Does Duolingo Work for Learning English? What Studies Found

Yes, for the first two levels. In the studies, people who finish the beginner part of a Duolingo course (levels A1 and A2) can read and understand at about that level. One study compared the app with English classes. In grammar and vocabulary, the app did as well as the classes or better.

But the studies stop there. We found no study of a learner of English above level A2. We found no study that tested a real conversation. And in that comparison with classes, the classes were better for listening.

You should know who is writing. Zapp! English sells an English course, so we are not neutral. This page reports what the studies found. Every study is in the list at the end, so you can check what we say.

Does Duolingo work for learning English?

Most studies of Duolingo are about English speakers who learn Spanish or French. Only a few tested people who learn English. These are the four largest that we found.

Study	Who ran it, who paid	Who was tested	What was tested	What was found
Duolingo’s English courses (report 2022, journal paper 2024)	Duolingo’s own researchers	245 adults who had finished the A2 lessons: 81 Japanese speakers, 72 Spanish speakers and 92 Portuguese speakers. They said they knew little or no English before, and used nothing except Duolingo	One reading test and one listening test at the end, from an outside test company. No test at the start. No second group	On average, just above A2 in both (“early B1”, the authors write). For the Spanish speakers, listening was one step lower than reading. Speaking and writing were not tested
Duolingo and classes in Spain (2025)	Researchers at the universities of Sheffield and Nottingham. Duolingo paid. The researchers write that they planned and did the study independently	337 adults who were starting level A2, almost all in Spain. 221 used Duolingo. 116 went to classes at an Escuela Oficial de Idiomas. 16 weeks	A grammar and listening test and two vocabulary tests, at the start and at the end. No speaking test	Both groups improved a little. The app group did better in grammar and in recognising words. The class improved more in listening
The first two months (Duolingo report, 2024)	Duolingo’s own researchers	140 beginners in their first two months. 70 of them were learning English	Short spoken and written tasks: for example, say hello, or answer “What’s your name?”	90% correct on average. The report says: “Learners did not engage in an actual conversation”
A secondary school in Turkey (2025)	A university researcher and a school teacher. No funding	60 pupils aged 14 and 15. 32 used Duolingo for at least one hour a week for 12 weeks. 28 were taught in the usual way	Listening, speaking, reading and writing tests	The Duolingo group improved in all four. We could read only the summary of this study, and it gives no numbers

We also found nine smaller studies of English learners, in Saudi Arabia, Iran, Turkey, Indonesia, Colombia and Ecuador. Most have fewer than 50 learners, and most tested one skill: vocabulary, listening or speaking. All nine report improvement. We could read only the summaries of most of them, so they are not in the table.

The first study in the table is good news, with one limit. The tests came from an outside company, and a person from that company watched each test online. But nobody tested the learners at the start. They only said that they knew little or no English. So we know where they finished. We do not know for certain how much they learned with the app. The authors say this themselves, and they say that future studies need a test at the start and a second group.

Think also about who is in a study like this. Duolingo sent emails to 32,935 Spanish speakers who were near the end of A1 or in A2. 2,034 answered. 818 met the conditions. 263 took the test, and 72 of them had finished A2. A study like this shows what the course can do for people who finish it. It does not show how many people finish.

Is Duolingo as good as English classes?

One study compared them, in Spain. It is the most useful study for anyone who is learning English.

The researchers followed 337 adults for 16 weeks. All were starting the lessons for level A2. 221 of them used Duolingo, and were asked to study for 30 minutes a day. The other 116 went to classes at an Escuela Oficial de Idiomas, a public language school, for four hours a week. Everybody took the same tests at the start and at the end.

What they found:

• Both groups improved. The improvements were small in both.
• Grammar and recognising words: the app did better. The Duolingo group started with higher scores, and the researchers allowed for that. The app was still ahead.
• Listening: the class did better. In the listening part of the test, the class went from 63.8% to 67.2%. The Duolingo group went from 63.0% to 65.0%.
• Writing the right word in a sentence: no difference between the groups.

The study has limits, and the researchers list them. The learners chose their own method, so the two groups were not equal at the start. There was no speaking test. And the listening test was simple: you hear a short sentence and choose the word that you heard.

Is Duolingo good for listening?

Listening is the weakest result in the studies, and that includes Duolingo’s own.

Duolingo’s researchers tested reading and listening in three groups of learners: beginners of Spanish and French (2020), learners of Spanish and French half-way through B1 (2021), and Spanish speakers learning English (2022). Each time, the listening score was lower than the reading score. This is not special to Duolingo. One of the reports notes that university students in classes also score lower in listening than in reading.

The 2022 report on the English course says it plainly. There are, it says, “still opportunities for improvement in developing learners’ listening comprehension skills on Duolingo”.

The study in Spain adds something new. At the end, the researchers asked the Duolingo learners which skill had improved most. 60% answered listening. But the researchers write that listening was the part of the test where the app group did not make a significant gain. Their conclusion is that people who learn with an app may think their listening has improved more than it really has.

Why was the class better for listening? The researchers suggest three reasons. First, a class gives more chances to speak and to hear other people speak. Second, the class heard more English, and less controlled English, than the app learners, whose practice the researchers describe as “controlled, scripted listening exercises”. Third, the people in the class also listened to more English outside lessons: videos, television, radio, podcasts and music. These are the researchers’ explanations. The study did not test them.

Who paid for the studies?

Mostly Duolingo, and it says so. Its own list of studies says: “We fund and conduct studies on the efficacy of our courses.” (We read that page on 13 October 2026.)

We found ten studies that were run by Duolingo’s own researchers (six) or by outside researchers with Duolingo’s money (four). Every study that tested more than 100 adults is one of these ten. Seven of the ten are about Spanish or French, not English.

Studies with no company money exist, but they are small. The best known followed nine people who learned Turkish for a semester. The others tested 52 to 64 adults, or classes of school pupils.

Study	Who ran it, who paid	Who was tested	What was tested	What was found
Spanish, United States (2012)	Two outside statisticians. Duolingo paid	196 adults started. 88 were in the final results. Eight weeks, 22 hours on average	A written placement test, at the start and at the end. No listening, no speaking	Scores went up. The claim “34 hours = one university semester” comes from this study (see the questions below)
Spanish and French beginners (report 2020, journal paper 2021)	Duolingo’s own researchers	225 adults in the United States who had finished the A1 and A2 lessons	Reading and listening tests at the end. Compared with published results for university students	Similar to students after four semesters. “No other skills were assessed”
Speaking in Spanish and French (report 2021)	Duolingo’s own researchers	258 adults who had finished the A1 and A2 lessons	A 15-minute speaking test on a computer. The computer gave the marks	66% of the Spanish learners and 53% of the French learners reached A2 or above
Spanish and French, half of B1 (report 2021)	Duolingo’s own researchers	340 adults who had finished half of the B1 lessons	Reading and listening tests at the end. Compared with published results for students	Similar to students after five semesters, or a little higher. The authors write that the findings “should not be overgeneralized”
Spanish at a US university (2024)	One university researcher and two Duolingo researchers. Duolingo paid	69 students started. 48 finished. Three months, about 27 hours	Reading, listening, speaking and writing, at the start and at the end	All four improved, by about one step inside the beginner level
Spanish and French for six months (2023)	University researchers. Duolingo paid	287 adults whose first language is English	A text with letters missing, and repeating sentences aloud, at the start and at the end	Small improvements. The authors call them “relatively modest”
French: app, class or both (2026)	University researchers. Duolingo paid	183 beginners in the United States: 65 used only Duolingo, 58 had a university class, 60 had both. 16 weeks	Vocabulary, grammar and spoken answers to recorded questions, at the start and at the end	All three groups improved by about the same amount
Turkish for a semester (2019)	University researchers. No company money stated	9 adults	Language tests and interviews	They improved. More study time went with more improvement. Some were frustrated with the lessons
Turkish: Duolingo or Babbel (2023)	University researchers. They report no conflict of interest	59 adults: 32 used Duolingo and 27 used Babbel. Eight weeks	A test of six skills at the end	Both groups made progress. No difference between the two apps
Spanish at a US primary school (2018)	A researcher at a school and one at a university. No company money stated	Pupils in the third and fourth years of school. 12 weeks: Duolingo or normal Spanish lessons	A 50-question vocabulary and grammar test	No difference between Duolingo and the lessons
Italian: Duolingo or a slideshow (2019)	University researchers. A US government research office paid	64 students. Seven sessions at home	Tests at the end	No difference in learning. The Duolingo group enjoyed it more
French students invited to use Duolingo (2021)	University researchers. The European Union paid	52 university students of French	Listening, reading and writing tests	An invitation to use the app did not raise scores. Students who also learned to organise their study used it more and wrote better

Does it matter who paid? It is a reason to read carefully. It is not a reason to throw the studies away. Duolingo’s researchers used tests that outside companies made and marked. They publish the limits of each study. And the studies that Duolingo paid universities to do were not all good news for Duolingo. One found that classes were better for listening. Another called its own results “relatively modest”.

A teacher who reviewed Duolingo’s reports in 2022 reached a conclusion that is fair in both directions. The claim that a Duolingo course equals a number of university semesters “is not supported by these studies”, the review says. But “the Duolingo users did make progress”, and there was “no lack of transparency”.

One more thing to know. A review in 2021 looked at 35 studies of Duolingo and found “very little conclusive evidence”. But it was written before every large study on this page, so it describes the small early studies.

Can you become fluent with Duolingo?

Nobody knows, because nobody has tested it. That is not the same as “no”.

Duolingo’s own answer is careful. On its page of common questions it writes that “no single course, app, method, or book can help you reach all your language goals”. And then: “So, can Duolingo make you fluent? It’s up to you and how you use it!” (Read on 13 October 2026.)

Several studies tested speaking. Here is what the learners had to do:

• 2021: 258 learners of Spanish or French took a speaking test on a computer. They read sentences aloud, repeated sentences and gave short answers. A computer gave the marks.
• 2023 and 2024: in two studies, learners of Spanish or French heard sentences and repeated them aloud.
• 2024: beginners answered short recorded questions, like “What’s your name?” The report says: “Learners did not engage in an actual conversation.”
• 2026: beginners of French answered recorded questions in six short scenes. Learners who used only Duolingo did as well as learners in a university class.
• 2019, Ecuador: teachers watched 44 university students speak English and marked them with a checklist, before and after three months. The class that used Duolingo in guided lessons improved more than the other class. It is the only study we found where a person judged the speaking face to face. It is small, and the marks came from teachers at the students’ own college.

In all the other studies where we could read how speaking was tested, the learner speaks to a recording or a computer. We found no study that tested a conversation between a learner and another person. And we found no study that tested whether learners can follow other people who are talking to each other. The 2026 study, which Duolingo paid for, says the same about research on language apps in general: there has been “minimal investigation” of learning in situations like natural conversation.

What about the new AI features? A paid plan called Duolingo Max has a “Video Call”, where you speak with one of the app’s characters. On 30 March 2026 Duolingo wrote that Japanese speakers who used Video Call twice a day for a month “improved more on a standardized speaking test” than learners who did not use it. The post gives no numbers, no name for the test and no count of learners. So we cannot say more than that.

What level of English can you reach with Duolingo?

There are two different answers, and it helps to keep them apart.

What the course contains. On 22 April 2026 Duolingo announced that learners of English can now “learn through B2”. Before that, it wrote, most learners “could only complete content up to A2”.

What the studies tested. The studies of English learners tested people at the end of A2 or earlier. We found no study of anyone who finished the B1 or B2 lessons in English. One study tested learners of Spanish and French half-way through B1.

So what does A2 mean for listening? The Council of Europe wrote the levels. It describes an A2 listener like this: you “can understand enough to be able to meet needs of a concrete type, provided people articulate clearly and slowly”. To “keep up with an animated conversation” between fluent speakers is part of B2. A discussion at natural speed, where people talk over each other and use colloquial expressions, appears in the description of C1.

The distance from A2 to B2 is the part that no study of Duolingo has measured for English.

How long does it take, and do people finish?

Duolingo’s studies of English did not record hours. Two figures come from other courses. In the study of Spanish and French beginners, the typical learner needed 112 hours to finish the A1 and A2 lessons. And the 2024 paper says that in “another course” it takes about 122 hours. At 15 minutes a day, 122 hours is about 16 months. At 30 minutes a day, it is about 8 months.

Many people stop before that. It happens in every study, even when learners are paid:

• Spain, 2025: 356 people started in the Duolingo group, and 135 stopped within 16 weeks. They received 100 euros if they finished. The class lost the same share: 72 of 188.
• United States, 2024: 69 university students started. 48 finished three months.
• United States, 2012: 196 people started. 88 were in the final results after eight weeks.

We found no published figure for how many ordinary learners, outside a study, reach the end of A2.

What should you add, and when?

Stay with the app while it helps you. The studies say that it builds grammar and vocabulary at A1 and A2.

From A2, add the two things that no study of the app has tested: listening to real conversation, and talking with people.

This is also Duolingo’s advice. Its page for learners who feel stuck after A1 and A2 says: “You’re ready for more authentic materials.” It tells them to use “texts and audio not meant for learners”. (Read on 13 October 2026.)

Real conversation is different from sentences that were written for a lesson. People hesitate, repeat words, start again and speak at the same time. One researcher compared textbook dialogues with real ones and counted about seven times more hesitations in the real ones. In another study, with 171 university students of Spanish, the students who heard unscripted recordings scored lower than the students who heard scripted versions of the same content. These two studies are about textbooks and classes. They are not about Duolingo.

You can try it now. In this recording, six people talk about the food they cook and eat. Nobody is reading from a script. It is at level B1, with British and American accents. Listen once, and see how much you can follow.

After you listen, you can check your understanding with four questions on the page for Zapp! English Listening 2.16 – Food. There are more recordings, sorted by level, in Zapp! English Listening, Zapp! English Colloquial and Zapp! English Vocabulary & Pronunciation. All the audio is free.

The text matters too. If you can read the exact words after you listen, you can find the parts that you did not hear.

And talk with people: in a class, with a teacher, or with friends who speak English. In Spain, the researchers thought that speaking practice in class was one reason why the class improved more in listening.

We have to be honest about our own part as well. Nobody has tested what happens when Duolingo learners add real listening at A2. The reasons come from research on listening in general, and from the descriptions of the levels. They do not come from a study of the app.

Get the complete courseAll workbooks and audio — €39

Frequently asked questions

Is Duolingo effective?

For beginners, yes, in the studies: it builds reading, grammar and vocabulary at levels A1 and A2. Almost all of those studies were run or paid for by Duolingo. For English, nothing has been tested above A2.

Can you become fluent with Duolingo?

No study has tested it. Duolingo itself writes that “no single course, app, method, or book can help you reach all your language goals”. We found no study that asked Duolingo learners to have a conversation with a person.

Is Duolingo enough to reach B1?

Perhaps, for reading and listening tests. In Duolingo’s own study, 245 learners who finished the A2 lessons scored at “early B1” on average. They were not tested at the start, and their speaking and writing were not tested.

How long does it take to learn English with Duolingo?

Duolingo’s studies of English did not record hours. In other Duolingo courses, the A1 and A2 lessons took about 112 to 122 hours. At 15 minutes a day, that is 15 or 16 months. No study has measured the time to B1 or B2.

Is Duolingo good for listening?

It is the weakest result. Listening scores were lower than reading scores in Duolingo’s own studies. In the one comparison with English classes, the class improved more in listening, and 60% of the app learners believed that listening was their most improved skill.

Is Duolingo better than English classes?

In the one study, each was better at something. After 16 weeks in Spain, the Duolingo group was ahead in grammar and in recognising words, and the class was ahead in listening. Speaking was not tested.

Is it true that 34 hours of Duolingo equal a university semester?

That number comes from a study in 2012, which Duolingo paid for. Adults in the United States learned Spanish for eight weeks. The researchers used the scores to estimate that a beginner needs between 26 and 49 hours (34 on average) to reach the pass mark for the second semester on a written placement test. It was not a comparison with real students, and it was not about English.

Is the Duolingo English Test the same thing?

No. The Duolingo English Test is an exam, a different product from the same company. This page is about the Duolingo app for learning English.

Where the numbers come from

We read Duolingo’s own pages on 13 October 2026. They may have changed since then.

• Duolingo and classes in Spain: González-Fernández, B., & de la Viña, I. (2025). The effectiveness of app-based and classroom-based instruction on L2 learning and motivation. Language Learning & Technology, 29(1), 1–18. doi:10.64152/10125/73656
• Duolingo’s English courses: Jiang, X., Peters, R., Plonsky, L., & Pajak, B. (2024). The effectiveness of Duolingo English courses in developing reading and listening proficiency. CALICO Journal, 41(3), 249–272. doi:10.1558/cj.26704; and Jiang, X., & Pajak, B. (2022). Reading and listening outcomes of learners in the Duolingo English course for Spanish speakers. Duolingo Research Report DRR-22-08.
• The first two months: Kittredge, A., Peters, R., Neumann, F., & Jiang, X. (2024). Duolingo learners can start a conversation after 4–6 weeks of app use. Duolingo Research Report DRR-24-01.
• A secondary school in Turkey: Kazu, I. Y., & Kuvvetli, M. (2025). Digital language learning with Duolingo: assessing its impact on listening, speaking, reading, and writing skills. Journal of Computers in Education. doi:10.1007/s40692-025-00355-0
• Spanish, 2012 (“34 hours”): Vesselinov, R., & Grego, J. (2012). Duolingo effectiveness study: Final report.
• Spanish and French beginners: Jiang, X., Rollinson, J., Plonsky, L., Gustafson, E., & Pajak, B. (2021). Evaluating the reading and listening outcomes of beginning-level Duolingo courses. Foreign Language Annals, 54(4), 974–1002. doi:10.1111/flan.12600; and Duolingo Research Report DRR-20-04 (2020).
• Speaking in Spanish and French: Jiang, X., Rollinson, J., Chen, H., Reuveni, B., Gustafson, E., Plonsky, L., & Pajak, B. (2021). How well does Duolingo teach speaking skills? Duolingo Research Report DRR-21-02.
• Half of B1: Jiang, X., Chen, H., Portnoff, L., Gustafson, E., Rollinson, J., Plonsky, L., & Pajak, B. (2021). Seven units of Duolingo courses comparable to 5 university semesters in reading and listening. Duolingo Research Report DRR-21-03.
• Spanish at a US university: Smith, B., Jiang, X., & Peters, R. (2024). The effectiveness of Duolingo in developing receptive and productive language knowledge and proficiency. Language Learning & Technology, 28(1), 1–26. doi:10.64152/10125/73595
• Six months: Sudina, E., & Plonsky, L. (2023; in print 2024). The effects of frequency, duration, and intensity on L2 learning through Duolingo. Journal of Second Language Studies, 7(1), 1–43. doi:10.1075/jsls.00021.plo
• French: app, class or both: Kim, Y., Payant, C., Skalicky, S., & Namkung, Y. (2026). Comparing the effectiveness of Duolingo, Classroom instruction, and Classroom + Duolingo instruction conditions on beginner-level French language development. Studies in Second Language Acquisition, 48(3), 680–707. doi:10.1017/S0272263126101521
• Turkish for a semester: Loewen, S., Crowther, D., Isbell, D. R., Kim, K. M., Maloney, J., Miller, Z. F., & Rawal, H. (2019). Mobile-assisted language learning: A Duolingo case study. ReCALL, 31(3), 293–311. doi:10.1017/S0958344019000065
• Duolingo or Babbel: Kessler, M., Loewen, S., & Gönülal, T. (2023; in print 2025). Mobile-assisted language learning with Babbel and Duolingo: comparing L2 learning gains and user experience. Computer Assisted Language Learning, 38(4), 690–714. doi:10.1080/09588221.2023.2215294
• Primary school: Rachels, J. R., & Rockinson-Szapkiw, A. J. (2018). The effects of a mobile gamification app on elementary students’ Spanish achievement and self-efficacy. Computer Assisted Language Learning, 31(1–2), 72–89. doi:10.1080/09588221.2017.1382536
• Duolingo or a slideshow: James, K. K., & Mayer, R. E. (2019). Learning a second language by playing a game. Applied Cognitive Psychology, 33(4), 669–674. doi:10.1002/acp.3492
• Students invited to use Duolingo: García Botero, G., Botero Restrepo, M. A., Zhu, C., & Questier, F. (2019; in print 2021). Complementing in-class language learning with voluntary out-of-class MALL. Computer Assisted Language Learning, 34(8), 1013–1039. doi:10.1080/09588221.2019.1650780
• Speaking in Ecuador: Armas Pesantez, P. R., Reinoso Espinosa, A. G., Macías Silva, E. C., & Aquino Rojas, M. A. (2019). Incidencia de Duolingo en el desarrollo de las habilidades comunicacionales verbales del idioma inglés a nivel de educación superior. European Scientific Journal, 15(16), 29–44. doi:10.19044/esj.2019.v15n16p29
• Review of 35 studies: Shortt, M., Tilak, S., Kuznetcova, I., Martens, B., & Akinkuolie, B. (2021; in print 2023). Gamification in mobile-assisted language learning: a systematic review of Duolingo literature from public release of 2012 to early 2020. Computer Assisted Language Learning, 36(3), 517–554. doi:10.1080/09588221.2021.1933540
• The teacher’s review: Taylor, C. (2022). A review of Duolingo’s in-house research reports. Computer Assisted Language Learning Electronic Journal, 23(3), 156–162.
• Duolingo’s own pages: its list of studies at duolingo.com/efficacy/studies; “Can Duolingo make me fluent?” (18 November 2024); “How to pull yourself out of the intermediate slump” (27 February 2024); “Our Video Call feature boosts learners’ speaking skills” (30 March 2026); “Our top courses now teach advanced content” (22 April 2026). All are on blog.duolingo.com.
• The levels: Council of Europe (2020). Common European Framework of Reference for Languages: Companion volume, pages 48–49.
• Textbook dialogues and real ones: Gilmore, A. (2004). A comparison of textbook and authentic interactions. ELT Journal, 58(4), 363–374. doi:10.1093/elt/58.4.363
• Scripted and unscripted recordings: Wagner, E., & Toth, P. D. (2014). Teaching and testing L2 Spanish listening using scripted vs. unscripted texts. Foreign Language Annals, 47(3), 404–422. doi:10.1111/flan.12091