Choose Your Language

We support 31 NAATI CCL language pairs. Select your language to learn more about how our AI-powered practice platform can help you prepare.

Mandarin (普通话)

Three million comes back as 300 wàn: Mandarin regroups big numbers in ten-thousands, and the CCL makes you re-cut the digits live. Practise both directions on standard Pǔtōnghuà audio, marked segment by segment.

Punjabi (ਪੰਜਾਬੀ)

NAATI's Punjabi CCL is Eastern Punjabi in Gurmukhi; Shahmukhi-script Western Punjabi speakers usually sit Urdu. Every rendering gets scored both ways here, so the exam register settles before you book.

Hindi (हिन्दी)

Aap for the caseworker, tum for a mate, tu for almost nobody: every Hindi 'you' is a register call English never flags. Two-way CCL dialogues score each segment against the original.

Vietnamese (Tiếng Việt)

Before you say anything in Vietnamese you have picked anh, chị, em or bác, and with them the age and respect English hides. Dialogue practice both ways shows where kinship terms and dropped tenses cost you.

Nepali (नेपाली)

One English 'you' fans out into ta, timi, tapai and hajur in Nepali, and each tier bends the verb with it. Two-way dialogues test whether the tier you chose survives the segment.

Tamil (தமிழ்)

'Appointment', 'insurance', 'rent': everyday Tamil borrows the nouns and adds a case ending, but the CCL wants the real Tamil word. Rehearse the into-Tamil direction on dialogues that mark every segment.

Korean (한국어)

Korean counts twice over: hana, dul, set for ages and items, il, i, sam for money and dates, and one dialogue can demand both. Practice on clean Korean audio catches the wrong-system reach early.

Arabic (العربية)

You grew up speaking ʿāmmiyya; the consent form runs on fuṣḥā you read daily but rarely say aloud. Two-way Arabic dialogues build that formal register one scored segment at a time.

Telugu (తెలుగు)

NAATI accepts Coastal, Rayalaseema and Telangana Telugu alike; what wobbles is formal output after a workday spent in English. Dialogues here score that output before the exam does.

Urdu (اردو)

NAATI's Urdu policy allows 'school' and 'gas bill'; wholesale Urdish it does not. Two-way practice trains the standard register and marks the segments where English slipped back in.

Spanish (Español)

Billón means a million million in Spanish, and embarazada never meant embarrassed. Practice dialogues in both directions drill the look-alikes and mirrored decimals under a clock.

Malayalam (മലയാളം)

Malayalam stacks a whole English clause into one verb-final word, while daily Manglish borrows the nouns straight back. Scored dialogues work the English-into-Malayalam direction across ten-odd domains.

Gujarati (ગુજરાતી)

In Gujarati there are two 'we's: aapne folds the listener in, ame leaves them out, and the wrong one shifts who a sentence includes. Two-way rehearsal scores each rendering for register and the lakh-crore swap.

Filipino (Filipino)

Po and opo sit naturally in a GP consult; a full English clause mid-sentence does not. Filipino CCL practice marks where Taglish stops being a loanword and starts being a dropped rendering.

Japanese (日本語)

Sonkeigo raises the listener, kenjougo lowers you, and English never says which the room wants. Two-way Japanese dialogues are measured for the meaning carried across, never the accent.

Sinhala (සිංහල)

Written Sinhala and spoken Sinhala barely share a verb, and nerves reach for the bookish ones. Practice holds you to the spoken register in both directions and marks what drifted.

Bengali (বাংলা)

Apni, tumi or tui sets the verb ending, so one wrong 'you' bends the whole Bengali sentence after it. Two-way dialogues build the standard register, whatever home spoke: Dhaka, Kolkata, Sylheti, Chittagonian.

Kannada (ಕನ್ನಡ)

'Car-u', 'office-u': Bengaluru speech tidies borrowed English nouns with a vowel, and the CCL wants them back in Kannada. Two-way dialogue practice checks every segment for what went missing.

Cantonese (粵語)

Written Chinese read aloud lands stiff at a clinic counter; the Cantonese CCL wants spoken hau2 jyu5. Practice runs on Cantonese audio, not Mandarin, and scores every segment you speak.

Portuguese (Português)

Portuguese writes 15,50 where English writes 15.50, and bilhão changes size across the Atlantic. Brazilian or European, you rehearse both CCL directions with each segment marked for what changed.

Turkish (Türkçe)

Turkish grammar declares how you know a thing: -di for what you saw, -mış for what you were told. Two-way practice makes that call routine, along with the sen-siz line.

Persian (فارسی)

A Persian refusal can be taarof, politeness that means its opposite, and a literal rendering walks the wrong message into English. Farsi is its own test, apart from Dari and Hazaragi, and both directions count.

Indonesian (Bahasa Indonesia)

Indonesian marks no tense on the verb; sudah, sedang, akan and belum do that work, and they are easy to shed under pressure. The two-way dialogues want clean standard Indonesian, not bahasa gaul.

Marathi (मराठी)

Marathi keeps three genders, and in several tenses the verb must agree with them, a match English never rehearses. Shuddh-register dialogues run both ways, and every attempt comes back marked.

Thai (ภาษาไทย)

Thai has no neutral 'I': phom or dichan, then khrap or kha on the end, sets speaker and politeness before anything else is said. Both CCL directions run here, and the into-English leg supplies the tense Thai leaves off.

Swahili (Kiswahili)

Swahili sorts nouns into around eighteen classes, and the verb, adjective and number all bend to match. Kiswahili sanifu, minus the Sheng reflex, is what the two-way dialogues train and mark.

Yoruba (Yorùbá)

Yoruba writes its three tones into the spelling, so igba (two hundred), ìgbà (a time), igbá (a calabash) and ìgbá (a garden egg) differ by pitch alone. Both CCL directions run on standard Yoruba here, marked segment by segment.

Igbo (Igbo)

Igbo carries meaning in tone, and standard writing leaves the marks off: ákwà is cloth, àkwá is egg, four letters either way. The dialogues run both ways in izugbe, the standard nobody grew up speaking, and each segment comes back marked.

Malay (Bahasa Melayu)

Malay and Indonesian look like the same language on the page and are separate CCL tests, so the habit that costs marks is drifting into the Indonesian word mid-segment. The dialogues run both ways in standard Malay, and each segment comes back marked.

Russian (Русский)

In Russian a number decides the case of the noun after it — two changes the ending one way, five another — so a figure you heard correctly can still come out wrong. The dialogues run both ways and every segment is marked on its own.

Hungarian (Magyar)

Hungarian builds a whole English clause into one suffixed word, so the shape of the sentence only settles once you know where it ends — awkward when you are interpreting a segment you heard once. The dialogues run both ways and are marked segment by segment.

Can't find your language?

We're always adding new languages. Tell us which one you need and we'll prioritise it.

Try it in your language

Free practice test with AI feedback — free account, no credit card.

Start Free Practice Test

No credit card required to start.