How NAATI CCL Scoring Works for Nepali Speakers: the 29/45 Floor Nobody Mentions

Almost every Nepali candidate we speak to knows the number 63. Far fewer know the number 29 — and 29 is the one that quietly fails people who thought they had done enough. You can walk out of the test having scored 64 out of 90, above the famous pass mark, and still receive a fail letter six weeks later. This is how that happens, and how the marking works underneath it.

The floor: 29 out of 45, on each dialogue

Start with the shape of the test. You get two dialogues. Each is a conversation of roughly 300 words between an English speaker and a Nepali speaker, played to you in segments of about 35 words that you interpret aloud from memory. Each dialogue is worth 45 marks, so the test totals 90.

Most guides stop at 63. But the pass rule has a second half: each dialogue must reach 29 out of 45 on its own. Not your average. Not your total. Each one, independently.

Four results, and why three of them fail

The rule is easiest to see as arithmetic. Here are four score combinations, all plausible, and what NAATI does with each.

Look at the first two rows together. Identical totals, identical outcomes, and improving your best dialogue from 36 to 40 bought you nothing. That is the practical consequence of the floor: marginal effort is worth far more on your weak side than on your strong one. Most candidates study the opposite way, because practising what you are already good at feels better.

Why it is usually the second dialogue

In principle either dialogue can fall through the floor. In practice there are three predictable reasons a candidate's weaker dialogue is weaker, and none of them is about how well you speak Nepali.

1. 1
   The topic lottery

   The two dialogues sit in different domains — the kind of ground a GP consultation, a tenancy dispute, a Centrelink appointment or a school meeting covers. You have no say in which two you get. If one lands on the domain you have never had a conversation about in Nepali, that is the dialogue that dips.
2. 2
   Concentration debt

   Holding 35 words in memory, restructuring them and delivering them cleanly is heavy work. By the second dialogue the effort of the first is still on your shoulders. Candidates report the same thing again and again: the omissions cluster late.
3. 3
   Direction asymmetry

   Interpreting into English and interpreting into Nepali are different skills, and almost nobody is equally strong in both. Many Nepali speakers raised or educated partly in English find the into-Nepali direction the harder one — the vocabulary for tenancy, insurance or clinical detail simply gets used less at home.

How marks actually come off

Now the mechanism underneath the numbers. The 45 marks on a dialogue are not awarded for confidence or fluency. The large majority reward accurate transfer of meaning — every idea the speaker expressed arriving intact in the other language — and a smaller but real share rewards quality of language: natural grammar, an appropriate register, clear delivery.

When meaning goes wrong, markers classify it in one of three ways.

Errors are weighted. Dropping a softening word costs a sliver. Changing “one tablet” to “two” costs a chunk, because a pharmacist would act on it. The weighting follows real-world damage, which is a useful instinct to internalise: ask yourself what a real caseworker or clinician would do with what you just said.

One segment, marked

Here is a housing segment of the kind that turns up in the tenancy domain — the sort of thing that decides a dialogue. Read the source, then the candidate's version.

There are six units of meaning: the amount ($40), the frequency (a week), the start date (3rd of March), the notice period (60 days), the tribunal option, and the condition attached to it (before the increase starts). A full-mark rendering carries all six.

That single segment sheds marks four separate ways:

• Distortion — 30 days instead of 60. A specific, confident, wrong number. The tenant may believe they have missed a deadline they have not missed. This is the costliest error in the segment.
• Omission — “a week”. $40 a week and $40 a month are wildly different sums. Dropping the frequency changes the size of the increase, not just its wording.
• Omission — the deadline condition.“Before the increase starts” disappeared, so the tribunal option now sounds open-ended. The tenant loses the only piece of timing that makes the option usable.
• Addition — “they usually side with the tenant”. Nobody said this. It is invented legal advice, and inventing it is a serious addition regardless of whether it happens to be true.

Note what did not go wrong. The candidate understood every word. Their Nepali was fluent. They lost marks on retention and on discipline — and a handful of segments like this across one dialogue is precisely the slide from a comfortable 33 down through the 29 floor.

The one language-quality trap worth naming

Most quality-of-language marks look after themselves if you speak naturally. One does not: register. Nepali marks politeness levels grammatically — the well-documented tapai / timi / ta distinction — and the speakers in a CCL dialogue are usually strangers in an official setting. A doctor, a tribunal officer or a bank staff member addressing a client is not speaking the way you speak to a cousin. Slipping into the familiar form under pressure is a common, avoidable quality deduction.

What does not cost you marks

Anxiety about the wrong things is its own drain on your score. For the record, none of these are penalised:

• Your accent. Marks follow meaning and clarity, not how neutral you sound.
• Thinking time. There is no speed bonus. A composed, complete rendering beats a fast, leaky one every time.
• Using the repetition. You get one repetition per dialogue, with no penalty. Candidates routinely save it for a segment that never comes and finish the dialogue with it unused — which is a mark-losing habit, not a virtue.
• Clean self-correction. Fix a slip clearly and the corrected version is what stands.

Marking your own practice against the floor

The habit that matters most is scoring each practice dialogue separately, out of 45, and treating the worst one as your true level. After each run, work through this.

“With room to spare” is doing real work in that list. Clearing 29 in calm practice at home is not the same as clearing it in the test, and a target of 33 or 34 per dialogue gives you the margin a nervous day eats up.

Why the floor is worth this much attention

The NAATI CCL costs around AUD $814 per sitting, and there is no re-sit discount — a second attempt costs exactly what the first did. Results take roughly 4 to 6 weeks to arrive by email, so a fail on the per-dialogue floor costs you the fee again plus a couple of months of your migration timeline. Passing is worth a flat 5 points under the credentialled community language factor, identical for the 189, 190 and 491 visas, and identical whether you score 63 or 89. There is no prize for a high mark and a heavy price for one weak dialogue — which tells you exactly where to spend your preparation.

If the marking model is new to you, the full breakdown of how CCL marking works goes deeper on error weighting, and the guide to the 5 PR points covers the migration side.

Quick check

Frequently asked questions