The CELI: one Italian exam, split three ways by who is sitting it

CELI generici are for adulti scolarizzati and run the full ladder: CELI Impatto (A1), CELI 1 (A2), CELI 2 (B1), CELI 3 (B2), CELI 4 (C1), CELI 5 (C2). Two complete sessions a year, in June and November.

CELI a are for adolescents aged 12 to 18, in three levels — CELI 1 a (A2), CELI 2 a (B1), CELI 3 a (B2) — and they are held once a year, not twice.

CELI i are for adult migrants: CELI Impatto i (A1), CELI 1 i (A2) and CELI 2 i cittadinanza (B1).

The CVCL is precise about what does and does not differ between the adolescent line and the general one, and the precision is the interesting part. What differs is the text genres and the content, selected for an adolescent's age, interests, needs, experience and knowledge of the world — factors the CVCL says affect the ability to handle one subject rather than another. What is held constant is the CEFR reference in terms of formal and structural knowledge, the testing methods, the development process, and — stated explicitly — the same spendibilità: the same usefulness.

So a CELI 3 a and a CELI 3 are the same claim about a person's Italian, made from material a fifteen-year-old can engage with. That is a different design from every other exam here, which either offers one paper to everyone or, like the DELF, varies the version while saying much the same thing about equivalence. What no other body in this dataset does is add a third line for a specific migration status.

This is the one place the CVCL attaches its certificates to an outcome, and it does so in two tiers rather than one.

The Ministry recognises the CELI 3 (B2) as valid evidence of Italian for enrolling at an Italian university within the contingent each university reserves for non-Italophone applicants. That qualifier matters: it is admission through the foreign-student quota.

*The CELI 4 (C1) and CELI 5 (C2) are valid to attest Italian a parità di condizioni with Italophone applicants* — on equal terms with people who grew up speaking it, outside the quota.

That is a distinction the other exams in this series do not draw. The DELF/DALF page, written earlier today, states that the DELF B2 and increasingly the DALF C1 open French universities — a drift described in prose. Italy has written the same drift into a rule: B2 gets you in through one door, C1 through the other. If your goal is to compete with domestic applicants rather than for a reserved place, the CELI 3 is not the exam.

The CELI certificates also carry the ALTE Q Mark, awarded after audit by the Association of Language Testers in Europe.

The CELI Impatto i is written for people arriving from a second-language literacy course — the CVCL says so directly — and its structure reflects that.

The written part has two components, reading comprehension and written production and interaction, mixing objective tasks (binary choices on texts, matching) with subjective ones (filling in forms). The oral is a guided conversation on material the CVCL sends to each centre.

And then this: La comprensione dell'ascolto è legata alla produzione orale andando a costituire un'unica prova. Listening comprehension is tied to oral production and forms a single test. There is no separate listening paper. One level up, at CELI 1 i (A2), listening reappears as its own component, Prove C.

Here is why that is worth the trip, and the comparison is ours. This site published a guide to the DELF and DALF earlier today whose central finding was that the DALF C2 stops scoring four skills and scores two, fusing comprehension with production. That happens at C2 — the very top — because at mastery level, separating what someone understood from what they did with it is an artificial exercise.

The CELI does the same fusion at A1, for the opposite reason: a candidate who may not yet read fluently cannot be given a listening paper to fill in, because the paper itself is the obstacle. Separating the skills would measure literacy instead of listening.

So two exam families in this dataset change what counts as a separately-scored skill as the ladder moves — and they change it at opposite ends, in the same direction, for opposite reasons. Fusion happens wherever splitting the skills would measure the wrong thing. That is a better description of what a 'skill' is on a language exam than any of these bodies offers on its own.

The CVCL lists the statutes its migrant certificates satisfy, which no other exam body documented here does:

DM 04/06/2010 and DPR 179/2011 for the framework; the amendment to Art. 9.1 of Law 91 of 05/02/2002 for the CELI 1 i (A2); and Legislative Decree 113/2018 for the *CELI 2 i cittadinanza (B1)*.

The name of that last certificate is doing the work: cittadinanza. B1 is the citizenship level, A2 the long-term residence level, and the CVCL states that the A2 and B1 orals in particular use tasks referring to social and civic life in Italy — so the content is not neutral, it is about living there.

The statute behind the name, read on normattiva.it on 11 September 2026: article 9.1 of Law 91/1992 makes citizenship by marriage or residence conditional on un'adeguata conoscenza della lingua italiana, non inferiore al livello B1, with the Constitutional Court's 2025 exemption for people who cannot learn. Italy's B1 against seventeen other countries is in the citizenship-level comparison.

The stated purposes are worth reading as written, because they are unusually candid for an exam body. The CELI i exists to be an incentive to begin and continue a course of study, offering un riconoscimento tangibile e formale del lavoro svolto; to answer the integrative motivation behind learning a second language; and to be a title valid for meeting legal obligations. An exam that names motivation as one of its own design goals is a different object from one that only measures.

For the practical shape: the CELI 1 i (A2) runs 2 hours 10 minutes in total, across reading comprehension, written interaction, listening comprehension and an oral, in two booklets.

This is the most unusual passage on the CVCL's pages and it deserves stating plainly.

For candidates registering in Italy for the CELI 1 immigrati (A2) solely to meet the language requirement for the EU long-term residence permit, the CVCL gives notice that since 1 January 2024 every institution belonging to the CLIQ association has adopted specific procedures in cases of verified illiteracy.

To learn what those procedures are, the candidate must go physically to a CELI exam centre in Italy — or to the CVCL in Perugia if sitting there — before registering for the A2 certification.

And then the sentence that makes it real: the procedures can be explained only by authorised exam centres, to candidates who ask in person, which means that any request received by email or telephone cannot be answered.

An exam body stating in writing that it will not answer a question remotely has no parallel elsewhere in this dataset. Read charitably, and it probably should be, it is a safeguard: whether a candidate is unable to read is not something to establish over email, and the accommodation depends on meeting them. Read practically, it is a real barrier placed in front of exactly the candidates least equipped to cross it — you must already know to travel to a centre, before registering, to find out that an accommodation exists.

If you are preparing someone in this situation, the actionable part is simply: go to the centre first, and do not register online first. That order is the whole provision.

Two gaps, both deliberate.

We have not verified the pass marks or the sectional rules. The CVCL publishes valutazione e punteggi, tabella di conversione dei punteggi and separate written and oral scoring criteria as a PDF per level rather than as a single stated rule, and this site's standing gate is that a number appears only when we have read it on the body's own document. The comparison table therefore says not verified for the CELI's sectional column rather than guessing, and it will say so until someone reads all of them.

We have not verified a fee. No figure appeared on the pages we read. That is not the same as the CVCL publishing none — which is what this site says of the Goethe-Institut and CAPLE after looking — so the honest cell is again not verified.

And the CILS is missing. Italy has a second state certificate, run by the Università per Stranieri di Siena, and on 30 August 2026 its level pages returned a server error while the site's navigation would not open. That is a fact about one day, not a judgement about the exam, and it is recorded here rather than filled in from somewhere else. Italy is one of several languages here served by more than one state certificate — German, Portuguese and Spanish are others — and it is the one where only half the picture is documented so far.