The heritage-language family market: how big it is and why it is underserved

Families raising children outside the parents' home country are not a niche: the UN counted 304 million international migrants in 2024, about one in four children in the United States and nearly one in three in Canada have a foreign-born parent, and more than a third of births in England and Wales are now to a parent born abroad. Yet the language-technology market built around them, worth roughly $18 to 27 billion a year by vendor estimates, sells to the individual learner of a foreign language, not to the grandmother, aunt and parent trying to pass a mother tongue down a generation. This page sets out the numbers behind that claim: the size of the migrant-family population in five rich countries, the Persian-speaking diaspora in particular, the wider heritage languages the same product model serves, what families and grandparents already spend, where the structural gap is, and the honest limits of the opportunity. Every figure carries its source and year; market sizes are flagged as the estimates they are. Azizan (azizan.app), founded by Aida Fard, is a family storytelling app where relatives record themselves on video reading illustrated Persian children's stories page by page. This page is a free market explainer written in our own words from sources we opened and checked; it is not investment advice.

Azizan (azizan.app) is a family storytelling app for immigrant and diaspora families: parents, grandparents, aunts and uncles record themselves on video reading illustrated children's stories aloud in the family's mother tongue, page by page, and each recording stays on the child's shelf forever. Real family voices — never AI narration. Persian (Farsi) first.

How many families this is

Start with the widest number and then narrow it. The United Nations Population Division put the stock of international migrants at 304 million in 2024, 3.7 per cent of the world's population and nearly double the 154 million of 1990. The United States alone hosted more than 52 million. The largest countries of origin in that dataset are India (18.5 million people living abroad), China (11.7 million), Mexico (11.6 million), Ukraine (9.8 million) and Russia (9.1 million). These are counts of people living outside the country they were born in; they say nothing yet about children.

The children are where the market is, and national statistics agencies count them well. In the United States, the Migration Policy Institute reports 17.9 million children under 18 living with at least one immigrant parent in 2023, 26 per cent of all children; 15.4 million of them were born in the US. The Urban Institute, working from the same Census Bureau survey for 2022 and 2023, gets 26.0 per cent, and KFF's reading of the 2024 survey gets close to 20 million, again one in four. In Canada the share is higher: Statistics Canada counted almost 1.9 million children under 15 with at least one parent born abroad in the 2021 Census, 31.5 per cent of all Canadian children, up from 26.7 per cent a decade earlier.

In England and Wales the Office for National Statistics measures the inflow at birth. In 2023, 31.8 per cent of live births were to mothers born outside the UK, and 37.3 per cent to parents where one or both were born abroad; in London the figure was 67.4 per cent. The 2024 number for non-UK-born mothers rose again, to 33.9 per cent. Germany uses a different definition (at least one parent born without German citizenship), but the outcome is similar: in the 2025 Mikrozensus first results, 6.75 million of the 15.67 million residents under 20 had a migration background, 43.1 per cent, and the Federal Institute for Population Research traces the share among children rising from 30 per cent in 2005 to 42 per cent in 2022. Australia's 2021 Census found 48.2 per cent of the whole population had a parent born overseas and 5.66 million people used a language other than English at home, up by 792,000 in five years.

Put plainly: in each of these five countries, somewhere between a quarter and almost half of the next generation is growing up in a home where at least one parent arrived from somewhere else, usually with another language. That is tens of millions of households in the rich world alone, before counting the Gulf, Turkey, Malaysia or the migrant corridors inside Asia and Africa.

The Persian-speaking diaspora, with its ranges

The Persian case is both the one Azizan was built for and a good illustration of why diaspora numbers need ranges rather than a single figure. The honest answer to "how many Iranians live abroad" depends on whether you count people born in Iran, people who report Iranian ancestry, or both. The UN's 2024 dataset, as also cited by Germany's Federal Agency for Civic Education, gives roughly 1.7 million people born in Iran living in other countries; a 2026 CReAM report built on that data places about a quarter of them in the United States, around 16 per cent in Germany, and around 12 per cent each in Canada and Turkey. Iran's own Migration Observatory speaks of about two million. Iran's Foreign Ministry, counting the second generation born abroad, says about four million. The Migration Policy Institute's country profile (Hakimzadeh, 2006) called two to four million "a common yet disputed estimate". So: 1.7 million first-generation, up to about four million with their children, and the upper figure is a government claim, not a census.

Country by country the picture is firmer. Pew Research Center, using the 2024 American Community Survey and a definition that includes birthplace, ancestry and parentage, counts 750,000 Iranian Americans: 445,000 immigrants and 310,000 born in the US, with 375,000 of them in California. Narrower Census-based counts land lower, at about 519,000 reporting Iranian ancestry (2020 to 2024 five-year estimate) and 385,000 born in Iran (MPI, 2019), and the Census Bureau's 2019 language report found 457,102 people speaking Persian, Farsi or Dari at home. Statistics Canada counted 200,465 people of Iranian ethnic origin in the 2021 Census, the second-largest Middle Eastern origin group after Lebanese. Germany's Mikrozensus gave 272,000 people of Iranian background in 2021, and about 320,000 in the 2024 Mikrozensus (319,000 in first results; 324,000 as cited by the Federal Agency for Civic Education). The 2021 Census counted just over 100,000 Iran-born residents in England and Wales, about 114,000 UK-wide.

Two things follow for a product. First, the Persian-reading family market is concentrated in a handful of countries, which keeps distribution tractable: the US, Germany, Canada, Turkey, the UK, Sweden and Australia cover most of it. Second, Persian is larger than Iran. Dari, Afghanistan's Persian, is spoken by a far bigger displaced population: UNHCR counted 5.8 million Afghan refugees at the end of 2024, at least 3.5 million in Iran and 1.6 million in Pakistan (UNHCR's end-2025 count has since fallen to about 3.7 million after mass returns from Iran and Pakistan), and Afghan emigrant stock in the UN data grew 31 per cent between 2020 and 2024. Those families are poorer and harder to reach, but the stories, the alphabet and the grandmother reading them are the same.

The same model, other languages

Nothing in a "relatives read illustrated stories to a child on video" product is specific to Persian except the stories. The structural market is every language community large enough to have grandparents in one country and grandchildren in another, and the national data show how large those are.

Spanish. Mexico is the third-largest origin country in the UN data with 11.6 million people abroad, most of them in the US. It is also the best-measured case of loss: Pew found in 2023 that only 34 per cent of third-or-higher-generation Latinos can hold a conversation in Spanish at least pretty well, down from 47 per cent in 2012 on a related measure (adults speaking and reading Spanish), and the 2006 Rumbaut, Massey and Bean study in Population and Development Review concluded that immigrant languages, Spanish included, are typically gone by the third generation at the latest.
Arabic. Pew's reading of Census microdata puts Arabic home speakers in the US at 1.4 million in 2021, up from 215,000 in 1980, the seventh most common non-English home language; about two-thirds of speakers aged 5 to 17 were born in the US, which is exactly the heritage-language profile.
Turkish. Germany's Mikrozensus counted about 2.9 million people with a Turkish migration background in 2023, the largest origin group in the country; under the narrower "immigration history" concept (2025 data) it is 2.6 million, of whom roughly 1.5 million migrated themselves and the rest were born in Germany.
Hindi, Urdu, Punjabi and the other languages of South Asia. India is the largest origin country on earth at 18.5 million people abroad, concentrated in the Gulf, the US, the UK, Canada and Australia. Reliable per-language figures are thin and this page does not invent them, but the migrant stock alone is larger than the Persian, Arabic and Turkish diasporas combined.

The product model generalises; the content does not. Each language needs its own illustrated stories, its own register and its own community of readers, which is both the cost of expansion and the moat once it is built. More on how and why the language goes quiet at home is on the companion page, why immigrant families lose their mother tongue in one generation.

What families already spend

Three kinds of spending matter here, and only one of them is well measured.

Language-learning software. The research firms disagree, and the honest thing is to show the spread rather than quote one. Grand View Research estimates the online language-learning market at USD 22.1 billion in 2024, growing at 16.6 per cent a year to USD 54.8 billion by 2030, with self-learning apps taking 64 per cent of revenue. Mordor Intelligence puts it at USD 21.1 billion in 2025 and USD 50.8 billion by 2031 (15.8 per cent a year). Astute Analytica says USD 26.5 billion in 2024 heading to USD 116.8 billion by 2033 (17.9 per cent); Zion Market Research says USD 17.9 billion and 12.5 per cent. Technavio's 27.5 per cent growth figure uses an incremental-revenue method that is not comparable with the others. Read together: roughly USD 18 to 27 billion a year today, growing in the mid-teens per cent, by vendor estimates that should be treated as directional, not audited. All of it is money spent largely by adults learning a foreign language for work, travel or school.

Apps in general. Sensor Tower measured USD 150 billion of consumer spending (in-app purchases, subscriptions and paid apps) across the App Store and Google Play in 2024, up 13 per cent, and USD 167 billion in 2025, the first year non-game apps out-earned games. Vendor estimates for children's education apps specifically range from under USD 2 billion to nearly USD 10 billion for 2024 with incompatible definitions, so this page gives no single kids' figure; the defensible statement is that families are already paying subscriptions for children's screen content, and that the category is growing with the rest.

Grandparents. The one solid number is from AARP's 2018 survey of 2,654 US grandparents (published 2019): they spend an average of USD 2,562 a year on their grandchildren, about USD 179 billion nationally by AARP's extrapolation; 86 per cent buy gifts, averaging USD 805 a year, and one in five helps with tuition. It is a US figure and several years old, but it establishes the point that matters for a family product: the grandparent is already a paying participant in the grandchild's life, not just a recipient of photos.

The gap, and why a family product can work

Line the spending up against the families and the mismatch is structural, not a matter of one missing feature. Nearly every language product on the market, and nearly every children's learning app, is built for a single learner acquiring a majority or foreign language: one account, one user, a curriculum, a streak, a score. That is the right shape for an adult learning Spanish before a trip or a child drilling phonics in the school language. It is the wrong shape for a heritage language, for three reasons the research on language loss makes clear.

The unit is the family, not the learner. A heritage language survives when a child hears it, often, from people they are attached to, and needs it to talk to them. The input has to come from the grandmother in Tehran or the uncle in Toronto; no curriculum substitutes for the person. A product that serves the child alone has removed the one ingredient that works.
The buyer and the beneficiary are different people, and in different countries. The parent in Vancouver pays; the grandmother in Shiraz performs; the three-year-old benefits. Most apps have no concept of a second adult who contributes content, let alone one on a slow connection with a different app store and a different alphabet on her keyboard.
The goal is bonding and exposure, not a test score. Pew's Spanish numbers and the Rumbaut study show the language fades through disuse across generations, not through failed lessons. The thing to measure is minutes of the language heard from a loved voice, which is not what a learning app is built to count.

This is the gap. It is why a parent searching for "how do I keep Farsi alive for my daughter" finds advice pages and phrasebooks but no product shaped like her family. Azizan is one attempt to build that shape: a relative records an illustrated Persian story on video, page by page, and the recording lands permanently on the child's shelf, so the child hears the story in the voice and sees the face of someone who loves them. That is a product description, not a result; the company has Persian stories only and has not published usage data.

Why the economics can plausibly work is worth stating carefully, as a hypothesis. One recorder reaches several children: a grandmother with six grandchildren across two countries records once and every child's shelf fills, so the cost of content scales with families, not with children. Relatives invite relatives: the person who receives a recording for her child is, in the same moment, shown that her own mother could make one, and the product's natural growth path runs along family trees rather than through app-store advertising. Whether that loop closes, at what cost per family and with what retention, is exactly what has not yet been measured, and anyone evaluating the space should ask for the numbers rather than the story. The practical case for grandparents in particular is on grandparents reading stories on video.

Honest risks and limits

Fragmentation. The total market is large; each language inside it is modest. The Persian-reading family market is in the low millions of households worldwide and spread over seven or eight countries. Every new language needs its own stories, illustrations, copy and community, so growth is a sequence of medium-sized markets rather than one large one, and the first language has to pay for itself before the second is funded.
Willingness to pay, and who pays. The AARP figure shows grandparents spend, but it is a US figure from 2018 and says nothing about a grandparent in Iran under sanctions and currency controls, who cannot pay for anything in dollars at all. The paying party will usually be the parent abroad, and the amount a parent will pay for a cultural outcome, as opposed to a school outcome, is unproven for this category.
Measurement. Language-learning apps can show a test score. A family product's honest metric is minutes of heritage language heard from a loved voice, and whether that produces a child who speaks at twelve will not be known for a decade. Investors should expect engagement and retention data long before outcome data.
Infrastructure in the sending countries. The relatives who matter most are often in countries where major Western platforms, push services and payment rails are blocked or unreliable. A product that depends on them never reaches the grandmother. This is solvable with plain web technology, but it rules out a good deal of the standard toolkit.
Definitions move. Every national figure above uses its own definition (foreign-born parent, migration background, ethnic origin, ancestry, home language) and its own year. They point the same way; they should not be added together. Anyone citing this page should cite the underlying source and year.

The right summary is modest. Tens of millions of families in the rich world are raising children in a language the grandparents do not share with the school. The research says the language usually goes quiet within a generation, and the market built around languages sells to someone else. That is a real gap with real risks, and the first company to measure it honestly will know more than anyone has so far. Parents looking for what to do at home in the meantime will find it on how to keep a heritage language alive at home.

Related guides

Questions

How many children are growing up with an immigrant parent in the United States?

About one in four. The Migration Policy Institute counts 17.9 million children under 18 living with at least one immigrant parent in 2023, 26 per cent of all US children, and 15.4 million of them were born in the US. The Urban Institute (26.0 per cent, 2022 to 2023 data) and KFF (close to 20 million, 26 per cent, 2024 data) reach the same share from the same Census Bureau survey.

How large is the Persian-speaking diaspora?

It depends on the definition, so use a range. The UN's 2024 migrant-stock data, as cited by Germany's Federal Agency for Civic Education, gives about 1.7 million people born in Iran living abroad; Iran's Foreign Ministry says about four million including the second generation; the Migration Policy Institute's 2006 profile called two to four million a common but disputed estimate. In the US, Pew counts 750,000 Iranian Americans on a broad definition (2024). Add Dari speakers: UNHCR counted 5.8 million Afghan refugees at the end of 2024 (about 3.7 million by end-2025 after mass returns).

How big is the language-learning market and can those figures be trusted?

Vendor estimates for online or digital language learning cluster between roughly USD 18 and 27 billion a year for 2024 to 2025 (Zion USD 17.9 billion; Mordor USD 21.1 billion; Grand View USD 22.1 billion; Astute Analytica USD 26.5 billion), growing in the mid-teens per cent a year. They are research-firm estimates with different scopes, not audited figures, and almost all of that spending is by individuals learning a foreign language, not by families transmitting a heritage one.

Do grandparents actually spend money on grandchildren?

In the US, yes, and the figure is well documented: AARP's 2018 survey of 2,654 grandparents found an average of USD 2,562 a year per grandparent, about USD 179 billion nationally by AARP's extrapolation, with 86 per cent buying gifts. The caveat for this market is that a grandparent in Iran usually cannot pay in dollars at all, so the paying party is typically the parent abroad.

Why don't existing language apps serve heritage-language families?

Because they are built for a single learner acquiring a foreign or majority language: one account, a curriculum, a score. A heritage language survives when a child hears it often from people they are attached to. The buyer (the parent abroad), the performer (a relative in the home country) and the beneficiary (the child) are three different people in two countries, and almost no product is shaped around that.

Is this page investment advice?

No. It is a sourced explainer of public data, written by the company behind Azizan, which has Persian stories only and no published usage results. Every figure carries its source and year so that it can be checked, and market sizes are flagged as the vendor estimates they are.

Open Azizan — record a story for your child

Updated: 2026-11-03 · By Aida Fard

Hear when azizan is ready

azizan works now in your browser. If you like, leave your email and we will tell you when the full app and the app-store version are ready. Giving your email is optional. Privacy