12, covers the broader research on the window the weekday plan is designed for.

About the author

James Wann is the founder of MandarinBytes and father of an 8-year-old daughter at a Taiwanese-heritage Saturday school in Seattle. Before MandarinBytes, he led personalization data at Disney across Disney+, Hulu, and ESPN, and spent three decades building software that adapts to individual users at consumer scale.

Research cited in this article

Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., & Rohrer, D. (2006). Distributed practice in verbal recall tasks: A review and quantitative synthesis. Psychological Bulletin, 132(3), 354–380.
Curdt-Christiansen, X. L. (2009). Invisible and visible language planning: ideological factors in the family language policy of Chinese immigrant families in Quebec. Language Policy, 8(4), 351–375.
King, K. A., & Fogle, L. W. (2006). Bilingual parenting as good parenting: Parents’ perspectives on family language policy for additive bilingualism. International Journal of Bilingual Education and Bilingualism, 9(6), 695–712.
Polinsky, M. (2018). Heritage Languages and their Speakers. Cambridge University Press.
Polinsky, M., & Kagan, O. (2007). Heritage languages: In the “wild” and in the classroom. Language and Linguistics Compass, 1(5), 368–395.
Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249–255.
Tse, L. (2000). The effects of ethnic identity formation on bilingual maintenance and development: An analysis of Asian American narratives. International Journal of Bilingual Education and Bilingualism, 3(3), 185–200.