Streaks are the most recognisable feature in language learning apps: a counter of consecutive days on which you did something. They work. People come back to protect the number. Our view is that this is precisely the problem with how apps present them. A streak measures whether you showed up, not whether you can understand or say more than you could last month, and apps should put a measure of skill at least as prominently on the home screen as the count of days. Until they do, learners are being encouraged to optimise for the wrong number.
What a streak actually counts
Duolingo, which popularised the feature, defines its streak as the number of days in a row you have completed a lesson. One lesson keeps it alive. The company is open about why the feature matters to it. In a post explaining how the streak builds habits, it reports that learners who reach a 7-day streak are 3.6 times more likely to complete their course, that adding animations to streak milestones increased the share of brand new learners returning after seven days by 1.7%, and that letting learners hold two Streak Freezes at once increased daily active learners by 0.38%.
Those are honest numbers, and they describe a feature tuned for one outcome: coming back tomorrow. The 3.6 times figure is a correlation, and people who were already motivated are the ones most likely to reach seven days. None of the experiments described measures what learners could do in the language.
What the research says about game mechanics
The best overview of gamified learning is a meta-analysis by Michael Sailer and Lisa Homner in Educational Psychology Review. Pooling experimental studies, it found significant small effects of gamification on cognitive learning outcomes (g = 0.49), motivational outcomes (g = 0.36) and behavioural outcomes (g = 0.25). The effect on cognitive outcomes held up when the authors looked only at the most rigorous studies. The effects on motivation and behaviour were less stable. Game elements such as a story and social interaction mattered for behavioural outcomes; points for their own sake were not singled out as the active ingredient.
An older line of research is more pointed. A meta-analysis of 128 experiments by Edward Deci, Richard Koestner and Richard Ryan in Psychological Bulletin found that rewards given simply for taking part or for completing a task significantly undermined free-choice intrinsic motivation, while positive feedback about performance enhanced it. Those were mostly laboratory studies, many with tangible rewards, and none tested a language app. But the distinction is useful. A streak rewards engagement. Feedback on what you can now do rewards competence. The research favours the second.
Time on task does matter. A semester-long case study of nine university students learning Turkish on Duolingo, published in ReCALL, found a positive, moderate correlation between time spent on the app and learning gains. The participants liked the app's flexibility and gamification, but also reported varying motivation and frustration with the instructional materials. What predicted progress was time, not consecutive days.
The strongest case for streaks
The best argument for the streak is that dropout, not inefficiency, is what defeats most self-taught learners. Language learning takes a long time, and a tool that keeps someone practising for months is worth more than a perfect method abandoned after a week. On that view the streak is not a distraction from progress but its precondition, and Duolingo's own data suggest it keeps people practising. A few minutes every day, the argument goes, beats an hour once a fortnight.
We accept most of that. Habit is real, and a short daily session is a reasonable way to learn vocabulary. Our objection is narrower. When the streak is the biggest number on the screen and one lesson is enough to protect it, it invites the minimum: the quick review at 11pm that keeps the flame lit and teaches little. A learner who has kept a 400-day streak may feel more accomplished than one who studied hard three days a week, and the app has done nothing to tell them otherwise.
Apps already have better numbers
This is fixable, partly because the measures exist. Duolingo itself reports a Duolingo Score, and says that in nine of its courses the content now runs to a score of 129, which it equates to B2, the fourth of six levels of the Common European Framework of Reference. A level on a recognised scale is the kind of number that tells learners whether their daily minutes are adding up.
What we would like to see, from every app that uses streaks:
- Skill first, days second. Show an estimated level, or progress within a level, at least as prominently as the streak.
- Let learners define a day. A streak that requires the learner's own target, ten minutes or a speaking exercise, means more than one that requires a single lesson.
- Show time, not just days. Weekly minutes of practice are a better guide to progress than an unbroken chain, according to the research above.
- Make the streak optional. Some learners find the pressure helpful. Others would rather study four days a week without feeling they have failed on the other three.
For learners in the meantime, the practical point is simple. Treat the streak as a reminder to practise, not as evidence that you are improving. Check your progress against something outside the app now and then: a level test, a conversation with a tutor or a native speaker, a podcast you could not follow last spring. If the number on the screen is growing and those are not, the number is measuring the wrong thing.