The image most parents carry of a child learning a second language is a kitchen table with flashcards on it, a kid being asked the French word for "spoon," a kid wriggling. Or it is an app on a tablet with green owls and confetti, the child tapping the right box and earning a streak. The version that actually works is closer to a different picture. It is a six-year-old asking, again, to read the book about the puppet who keeps lying, in a language the child cannot quite read yet, because she wants to find out whether he gets eaten by the shark this time. The second child will be bilingual in five years. The flashcard child probably will not.
The interesting thing is how settled the evidence on this is. Forty years of second-language research keep landing in roughly the same place, and the place is straightforward enough to summarise in a paragraph. Children acquire languages through stories they enjoy, in conditions where the meaning is just slightly ahead of what they can already understand, and they need a lot of those stories. Almost everything else, drills, flashcards, lessons, app streaks, is at best a small supplement and at worst an active distraction. That paragraph is the entire field.
What the studies actually show
The central name here is Stephen Krashen, who in a series of books across the 1980s and 1990s laid out what he called the input hypothesis. The gist is that people acquire languages by understanding messages, not by studying rules. The version of this finding that matters for children is what Krashen called Free Voluntary Reading. Kids who are given access to books they want to read in their target language, and time to read them with no quizzes attached, consistently outperform kids in traditional instruction on every measure anyone has thought to test, from vocabulary to grammar to writing to standardised exams. The effect sizes in some studies look almost embarrassing.
The most famous of these is the Book Flood research run by Warwick Elley in Singapore and Fiji in the late 1980s. Elley took several hundred classrooms of children learning English as a second language and replaced their structured drill curriculum with a simple intervention. He put a few hundred high-interest storybooks in each room and gave the children time to read. Reading scores in the book-flood classrooms moved about twice as fast as in the control rooms, with comparable gains in writing. The intervention did nothing fancy. It just put real books in front of children who wanted to read them.
A separate finding from the reading literature, the Matthew effect named by Keith Stanovich in 1986, explains why this compounds so brutally over a childhood. Children who read slightly more than their peers in any given year acquire slightly more vocabulary that year, which makes the next book slightly easier, which makes them read slightly more, and so on. Over five or six years the gap between a child who reads daily and a child who reads almost never widens to a chasm that schools find very difficult to close. The OECD's PISA reports keep finding the same thing on a global scale. Time spent reading for pleasure is one of the strongest single predictors of academic outcome, larger than family income once you control for everything else.
Why apps mostly do the wrong thing
None of this is what the standard children's language app delivers. The category is built around quick wins, taps, streaks, levels, and short timed tasks, because those are what produce visible engagement metrics and parent-friendly progress screens. They are also almost the opposite of what the research recommends. A kid who has played a green owl app every day for two years has often learnt to win that app and still cannot read a paragraph of the language without help. The mechanics generate the impression of progress without the substance. None of this is the app's fault, exactly. It is what you get when you design a children's product around what parents will pay for monthly rather than around the one thing the field agrees children need.
The deeper problem is that the drilled approach also kills the second prize that matters. A child who comes out of years of language drills usually has no particular love for the language and no instinct to read in it. A child who comes out of years of reading stories in another language usually loves both the language and reading itself, and will keep doing both for life without anyone making them. The research describes this as two separate outcomes. Parents tend to assume the love part is a bonus. It is the entire compounding mechanism.
Why children's books are unusually good at this job
There is a quiet reason children's classics keep showing up in language teaching that has nothing to do with sentiment. The good ones were optimised over the last two centuries to be addictive to a six or eight or ten year old, in a way that very little modern educational content is. Pinocchio gets nailed by a fox in chapter one and turned into a donkey by chapter twenty-five. Hansel and Gretel are abandoned by their parents on page two. The Jungle Book opens with a child wolf pack. Heidi is sent up a mountain to live with a hostile grandfather and a goat called Little Swan. None of these stories needed pedagogical wrapping. They were already designed to make a child unable to stop turning the page, which is, almost coincidentally, exactly the input condition Krashen spent his career arguing for.
Adapt one of those books down to the actual level a learning child can read, leave the plot and the menace and the strange specific details intact, and you have a piece of material that does the language-acquisition job and the love-of-reading job in the same fifteen minutes a day. You do not have to choose between them. You almost cannot do one of them without doing the other.
What it looks like done right
Concretely, the version that matches the evidence looks like this. The child has access to real books in the target language, at a level they can actually read with mild effort. The books are stories they want to finish, not curricular extracts. Help is available when the child asks for it, in the form of a small marginal translation or a tappable word, not a pop quiz at the end of the page. The child reads a short stretch every day, ten or fifteen minutes is plenty at this age, ideally at roughly the same time. No one tracks streaks. No one quizzes. When the child finishes a book, the next one is on the shelf. The parent's job, evidence-backed, is to keep the shelf stocked and otherwise stay out of the way. That is the whole method.
This is not a hard product brief. It is a hard parental discipline. The parent has to resist the urge to test, to correct, to ask the child to say something in French at dinner, to be impressed in front of relatives, and to celebrate the streak instead of the finished book. Those urges are the ones that quietly poison the acquisition curve we just spent six paragraphs describing. The instrument is patience.
The longer shelf
Storica's A0 to A1 shelf is built almost entirely out of the children's classics that the literature keeps recommending. Pinocchio in Italian, Heidi in German, The Jungle Book in English, Beauty and the Beast in French, Grimm's Fairy Tales in German, The Wizard of Oz in English. All adapted to a level a child who has been learning the language for one to two years can actually read, with the strangenesses left in, the puppet still hanged in chapter fifteen, the witch still melting at the end. A child who reads through that shelf comes out the other side with two habits running side by side, which is what the research has been saying for forty years is the only thing that really works.
I build Storica, a daily reading club for the language you're learning. The children's shelf in particular is built on the idea above: real books, the strangenesses left in, fifteen minutes a day, no streaks, no nags, just the next page.