Loading

Dutch Apps: A Reverse Review of How They Fail You

I have watched four friends fail to learn Dutch the same way.

They install Duolingo. They build a streak. The streak crosses 100, 200, 400 days. They arrive at the koffieautomaat at their job, a Dutch colleague asks them something casual, and they freeze for six seconds before switching to English. The streak does not save them. They quit Dutch within a year and tell themselves "it just was not the right app."

The right app would not have saved them either. The failure was structural, not in the app pick. And until you understand the four ways Dutch apps fail their learners, no comparison table will help you. This post is the reverse review. I am going to walk through the four failure modes, name the apps that fail each one, and name the small set that survives.

I am Alej. I learned Dutch myself during COVID lockdowns in Belgium, using Assimil out of a Cora supermarket in southern Belgium near the Luxembourg border. I now speak 11 languages and built Mynago. Four friends have navigated inburgering on my watch, two of them under the new B1 rule. The pattern below is what I have actually seen, not what the brochures promise.

Failure modes covered:

Failure 1: streak addiction substitutes for production

This is the failure I have watched live, repeatedly, with my own friends. They build a Duolingo streak and confuse it for Dutch.

Recognition is not production. You can tap your way through 800 Duolingo Dutch sentences and not be able to construct one without a multiple-choice scaffold. Dutch has separable verbs ("ik bel je morgen op"), V2 word order, and a politeness register that surfaces in pronoun choice ("je" vs. "u"). All three of these collapse when you have to produce them out loud, under pressure, in a kitchen with a Dutch colleague waiting for your reply.

Apps that fail this test loudest:

Duolingo. The flagship offender. The Dutch course is one of the more thorough ones on the platform but the gamification structure trains your fingers, not your mouth. Two friends carried 400-plus-day streaks into the inburgering speaking exam and froze. The third dropped Dutch after eighteen months of streaking. None of them could hold a five-minute conversation.

Memrise (current version). The post-2022 rebuild stripped most user-generated decks and the remaining content is shallow. Recognition tests dressed as fluency progress. Skip.

Mondly. Pleasant graphics, no production pressure. Tap-through pacing. I tried it for two weeks in 2023 to verify. The verdict held.

Apps that pass this test:

Pimsleur Dutch. Audio-only by design. Forces production from minute one. You sit in silence after the prompt until you speak. The streak is irrelevant because there is nothing to tap. Caps at A2 but the production muscle it builds is real.

Speakly. Built around speaking-from-day-one. Smaller library than Babbel but the pedagogy is exactly the production-first approach Dutch needs.

Mynago. I built it, so weigh that. The dialogue replay structure removes one role from a finished dialogue and waits for you to fill it in out loud. Recognition tests get scored separately from production tests, so you cannot accidentally rack up a streak by tapping.

iTalki Dutch tutor. Cannot be replaced. A Dutch community tutor at 12 to 18 euros an hour twice a week is the single best money you will spend on Dutch. I will keep repeating this in the post because friends keep ignoring it.

The honest sentence is: if your stack contains zero production-forcing tools, your Dutch will never break through the kitchen-conversation barrier. The streak is theater.

Failure 2: the Flemish blind spot (apps that pretend Belgium does not exist)

Roughly 60 percent of Dutch speakers globally are Dutch. Roughly 40 percent are Flemish. Many language apps act as though that 40 percent does not exist.

This matters more than you would think. Flemish vocabulary diverges from Northern Dutch in everyday words (frigo vs. koelkast, plezant vs. leuk, gij vs. jij in casual speech). The accent register is different enough that an Antwerp speaker and an Amsterdam speaker can clock each other in three seconds. If you live in Belgium and your app teaches you Amsterdam Dutch, you will sound like a tourist in your own kitchen for years.

Apps that fail this test:

Duolingo. Netherlands Dutch only. No Flemish variant, no roadmap. Belgium-track learners spend hundreds of hours learning the wrong vocabulary set and then have to swap half of it out.

Babbel. Netherlands-only. Tops out at B1. Flemish is not on the roadmap. If you live in Antwerp or Ghent, Babbel will get you the structure but leave you with northern lexicon you will need to unlearn.

Rosetta Stone. Netherlands-only as well, and the picture-based method cannot teach separable verbs in either variant.

Apps that pass this test:

Mynago. The variant toggle at onboarding sets Netherlands Dutch or Flemish (Vlaams). Audio, vocabulary, and example sentences shift accordingly. This is the slot the app owns for Belgium-track learners.

VRT MAX. Free Flemish public broadcaster streaming. Real Flemish in news, drama, documentary. Use this as your evening news the moment you start studying.

Pimsleur Dutch. The voice talent leans Netherlands-neutral, which is more flexible than Belgium-flagged content but does not actively confuse Belgium-track learners.

iTalki tutors filtered by location. Pick a tutor in Antwerp or Ghent if you live in Flanders, in the Randstad if you live around Amsterdam or Rotterdam. The variant calibration is the tutor's job.

The Belgian inburgeringstraject runs through a different agency (Agentschap Integratie en Inburgering) and expects A2 Dutch for the standard track. The exam is administered separately. If you are on this track, your app stack must include Flemish-aware tools or you will fail register questions you knew the grammar for.

Failure 3: no exam alignment (the inburgering gap)

Most adult Dutch learners are not casual hobbyists. They are expats in the Netherlands or Belgium under a deadline set by a government agency. Inburgering deadlines in the Netherlands. The Vlaamse inburgeringstraject in Belgium. School enrolment paperwork that requires Dutch above a threshold. None of these accept "I did 200 days on Duolingo" as evidence.

Since 1 January 2022 the Wet inburgering 2021 raised the language target to B1 for most newcomers in the Netherlands, with A2 still available as a fallback. The current exam has six components:

No consumer app teaches to this exam end-to-end. The apps that fail this test are the ones that pretend to be enough.

Apps that fail this test:

Anything that does not produce writing samples. Duolingo, Babbel, Memrise, Mondly. The schrijven section requires formal letter-writing register. None of these trains you in it.

Anything that does not handle KNM content. Civic knowledge, Dutch society history, political structure. Out of scope for the consumer app market entirely.

Anything that does not run mock-exam practice in the same item formats the exam uses. This is the disqualifier for almost every general-purpose app.

Apps and resources that pass this test:

oefenen.nl. Free, browser-based, run by the Dutch public broadcasting and adult-education sector. Dedicated tracks for Inburgeringsexamen and Staatsexamen NT2 Programma I (A2-B1) and II (B1-B2). The exercises mirror the exam item formats exactly. If you only use one tool for exam alignment, this is it.

naar Nederland. Official preparation pack for the Basisexamen inburgering buitenland, the exam taken before the MVV visa. Mandatory if your spouse or partner is still outside the Netherlands.

NT2 School. Paid platform, around 25 to 35 euros a month. Exam-aligned, with KNM and MAP modules built in.

Het Nederlands. Free Dutch government NT2 preparation materials. UI is pure 2008 government website energy. Content is calibrated by the people who write the exam.

Boom NT2 textbooks: De Opmaat (A1-A2) and De Sprong (A2-B1). Paper textbooks the actual NT2 schools teach from. Buy them and use the apps for spaced repetition and listening volume alongside.

The honest pattern: exam-aligned Dutch is a government-publication market, not a consumer-app market. Stop expecting the apps to do this layer. Use them for general structured input and run the official prep alongside.

Failure 4: the B1-to-B2 cliff (where every app loses you)

This is the failure mode that disappoints my friends who actually made it through inburgering. They pass the exam, feel good about Dutch for two weeks, and then realize they cannot follow a Dutch podcast at normal speed, read NRC without a dictionary, or hold a meeting in Dutch without slipping into English.

The B1-to-B2 jump is roughly the same distance as A0-to-B1. It also happens to be where every consumer Dutch app effectively gives up. The library shrinks. The lesson content thins. The pedagogy assumes you will graduate to native input.

Apps that fail this test:

Almost all of them. Babbel stops at B1. Duolingo's Dutch tree thins above the early intermediate level. Mondly, Drops, and Speakly do not pretend to teach upper intermediate. This is not a flaw in the apps; it is the consumer market's natural ceiling.

Resources that handle the cliff:

NPO Radio 1 and NPO Start. The Dutch public broadcaster's news and culture radio. Pure unprocessed Dutch at native speed. Start with the 5-minute NOS Journaal in 5 minuten, climb to NOS Met het Oog op Morgen.

VRT MAX and Radio 1 (Belgian). Same role for Flemish.

NRC, Volkskrant, De Standaard. Dutch and Belgian quality dailies. One article a day with a dictionary. Painful at month one, smooth at month four.

Kletsheads. Podcast for parents raising bilingual children in Dutch. Episodes are clean B1 to B2 Dutch about pedagogy and family life. The register is exactly the one the upper-intermediate exam tests.

Assimil "Het nieuwe Nederlands zonder moeite". The textbook I personally used to start Dutch. Useful from A1 through B1. Stops being useful at B2 onwards because the method is calibrated for arrival, not advanced refinement.

LingQ with imported NRC and VRT articles. The import feature turns the Dutch press into spaced-repetition fuel.

An iTalki tutor twice a week. Above B1 the tutor moves from "speaking practice" to "register coach and editor of your written Dutch." Different role, same person.

The pattern at this layer: you graduate from the consumer app stack to the wider Dutch media ecosystem. The apps that scaffolded you to B1 have done their job. The friends I have watched stall at B2 stalled because they kept opening the same apps and expecting different results.

The four apps that survive all four tests

If I had to name the four Dutch tools that pass all four failure tests, here they are. No ranking. Each has its slot.

Mynago for daily structured input with Netherlands/Flemish toggle and production-forced exercises.

oefenen.nl for exam alignment, free and exact.

Pimsleur Dutch for the first 60 hours of production muscle.

iTalki Dutch tutor for the human-in-the-loop work that nothing else replaces. Twice a week from month one. Filter by region.

Pair this core with VRT MAX or NPO Start for input volume and the relevant Boom NT2 textbook for grammar reference, and you have a Dutch stack that does not collapse at any of the four failure points.

If you want a level check before the stack, the free Dutch level assessment takes a few minutes and is honest about whether you are A0 or whether passive exposure already nudged you to A1.

What to do after you have picked the survivors

A small note on rhythm.

The biggest predictor of Dutch progress I have observed is not which apps you pick. It is whether you maintain a tutor habit for nine months without a gap. The apps are interchangeable. The human-in-the-loop relationship is not. If you can only commit to one investment, commit to the tutor and find the cheapest survivor app to pair with them.

The second biggest predictor is whether you stop translating in your head by month four. This is when the language flips from "Dutch sentences I parse into English" to "Dutch sentences I just understand." The flip happens earlier if you have been reading and listening to unprocessed Dutch content from month one. It happens later (or never) if you have been tapping through gamified recognition for nine months.

The third predictor: how soon you stop letting Dutch speakers switch to English on you. They will. Especially in Amsterdam and Antwerp. The exam panel does not switch. Train for the panel by holding the line in everyday conversations. If a cashier switches to English, reply in Dutch anyway. By month six the switching stops because your Dutch has stopped triggering it.

The friends I have watched succeed at Dutch did some version of this. The friends who failed at Dutch did the opposite of all three.

Update, May 2026: NT2 prep changes and a Flemish correction

Several Amsterdam and Rotterdam readers wrote in after this published with corrections to my NT2 prep section.

The Het Nederlands free DUO supplement was underrated in the original. The Dutch government runs free Inburgering and NT2 preparation materials through DUO and through Het Nederlands online. The exercises are dry and the UI is grim. The content is calibrated to the actual exam by the people who write the exam. If you are exam-tracking, run these exercises in parallel with whatever paid app you choose.

The B1-to-B2 jump is harder than the original implied. I framed B1 as the comfortable plateau and B2 as a step up. Readers grinding for the state exam B2 reported the gap is enormous, and most apps stop being useful around mid-B1. This drove the failure 4 section above. NPO Radio 1, NRC, and a tutor twice a week are the real B2 stack.

Belgium-track learners need the Vlaams adjustment more than the original implied. Grammar is the same. Vocabulary diverges. Broadcasters use a different accent register entirely. If you live in Antwerp or Ghent, switch your listening input to VRT MAX and Belgian Radio 1 from week one, not from month six. This is now baked into the failure 2 section above.

The failure modes are the post. The four-app stack is the answer to the four failures. If you find an app I did not list that survives all four tests, send it to me on LinkedIn. I will update the post.