Why can you finish a course and still not speak?
Because recognising and producing are different skills, and almost every app trains the first while implying it is teaching the second. Selecting the right answer from four options, or typing a word you were just shown, exercises recognition. Speaking requires retrieving a word with nothing to prompt it, forming a sentence in real time and tolerating the discomfort of getting it wrong in front of someone.
Lingoda is the only tool here that addresses this directly: live classes with certified teachers, in groups capped at five so everyone actually talks, running around the clock so a class fits any schedule. That is why it costs from about €80 a month rather than €6 — you are buying a teacher's time, not software.
Busuu is the middle path: exercises you write and speak are corrected by native speakers from the community, free, so at least what you produce gets human feedback rather than an automated score. Memrise adds an AI conversation tutor, which removes the anxiety of a first conversation even if it is not a substitute for one.
The honest sequence: use an app to build the vocabulary and the grammar so you have something to say, and add speaking practice before you think you are ready rather than after. Everyone waits too long.
Does frequency-based vocabulary actually work?
The arithmetic behind it is real: a small number of words covers a very large share of everyday speech, so learning them first reaches usefulness far faster than working alphabetically or by theme.
Speakly builds its whole method on this, teaching the 4,000 statistically most useful words on the argument that they cover most of what people actually say. Lingvist orders by frequency too, and adds something more unusual: it measures what you already know in the first session and stops testing you on it, so a learner with school French does not spend three weeks on words they knew at fifteen.
That adaptivity is the genuinely efficient part. Most apps make everyone start at lesson one; Lingvist finds your level immediately and spends every subsequent minute on the gap.
The limitation both share is that vocabulary is not a language. Lingvist is explicit that it does not teach grammar or speaking, and Speakly's grammar coverage is light. They are excellent at one component and need something else beside them — which is fine as long as you know that going in rather than discovering it at month four.
What is wrong with learning a language through English?
It adds a translation step you then have to unlearn, and it excludes a lot of people entirely.
Almost every major language app teaches from English: the explanations, the prompts and the example sentences assume English is your first language. For a Dutch, Polish or Romanian speaker learning Spanish, that means learning Spanish through a second language, with every explanation routed through a grammar that is not theirs and not the target's.
Mondly covers 41 languages taught from 33 native languages, which makes it the clear answer for anyone in that position — learning Spanish from Polish rather than from English removes a whole layer of indirection. That is a genuinely European feature rather than a marketing one, because Europe is where most learners are not native English speakers.
Babbel approaches it from the other side, writing courses per language pair rather than generating one course and translating it. A German speaker learning French gets material written for that specific pair, addressing the mistakes German speakers actually make — which is a different course from the one an English speaker needs.
Does gamification help or just feel like progress?
It helps with the habit and hurts the learning, and the balance depends on which one you are short of.
Streaks, leaderboards and points solve a real problem: language learning fails from stopping far more often than from bad material. Mondly uses daily lessons and leaderboards; Memrise makes vocabulary drilling engaging enough to sustain.
The cost is that gamified apps optimise for the metric they show you. Maintaining a streak with a two-minute lesson feels like progress and produces very little, and the pleasant sense of advancement can persist for a year without conversational ability following.
Babbel deliberately does less of this — its content is conversational rather than gamified, which means the habit needs more self-discipline and the material is closer to what you will actually use. Lingvist is austere to the point that some learners find it unmotivating, and very efficient for those who do not.
The practical rule: if you struggle to keep going, take the gamified option. If you keep going anyway, take the one with better material, because the streak is not the point.
What should you check before paying for a year?
Whether the language you want is genuinely supported or nominally supported. Mondly lists 41 languages, and its smaller ones have noticeably more repetitive content than the majors. Babbel covers 14 and Busuu 14, both well. Lingvist and Speakly have small catalogues but treat what they have seriously.
Whether you need proof of level. Busuu offers McGraw Hill certificates and Lingoda is CEFR-aligned with certificates, which matters for a job application, a visa or a university requirement — and no amount of app progress substitutes for a recognised level.
What the free tier actually contains. Busuu's is genuinely usable, Memrise and Mondly have real free tiers, and Babbel has no meaningful free tier at all — only the first lesson. Lingvist and Speakly offer limited free access.
And what happens to what you produce. All seven process in the EU or UK, and Busuu and Mondly both have non-European parent companies — Chegg in the US and Pearson in the UK respectively — which is worth knowing rather than assuming from the office address.
