Esperanto Propaedeutic Value: What the Studies Actually Show
By Michael McGettrickUpdated 11 October 2026
is genuinely quick to learn, and the studies back that. The famous follow-on claim, that learning it first makes your next language faster, was tested by the best-designed research and not found. Two claims get run together, and almost all the confusion lives in the gap between them.
| Claim | Verdict | Best evidence |
|---|---|---|
| Fast to learn | supported | Essex 2013, randomised: about 45% vs 14% |
| A springboard | not found | Essex 2013 and 2017, 225 and 178 children |
| The levelling effect | thin | lower variance, not a higher mean |
| "Five times faster" | publicity | David Kelso, project promotional material |
Quick check
Now use it: three quick questions on the chart above. The full practice is further down.
Pick the right one
0/3What the evidence supports. Pick the statement this page will actually stand behind.
(about the transfer claim, on the best evidence)
(what the 45% against 14% figure measures)
(about who ran most of the propaedeutic research)
The structural reasons it is fast are on how to actually learn Esperanto. This page is the evidence.
Two claims, not one
Claim one: is faster to learn than a national language. It has decent support and is not really controversial.
Claim two, the propaedeutic one: learning Esperanto first makes your next language faster. A year of Esperanto plus two of French beats three of French. It is the claim the movement leads with, and on the best evidence available it is the one to retire.
The mechanism, which is plausible
Esperanto makes grammatical machinery visible. A learner meets agreement, the object ending and word building in a form so regular the concept cannot be missed:
-jon the adjective as well as the noun).- I see the book. (the object marked with-n)
When the same concepts turn up in French, buried under irregularity, the learner already has a box to put them in. Researchers call this raising explicit metalinguistic awareness. It is a real cognitive story and not obviously wrong; it just has not shown up in the results.
The classic studies, and what is wrong with them
Two programmes get cited more than all the others:
| Study | When | Who and how many |
|---|---|---|
| Five-country | 1975 to 1977 | ILEI; 302 pupils in Belgium, , Greece, West Germany and the Netherlands |
| Paderborn | about 1975 to 1982 | Helmar s ; about 600 pupils from year three |
The five-country experiment ended with a joint week at St Gérard in Belgium in 1977 and is documented in Helmut Sonnabend's : (1979), a movement publication. Paderborn was built into a teaching called -a o, language-orientation teaching.
The problems are not small:
| Problem | Detail |
|---|---|
| Nothing randomised | intact school classes, so any class difference is a result difference |
| Who published | his own , and a journal he founded and edited |
| Thin pedigree | "the Sheffield study" ran in , Manchester; Williams called his conclusions impressionistic |
| No -analysis | nobody has ever pooled the |
s results also went out through a political foundation and in-house protocols; the one nearly-mainstream write-up appeared in Linguistik und Didaktik in 1980. When the researcher, the publisher and the editor are the same person, the result is not fraudulent and it is also not independent.
The one mainstream review is Maxwell (1988) in Studies in Second Language Acquisition. He came down cautiously in favour of the ease-of-learning claim, while noting that the results could be challenged mainly because of the wide range of variable factors involved. That is a sympathetic reviewer describing weak evidence.
Where the modern evidence lands
Since the mid-2000s, Tellier and Roehr-Brackin at the University of Essex have studied the question, attached to the British Springboard to Languages project. Applied linguists rather than advocates designed it, and they published their null results too.
| Study | Children | Design | Finding |
|---|---|---|---|
| Essex 2013 | 28 | randomised in one class | gain about 45%, French about 14% |
| Essex 2013 | 225 | cross-sectional | no lasting language-awareness edge, years later |
| Essex 2017 | 178 | intact groups | no benefit to later French on any of four skills |
Read the 45-against-14 figure properly
The randomised study ran about twenty-two and a half hours of teaching. Its figure is a comparison of how much each group learned of the language it was being taught. The Esperanto learners learned more Esperanto than the French learners learned French: clean evidence that is easier to learn. It is not a transfer result. Nobody's French improved in that study, because nobody in the Esperanto group was learning French yet.
On transfer, nothing
The 2017 study is the direct test. Four groups started on German, Italian, Esperanto, or Esperanto with grammar focus, then all took the same French course:
- French afterwards: no significant difference between the groups, overall or in any of the four skills tested (grammar, listening, reading, writing).
- The one positive signal: a language-awareness edge for the Esperanto-with-grammar group showed in the full 178 and vanished among the 116 who completed both parts, the signature of noise rather than an effect.
- Years later: the 225-child study found no lasting language-awareness advantage either.
What survives is a possible levelling effect: lower variance rather than a higher mean, so the class spread out less rather than did better. Even that is fragile: descriptive rather than inferential.
The citable verdict
Roehr-Brackin and Tellier's 2018 paper is free to read in the Essex repository, so you can check it rather than trust this page. It finds no evidence that Esperanto is a better springboard than another European language, and it makes the control criticism that undoes most of the older literature: much of it compares children taught Esperanto against children taught nothing at all. That shows something beats nothing, not that Esperanto beats another language.
Two cautions about the publicity
- The figure that pupils learn a language "up to five times more quickly" after Esperanto is David Kelso's, from Springboard to Languages' own promotional material. Never quote it as a result.
- Springboard to Languages produced five unpublished reports, and the evaluators themselves named teacher specialism as a confound.
- Amanda Barton and Joanna Bragg evaluated the project independently from 2006 to 2011; Roehr at Essex summarised their reports.
- The "four schools, 230 pupils" figure traces to the project's now-dead website.
What to actually believe
- is measurably quicker to learn. One randomised study plus the structural argument: this is the claim that holds up.
- The springboard claim should be retired. The best-designed studies looked for it, with hundreds of children, and did not find it.
- The levelling effect is the only survivor, and it is too thin to build an argument on.
None of this is a reason not to learn Esperanto. It is a reason to learn it for its own speed and its own community rather than as an investment in a French exam.
See also
- How to actually learn Esperanto - why it is fast, and the method that works.
- Esperanto today - where the speakers actually are.
- The Fundamento and the Akademio - who decides what is correct.